Rethinking Epistemology: Volume 2 9783110277944, 9783110277821

Citation preview

Rethinking Epistemology 2

Berlin Studies in Knowledge Research Edited by Günter Abel and James Conant

Volume 2

De Gruyter

Rethinking Epistemology Volume 2 Edited by

Günter Abel and James Conant

De Gruyter

Editors Prof. Dr. Günter Abel Technische Universität Berlin Institut für Philosophie Straße des 17. Juni 135 10623 Berlin Germany e-mail: [email protected] Prof. Dr. James Conant The University of Chicago Dept. of Philosophy 1115 E. 58th Street Chicago IL 60637 USA e-mail: [email protected]

ISBN 978-3-11-027782-1 e-ISBN 978-3-11-027794-4 Library of Congress Cataloging-in-Publication Data A CIP catalog record for this book has been applied for at the Library of Congress. Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available in the Internet at http://dnb.dnb.de. 쑔 2012 Walter de Gruyter GmbH & Co. KG, Berlin/Boston Printing: Hubert & Co. GmbH & Co. KG, Göttingen ⬁ Printed on acid-free paper Printed in Germany www.degruyter.com

Foreword The present volume is the second of two volumes which together form a publication which seeks to introduce a variety of new approaches to current problems in epistemology. Taken together the two volumes are the starting volumes of the Berlin Studies in Knowledge Research series. They are intended to open the series and indicate some of the variety of topics to which the subsequent forthcoming volumes will each be individually devoted. As the title which these volumes jointly bear – Rethinking Epistemology – indicates, their shared aim is to identify important topics in the theory of knowledge which have either been unduly neglected in recent philosophy or whose consequences for other areas of epistemology in particular and philosophy more generally have tended to remain unappreciated. This is therefore also the central aim of the present volume as well as one of the central aims of the entire series which this pair of volumes seeks to indicate. The series as a whole is animated by a number of subsidiary aims as well. It is closely associated, both in its selection of topics and its overall philosophical orientation, with the projects promoted by the Innovationszentrum Wissensforschung (IZW), or Center for Knowledge Research, at the Technische Universität Berlin. The goal of both the Center and the series is to foster systematic research into the variety of forms of knowledge that there are, as well as to uncover aspects of their underlying unity. The handful of remarks that follow are intended to provide only the barest outline of the intellectual program of the Center and its associated publication series. (A detailed discussion of the variety of forms of research into the nature of knowledge here envisioned and of what makes them valuable is provided in the opening contribution to the first volume.) In offering these remarks here, in the Foreword to this collection, we do not mean to suggest that the contributors to this volume necessarily conceive of themselves as all engaged in a common intellectual project, let alone the one sketched below. We mean rather merely to suggest that their work, considered collectively, is helpfully viewed as marking a certain trend in recent work in analytic philosophy in both the Anglo-American and German academy – a trend which, on the one hand, marks a notable departure from the topics

VI

Foreword

and concerns which previously formed the staple of analytic epistemology and which, on the other hand, marks a movement in the direction of a more capacious understanding of what knowledge is and what the philosophical study of it might require. For the purposes of both the volume and the series, the conception of the discipline of epistemology at work throughout is intended to be a maximally generous one – one which encompasses a study of the full variety of forms, practices and dynamics of knowledge, as well as their mutually interacting points of contact and their respective mechanisms of interpenetration. It is this generous conception of the discipline of epistemology that is indicated by the German expression Wissensforschung and its regrettably unattractive English counterpart “knowledge research”. On this conception, the topics of mainstream epistemology, such as a posteriori perceptual and a priori inferential knowledge, are asked to take their place alongside the full spectrum of other forms of knowledge that come with the entire spectrum of human cognition, from thought, speech, and self-consciousness to action, skill, and virtue. The series thereby seeks to bring about an expansion and revision of classic analytical conceptions of the philosophical disciplines of epistemology and the theory of science. The hope is to nudge the future philosophical study of knowledge into an attitude of greater openness to interdisciplinary and transdisciplinary approaches to its topic and to foster a greater appreciation of the irreducible plurality of the various forms of human knowledge. This leads to a reorientation of the discipline of epistemology, undoing some of its artificial restrictions in its conception of the scope of its topic, while introducing a greater emphasis on the heterogeneity of different forms of knowledge, hopefully thereby effecting the possibility of a nuanced study of the ways in which these heterogeneous forms mutually impinge upon and condition one another. Some felt need for such forms of inquiry is reflected in the increasing profusion of partially overlapping terminological pairs – pairs such as “implicit/explicit”, “discursive/non-discursive”, “theoretical/practical”, “knowing-how/knowing-that”, “conceptual/non-conceptual”, “codifiable/uncodifiable”, “first-person/third-person”, personal/subpersonal”, and “normative/non-normative” – and an increasing appreciation of the ways in which none of the distinctions drawn by any of these terminological pairs manages to align all that perfectly with those drawn by any of the others. Although it forms a less central part of the agenda of the present volume than that of some of the forthcoming subsequent volumes, a sub-

Foreword

VII

sidiary concern of the series as a whole is to help redefine the traditional theory of science, as well as the variety of recent forms of science studies which have sought to replace it. The hope is to foster a greater dialogue between these forms of inquiry and philosophy proper, so that the study of the acquisition and development of scientific knowledge can become more philosophically systematic, while philosophical reflection on science can grow more sensitive and attentive to the genuine diversity and irreducible historicity of the forms of scientific knowledge which have proliferated over the centuries. Finally, the series will seek to foster reflection on the ways in which scientific and everyday ways of knowing overlap and diverge with one another, while resisting the tendency to hold up either as the standard by which to measure the other. This last aspect of the orientation of the series reflects its concern with promoting proper philosophical reflection on the interface between the world as it is known to us in our everyday lives (what the Germans call the Lebenswelt) and the world as its nature is disclosed to us by the natural sciences – between what Wilfrid Sellars called the manifest image and the scientific image of the world. Philosophical comprehension of the relations in which these two images stand to one another requires an exploration of the manifold points at which philosophy, the humanities, the social sciences, and the natural sciences all interact with one another. A number of the subsequent volumes in this series will seek to explore some of these transdisciplinary issues. A final aim of this publication series is to promote and facilitate intellectual exchange between European and Anglo-American scholars of knowledge. The hope is to draw attention to key writings by European researchers which have gone relatively unnoticed in the Anglo-American discussion and to promote a broader consciousness of seminal recent English-language contributions which have yet to be received in Europe. Each volume will therefore contain contributions by both Anglo-American and European scholars working on closely related topics. The present volume of this new series has a focus which in one respect is comparatively narrow than that of the series as a whole: it targets some of the central philosophical questions of epistemology as they have been taken up in recent years by philosophers seeking to reshape our understanding of what knowledge is. The editors would like to express their indebtedness to a number of institutions and persons. First, we are grateful to the Technische Universität Berlin for establishing the Innova-

VIII

Foreword

tionszentrum Wissensforschung (Center for Knowledge Research). We would also like to thank the publishing house Walter de Gruyter (Berlin – New York) and above all Dr. Gertrud Grünkorn (Editorial Director) for supporting the idea of this series from its inception. We would also like to thank the staff of the Innovationszentrum Wissensforschung; without their effort and commitment the present volume would not have come into existence. Particularly, noteworthy in this regard were the services of Peter Remmers, Martin Wolf, Hadi Faizi, Can Atli, Katharina von Laer, Claudio Roller, Stefan Tolksdorf, Doris Schöps, Elisabeth Simon and Daniel Smyth (Chicago). Last but not least, we are indebted to the authors of this volume of the series. It is through their contributions that the idea of “Rethinking Epistemology” has been converted from a possibility to an actuality. Günter Abel (Berlin) & James Conant (Chicago)

Contents James Conant Two Varieties of Skepticism . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

I. First-, Second-, and Third-Personal Knowledge Quassim Cassam Knowing What You Believe . . . . . . . . . . . . . . . . . . . . . . . . . .

77

David H. Finkelstein From Transparency to Expressivism . . . . . . . . . . . . . . . . . . . . .

101

James Doyle On Finkelstein’s Account of the Distinction between Conscious and Unconscious States of Mind . . . . . . . . . . . . . . . . . . . . . . .

119

Will Small Practical Knowledge and the Structure of Action . . . . . . . . . .

133

Matthias Haase Three Forms of the First Person Plural . . . . . . . . . . . . . . . . . .

229

Benjamin McMyler Testimony, Address, and the Second Person . . . . . . . . . . . . . . .

257

II. Knowledge and the Problem of Dualism John McDowell Tyler Burge on Disjunctivism . . . . . . . . . . . . . . . . . . . . . . . . . .

289

Thomas Lockhart Motivating Disjunctivism . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

309

Adrian Haddock Disjunctive Conceptions of Experience and Perceiving . . . . . .

349

X

Contents

Martin Gustafsson Trusting One’s Senses: McDowell on Experience, Belief and Justification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

375

Matthew Boyle Essentially Rational Animals . . . . . . . . . . . . . . . . . . . . . . . . . . .

395

Notes on Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

429

Index of Persons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

431

Index of Topics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

435

Two Varieties of Skepticism James Conant This paper distinguishes two varieties of skepticism and the varieties of philosophical response those skepticisms have engendered. The aim of the exercise is to furnish a perspicuous overview of some of the dialectical relations that obtain across some of the range of problems that philosophers have called (and continue to call) “skeptical”. I will argue that such an overview affords a number of forms of philosophical insight.1

I. Cartesian and Kantian Varieties of Skepticism – A First Pass at the Distinction I will call the two varieties of skepticism in question Cartesian skepticism and Kantian skepticism respectively.2 (These labels are admittedly contentious.3 Nothing of substance hangs on my employing these rather than 1

2

The taxonomy is meant to serve as a descriptive tool for distinguishing various sorts of philosophical standpoint. It is constructed in as philosophically neutral a fashion as possible. The distinctions presented below upon which it rests are ones that can be deployed by philosophers of very different persuasions regardless of their collateral philosophical commitments. A philosopher could make use of these distinctions to argue for any of a number of very different conclusions. Some of the more specific philosophical claims that I myself express sympathy for in the latter part of this part (e. g., regarding how these varieties of skepticism are related to one another) do, however, turn on collateral philosophical commitments. The taxonomy presented here affords the terminological resources for a precise statement of these claims. It cannot and is not meant to do more. A reason for referring to the taxonomy offered here as “partial” is because the overview of varieties of skepticism set forth here is in no way intended to be exhaustive. On the contrary, much of the point of this paper lies in showing that philosophical illumination can be afforded by an attention to differences among varieties of skepticism. It should be possible to uncover further ways in which it can prove helpful to introduce further distinctions between varieties of skepticism – distinctions that are elided or overlooked in the admittedly schematic taxonomy furnished here. An additional, more particular reason for referring of the taxonomy offered here as “partial” is because it is in no way intend-

2

James Conant

certain other labels that might serve equally well to mark the distinctions to which I will be seeking to draw attention. What matters are the distinctions themselves.4) Each of these varieties of skepticism has its origin in a skeptical question – I will call these the Cartesian question and the

3

4

ed to accommodate (what I take to be) the variety of skepticism which (for lack of a better label) I will call Pyrrhonian skepticism: one which has its roots in ancient skepticism, which remains a (seemingly) live philosophical option until at least Montaigne, and which continues to remain a significant philosophical point of reference to nineteenth-century thinkers as different from one another as Hegel, Kierkegaard, and Nietzsche – each of whom regard it (though not entirely for the same reasons) as not only a distinct but also (as compared with modern varieties) as a philosophically comparatively superior form of skepticism. To try to take proper account of the similarities and differences between this variety of skepticism and the Cartesian and Kantian varieties respectively would make this (already too long) paper unmanageably long. I will, nonetheless, occasionally remark in a passing footnote on some aspect of the relation between the generic features of this comparatively ancient form of skepticism and those of the relatively modern varieties with which I am focally concerned here. A yet further variety of skepticism that some might think ought to be sharply distinguished from the Pyrrhonian, Cartesian, and Kantian varieties respectively is Agrippan skepticism (see Williams (2001), 61 ff.) The question of whether Agrippan skepticism is a genuinely distinct variety – and, if so, how it is best formulated in order to bring out its distinctive generic features – is not one that I can afford to take up here. There is much that any conscientious historian of philosophy will find to object to in my choosing to attach these labels to each of these varieties of skepticism: does not (what I am calling) Cartesian skepticism antedate Descartes?; is not the issue of skepticism a sideshow in Descartes’ philosophy and hence of relatively minor importance (compared to other things – say, the new science, or the quarrel with the Church) for an understanding of Descartes’s work as a whole?; are not features of (what I am calling) the Kantian problematic to be found equally in Descartes?, is not Kant more Cartesian than I suggest he is?, is not skepticism equally a sideshow in Kant’s larger endeavor?, and what about Hume? Such objections are directed at the aptness of these labels; and it would take considerable space to respond to each of them. The point of the present note is simply to remark that it would be a mistake to think that the integrity of this project (of distinguishing the varieties of skepticism I choose to label “Cartesian” and “Kantian”) is much threatened by such objections. Such objections speak only to the question whether I wouldn’t be better advised to re-label the varieties of skepticism with which I am here concerned. Later on, a brief attempt will be made both to explain why I choose these particular labels and to respond to a few objections of the above sort. For the most part, however, such historical worries will have to go unaddressed here. As we shall see, there are – in addition to historical – also philosophical grounds for suspecting the stability of these distinctions.

Two Varieties of Skepticism

3

Kantian question respectively; and each of these varieties of skeptical question leads to a skeptical paradox – which I will call the Cartesian paradox and the Kantian paradox respectively. I will call the imaginary philosopher who acquiesces in the Cartesian paradox a Cartesian skeptic, and the (even more) imaginary philosopher who acquiesces in the Kantian paradox a Kantian skeptic. The customary response to each of these paradoxes is to seek a way to entitle oneself to do something other than acquiesce in the paradoxical conclusion, by refuting or dissolving or diagnosing or by-passing the paradox in question. This gives rise to two varieties of philosophical problematic, flowing from such attempts to address each of these two sorts of skeptic, that I will call the Cartesian problematic and the Kantian problematic respectively. It is with these twin problematics that I will be primarily concerned here and which I will be discussing and referring to as “varieties of skepticism”. According to this unconventional idiom, the term “skepticism” (and its variants, such as “Cartesian skepticism” or “Kantian skepticism”) therefore refers not just to one particular sort of philosophical position (i. e., that held by one or another sort of skeptic) but rather to the wider dialectical space within which philosophers occupying a range of apparently opposed philosophical positions (such as “realism”, “idealism”, “coherentism”, etc.) engage one another, while seeking a stable way to answer the skeptic’s question in the affirmative rather than (as the skeptic himself does) in the negative.5 So, according to the terminology I am here introducing, a philosopher can be concerned with the Cartesian skeptical problematic without himself being a Cartesian skeptic; and, indeed, Descartes was such a philosopher. And, according to this terminology, 5

Such an inclusive use of the term ‘skepticism’, while unusual, is not unprecedented. For a similarly inclusive use, especially in connection with Cartesian skepticism, see, for example, Cavell (1979): I do not … confine the term [‘skepticism’] to philosophers who wind up denying that we can ever know; I apply it to any view which takes the existence of the world to be a problem of knowledge. A crucial step for me, in calling an argument skeptical, is that it contain a passage running roughly, “So we don’t know (on the basis of the senses (or behavior) alone); then (how) do we know?”. It is at this stage that philosophies break into Phenomenalism, Critical Realism, etc…. I hope it will not seem perverse that I lump views in such a way, taking the very raising of the question of knowledge in a certain form, or spirit, to constitute skepticism, regardless of whether a philosophy takes itself to have answered the question affirmatively or negatively. It is a perspective from which skepticism and (what Kant calls) dogmatism are made in one another’s image, leaving nothing for choice. (46)

4

James Conant

the same holds for the relation between being concerned with the Kantian skeptical problematic, being a Kantian skeptic, and being Kant. The following is an excerpt from the classic formulation of the Cartesian problematic: How often, asleep at night, am I convinced of just such familiar events – that I am here in my dressing-gown, sitting by the fire – when in fact I am lying undressed in bed! Yet at the moment my eyes are certainly wide awake when I look at this piece of paper; I shake my head and it is not asleep; as I stretch out and feel my hand I do so deliberately, and I know what I am doing. All this would not happen with such distinctness to someone asleep. Indeed! As if I did not remember other occasions when I have been tricked by exactly similar thoughts while asleep! As I think about this more carefully, I see plainly that there are never any sure signs by means of which being awake can be distinguished from being asleep…. Suppose then that I am dreaming, and that these particulars – that my eyes are open, that I am moving my head and stretching out my hands – are not true. Perhaps, indeed, I do not even have such hands or such a body at all.6

I wish to contrast the skeptical problematic which figures in the above passage with the one which figures in the following excerpt from Kant: The a priori conditions of a possible experience in general are at the same time conditions of the possibility of objects of experience. Now I maintain that the categories … are nothing but the conditions of thought in a possible experience…. [A]nd without such unity … no thoroughgoing, universal, and therefore necessary, unity of consciousness would be met with in the manifold of perceptions. These perceptions would not then belong to any experience, consequently would be without an object, merely a blind play of representations, less even than a dream.7

The problematic of the first of these passages centers on how to distinguish between dreaming that one is experiencing something and actually experiencing it. The problematic of the second of these passages centers on what it takes to be able to dream that one is experiencing something. That is to say, the second of these two problematics focuses on the conditions of the possibility of something that the first problematic takes for granted. I take the (apparent) difference here to be a consequential one. The most familiar way of formulating the contrast between these two problematics is as one of knowledge vs. the conditions of knowledge. Thus one is often told something along the following lines: The Carte6 7

The quotation is drawn from the fifth and sixth paragraphs of Descartes’s First Meditation (Descartes (1986)). Kant (1998), A111 f.

Two Varieties of Skepticism

5

sian wants to arrive at knowledge; the Kantian wants to arrive at the ground of the possibility of knowledge. But what does that mean? There are lots of ways of unpacking this contrast. I will briefly indicate a few of the different points upon which the accent can fall in an unpacking of this contrast. I will, however, want to claim that these apparently distinct formulations can be seen, in the end, to come to the same thing. Here are three ways one might go about unpacking this contrast: 1. Actuality vs. possibility. Cartesian skepticism takes the possibility of experience for granted; its question has to do with actuality. Hence the importance of the word ‘real’ in Cartesian formulations of the skeptical problematic: are things really as they seem? Kantian skepticism brings within the scope of its worry that which the Cartesian skeptic takes for granted: that experience possesses the requisite unity so much as to be able to be about something. Hence the importance of the word ‘possible’ in Kantian formulations of the skeptical problematic: how is experience (so much as) possible? 2. Being so vs. being so. 8 Cartesian skepticism calls into question the veridicality of one’s experience; Kantian skepticism calls into question the intelligibility of experience. The Cartesian problematic is concerned with the question: How can I know that things are as they seem? Hence the worry in the Cartesian problematic focuses on an inferential step from appearance to reality. The Kantian problematic is concerned with the question: How can things so much as seem to be a certain way? Hence the worry in the Kantian problematic focuses on the conditions of the possibility of the kind of unity presupposed by the Cartesian (i. e., that which the Cartesian seeks to infer from): what sort of unity must characterize a “play of presentations” for it to be more than “a merely blind play”, for it to possess the aspect of offering appearances – for it to possess the character of being of an object. 3. Truth vs. objective purport. The Cartesian wants to know which of his thoughts are true, which of his experiences are veridical. The Cartesian skeptic therefore worries about the transition from a sensory experience to a judgement, from a thought to (what Frege calls) 8

I owe this way of formulating the contrast to Stanley Cavell. See Cavell (1979), 45. (However, I make a use of this contrast here – to formulate the distinction between Cartesian and Kantian skepticism – that Cavell himself might not approve of.)

6

James Conant

its truth-value. Hence the Cartesian problematic inquires into the grounds of truth: given that this is what we are inclined to judge, do we know that we judge truthfully in so judging? The Kantian skeptic seems to deprive us of the resources for so much as being able to enjoy an experience (waking or dreaming), for so much as being able to frame a thought (true or false). The Kantian problematic inquires into the grounds of the possibility of being able to enjoy an experience, entertain a thought-content. The Kantian asks: What does it take to have thoughts that are vulnerable to how things are? The Kantian problematic is concerned, in the first instance, not with truth but with what it is to stick your neck out in thinking, with Kant calls the objective validity of judgment (the possibility of something’s being a candidate for truth or falsehood) – with what I will henceforth call the objective purport of judgement. In any area of philosophy in which one finds one of these two varieties of skepticism, one generally also finds the other. This is not to deny that in some areas of philosophy one of these problematics may come to seem more gripping or otherwise deserving of interest than the other. I do want to argue, however, that wherever one of these sorts of skepticism is possible, the other is also possible. (The fact that these same problematics can and often do surface in virtually every so-called “area” of philosophy is itself a ground for wondering whether philosophy is usefully divided into separate “areas” of inquiry as so many nowadays are prone to suppose.) I will briefly indicate a few of the guises in which the Cartesian and Kantian problematics respectively surface across a number of (supposedly distinct) “areas” of philosophy. But before I do so, it is worth noting one of the many respects in which the labels “Cartesian” and “Kantian” may seem tendentious: namely, as labels for some of the specific variants I shall be concerned to distinguish within (what I am here calling) the “Cartesian” and “Kantian” varieties of skepticism respectively.9 Descartes himself, for example, never poses most of the 9

I speak here, rather tentatively, of varieties (rather than genera) of skepticism, and of these varieties as subtending variants (rather than species) of Cartesian and Kantian skepticism respectively, in order to leave open questions concerning how these twin philosophical problematics relate to one another – questions such as: to what extent do they constitute distinct kinds of philosophical worry?, to what extent do they overlap?, or must they mutually exclude one another? I want the terminology I initially employ to introduce the philosoph-

Two Varieties of Skepticism

7

skeptical worries that I will be calling variants of “Cartesian skepticism”. (Indeed, it is arguable that he only ever clearly poses one of them – namely, skepticism about the external world.) And, though Kant arguably explores more of the possible variants of (what I will be calling) “Kantian skepticism”, he seems to be quite oblivious to some of them. Thus, in claiming that the philosophical problems I mention below represent variants of Cartesian and Kantian skepticism respectively, I am not making an historical claim about which problems are (and are not10) discussed in the writings of Descartes or Kant. Rather I am making a philosophical claim about a congruence to be found in the shape of the problems themselves, regardless of whose writings they appear in. In order to gain a clearer sense of what constitutes the unity of a single variety of skepticism, it will help first to consider some variants on a single variety. I will accordingly first present five examples of (what I ical phenomena under examination in a manner sufficiently neutral so as to allow the question “How, if at all, are these – Cartesian and Kantian – skepticisms related to one another?” to admit of a variety of answers, without prejudging any particular line of answer. However, as will become clear, I myself do favor a particular line – one that this tentative mode of formulation (in terms of “varieties” rather than “kinds”) is particularly concerned to leave room for (namely, that the apparent “kinds” in question here are, in the end, to be recognized as only apparently distinct kinds). 10 The aptness of these labels for the purposes of distinguishing the two varieties of skepticism at issue here are not meant to turn on any claim to the effect that an interest in the other problematic (i. e., the one that does not bear his name) is absent from the writings of either Descartes or Kant. So it does not imply a denial that Kant was interested in Cartesian skepticism. As we shall see later on, Kant addresses a variant of Cartesian skepticism (he calls it “problematical idealism”) and seeks in “The Refutation of Idealism” to show how the proper treatment of (what he himself calls) “skepticism” contains as one of its corollaries the untenability of all such forms of idealism. Nor does it turn on a denial that there are incipient forms of a Kantian problematic to be found in Descartes’s writings. (Though I do think that such a problematic never comes fully into view in his pages as a full-blown, self-standing variety of skepticism). For some discussion of the presence of an incipiently Kantian problematic in Descartes’s thought, see Conant (1992). The evil demon hypothesis of The First Meditation (unlike the dreaming hypothesis), if pressed hard enough, does unfold into (what I call) a Kantian skeptical problematic. The most vivid appearance in the Cartesian corpus of such a problematic is to be found in the treatment of the problem of the creation of eternal truths. But such a problematic is pervasively present in the Cartesian corpus (albeit very incipiently) in the treatment of the fundamental Cartesian issue of what it is for an idea to possess objective reality.

8

James Conant

will call) Cartesian skepticism. I will then go on to present five examples of Kantian skepticism.

II. Five Examples of Cartesian Skepticism I begin with the classic case: Cartesian skepticism about perception. The important thing to see here is that this case represents just one instance of a far broader genre: 1. Cartesian skepticism about perception. How can I know things are as my senses present them as being? Is there really an external world? I am having an experience of a certain sort (say, that I am here in my dressing-gown, sitting by the fire) but how can I know that things are as my experience presents them as being? I can have experiences that are indistinguishable from this one (in which I appear to be here in my dressing-gown, sitting by the fire), such as when I am dreaming, and yet things are not as they appear. The case under consideration is a best case of knowledge, and yet there still seems to be room for the question: How can I know that I am not, in fact, lying undressed in my bed dreaming that I am here in my dressing-gown, sitting by the fire? For there are no marks or features that allow one conclusively to distinguish waking from dreaming states. This leads to the following conclusion: if I don’t know this, then how can I be said to know anything? Why should I ever trust the testimony of my senses? Should I ever endorse the appearances with which my senses present me? The gap the Cartesian skeptic regarding perception seeks to bridge is from his own mind to the outer world. The outer world is hidden behind the veil of perception. The paradox lies in our apparent inability to answer the following question: How can I penetrate the veil of sensory ideas and attain a view of what is really happening outside of my mind? In the above case, the knowing subject lies on the mind side of a mind/ world divide. The Cartesian skeptic of perception is left with a problem of the following form: what must I add to (mere) perception to turn it into knowledge? The Cartesian epistemologist’s analysis of our perceptual situation turns here on a particular way of conceiving wherein the distinction consists between a veridical and a non-veridical perception. Both cases are conceived as sharing something in common – as possess-

Two Varieties of Skepticism

9

ing what John McDowell calls a highest common factor.11 Thus the Cartesian epistemologist is drawn to understand the concept of an apparent perception as follows: An apparent perception of X is the component in a subject’s experience which figures in both a veridical and a non-veridical perceptual experience of X.12 This gives rise then to a further question: what is the something extra which a veridical perception has and a non-veridical perception lacks in virtue of which the latter is veridical and the latter merely apparent. This gives rise to the search for the special something which one must add to a merely apparent conception in order to turn it into a veridical perception. Once the problem is conceived in this form it issues in Cartesian skepticism. The question which animates the skeptic in the above case is the following: how do we bridge the gap from mind to world – from inner to outer? In order to separate the essential from the accidental within this genre of skepticism, it is helpful to consider the following case next: 2. Cartesian skepticism about other minds. The person before me is acting for all the world as if he were in pain, but how can I know that he is pain. Is he pretending? Or is he really in pain? The case of someone convincingly pretending to be in pain can be indistinguishable from the case of someone actually being in pain. Thus I am not able to conclude that I know he is in pain. For there are no marks or features that conclusively distinguish a case of someone’s pretending to be in a state from his being in that state. The conclusion here again takes the form: if I don’t know this, how can I ever be said to know what someone else if feeling? The Cartesian problematic here is focused on the problem of how to underwrite the testimony of the human body. The gap the Cartesian seeks to bridge here is from the other’s outer bodily movements to his inner states. The inner world of the other is hidden behind the veil of the body. This version of the Cartesian skeptic asks: How can I penetrate the screen of the other’s body and attain a view of what is really happening inside the other himself ? In this case, the direction of travel across the mind/world divide has been reversed – the task is now to travel from outer to inner, rather than in the opposite direction – but (what I will call) the form of the prob11 McDowell (1998). 12 For further discussion of this point, see Thomas Lockhart’s paper in this volume.

10

James Conant

lematic is the same. The Cartesian skeptic about other minds is left with a problem of the following form: what must I add to (mere) bodily behavior to turn it into an expression of an inner state? The Cartesian epistemologist’s analysis of bodily expressions of our mental life turns here on a particular way of conceiving wherein the distinction consists between a genuine and a merely apparent expression of an inner state. Both cases are again conceived as sharing something in common – as possessing a highest common factor. Thus the Cartesian epistemologist of other minds is drawn to understand the case of someone’s pretending to be in pain as follows: The outward bodily pain-behavior is a common component which can figure equally in both a case of someone’s pretending to be in pain and someone’s being in pain. This gives rise then to a further question: what is the something extra which the case of someone’s genuinely being in pain has and someone’s pretending to be in pain lacks in virtue of which the latter can yield knowledge of another mind and the latter cannot. This gives rise to the search for the special something – the inner state as such – which one must add to the psychologically neutral layer of mere bodily behavior in order to turn the latter into a case of genuine pain behavior. Once the problem is conceived in this form it issues in Cartesian skepticism, only now it is the screen of the body (rather than the veil of perception) which stands in the way of our being able to gain knowledge of what we want. What makes the behavior count as genuine pain behavior is something which blankly obtains on the other side of the screen, beyond the reach of the gaze of the subject who seeks knowledge of the other’s mental states. The pain behavior in this case – like the sensory experience of the outer world in the philosophy of perception case – is conceived in terms which render it by its very nature unable to ground a claim to knowledge, for that which it presents us with (mere sensory experience in the one case, mere bodily behavior in the other) is construed in such a way that it, of necessity, falls short of that (the external world in the one case, the other’s inner state in the other) which we seek knowledge. Let us now move to (what is taken to be) a different area of philosophy altogether: the philosophy of language. This is an area of philosophy in which interest in Kantian skepticism is currently far more dominant than interest in Cartesian skepticism. It is worth noticing, however, that one can also identify a problematic here which exhibits a Cartesian form:

Two Varieties of Skepticism

11

3. Cartesian skepticism about language. How can I know that my interpretation of something (a text, an utterance, a sign-post) is correct? How can I be sure that this is what is really meant? I know how this sort of sign (say, a sign-post in the shape, say, of a pointing arrow) is usually to be interpreted, but how do I know that my interpretation in this case is the right interpretation? The physical appearance of the sign (text, utterance) could be indistinguishable across two different contexts of use, and yet the sign can have entirely different meanings in those two different contexts. Thus to understand the sign, I must first interpret it, and it is always possible that my interpretation is incorrect – that it applies to the one context, but not the other. Hence I cannot conclude that I know what the sign here means. But if I don’t know this, how can I ever be said to know what something means? The gap that the Cartesian seeks to bridge here is between his understanding of the meaning of a sign and what the sign actually means. The actual meaning of the sign is never present to our view, but only our interpretation of it. This version of the Cartesian skeptic asks: How can I penetrate the penumbra of interpretations which stand between me and the sign and attain knowledge of the meaning of the sign? The foregoing problem has far more currency presently in English Departments and Law Schools and Art History Departments – that is, in forms of academic and intellectual inquiry which must arrive at a determinate understanding of a particular stretch of text or picture – than it currently has in Philosophy Departments, where the practical task of achieving an understanding of something’s meaning can be sacrificed to the broader question of whether such understanding is so much as possible. What matters for our present purposes is that it is a form of problem which can arise. The Cartesian skeptic of language is left with a problem of the following form: what must I add to a mere string of signs in order to turn it something which can support knowledge of their meaning The Cartesian epistemologist’s analysis of our linguistic situation turns here on a particular way of conceiving wherein the distinction consists between a veridical and a non-veridical understanding of meaning. Both cases are conceived as sharing something in common – a highest common factor – in this case the particular set of signs which can support differing interpretations. Any case of understanding the meaning of a sign is then analyzed as a case of a mere sign plus an interpretation (on the model of a mere sensory experience plus an outer fact

12

James Conant

which obtains without in the world, or a mere bodily experience plus an inner state which obtains within the subject). Thus the Cartesian linguist is drawn to understand the case of someone’s misunderstanding a linguistic performance as follows: The outwardly perceptible physical sign is a common component which can figure equally in both a case of someone’s misunderstanding the sign and a case of someone’s understanding the sign. This gives rise then to the further question: what is the something extra which must be added to the physical sign, considered in itself, so that the conjunction of the sign plus that extra element can yield knowledge of meaning. This gives rise to the search for the special something – the criterion of meaning – which one must add to the semantically neutral layer of mere linguistic noises or marks in order to turn the latter into a case of genuinely meaningful linguistic behavior. It is also worth noticing that problems of a Cartesian form arise no less in practical philosophy than in theoretical philosophy. Consider the following case: 4. Cartesian skepticism about intentional action. The person before me appears to be raising his hand, but how can I know that he is raising his hand. The movement of his limbs is certainly indistinguishable from what it would be if he were raising his hand. But it is possible for a human hand to shoot up involuntarily, without the subject acting upon an intention to raise it. And these two sorts of cases – the case of a mere bodily movement and the case of an intentional action – can be indistinguishable from one another. So how can I know that the other is really acting on an intention, i. e., that is his movements are genuine expressions of agency, rather than the mere effects of biological (or psychological, or some other sort of) causes beyond his control. The gap that the Cartesian seeks to bridge here is between a merely bodily movement and the intention that lies beneath (or behind) it. This version of the Cartesian skeptic asks: How can I get beyond an intervening set of bodily events (whose etiology is explicable in, say, purely physiological terms) and arrive at knowledge of the intentions which gave rise to them? The Cartesian skeptic about intentional action has a problem of the following form: what must I add to a mere bodily movement in order to turn it something which can support knowledge of intentional action. The Cartesian epistemologist’s analysis of the situation of the agent turns here on a particular way of conceiving wherein the distinction

Two Varieties of Skepticism

13

consists between a bit of genuine agency and a mere bodily movement. Both cases are again conceived as sharing something in common – a highest common factor – in this case the mere bodily movement. Any case of intentional agency is then analyzed as a case of a mere bodily movement being preceded by an intention which is then assigned to it as its cause. Thus the Cartesian philosopher of action is forced to grapple with the question of what is left of an intentional action once we subtract the mere bodily movement (on the model of what is left of a case of genuinely veridical sensory perception once we subtract the mere appearance, etc.). This triggers the search for the suitably connected something extra which must be linked to the bodily movement, so that the conjunction of bodily movement supplemented with that extra element can yield a case of intentional action. It is instructive to consider a final case of the Cartesian problematic, this time from the field of aesthetics. Skeptical problems in the realm of aesthetics are particularly interesting because the feedback loop between theory and practice in the contemporary sphere of the arts is of a sufficiently direct nature that it is possible to find any number of cases in which instances of the practice of art itself has come to reflect the philosophical assumptions implicit in a skeptical understanding of what art is. Philosophical reflection in this sphere is no longer on a set of practices for making knowledge-claims whose self-standing nature as a practice is able to continue unaffected by the character of such forms of reflection.13 The following might therefore strike one as equally aptly expressing either a general (and thus contentious) philosophical conception of what art is or a fairly banal description of the self-understanding implicit in certain forms of avant-garde artistic practice: 5. Cartesian skepticism about art. While working at home, I have accidentally turned over a can of paint, creating a pattern of splatter across a piece of canvass that happened to be lying on the ground below the can. From one point of view, what we now have is a waste of some perfectly good paint and a perfectly good canvass. But, let us imagine that downtown, in the Museum of Modern Art, there happens to be hanging on the wall a canvass, sensuously indistinguishable from the one before me, but it bears the caption “Splat13 Similar examples of such a feedback loop between epistemic practice and skeptical theory could be adduced from other areas of human life, such as, for example, from the relation between the practice and theory of law.

14

James Conant

ter, Artist: James Conant”. The object in the museum is declared by the curators of that institution to be a work of art. From an epistemological point of view, apart from their respective locations, one in my home and one in the museum, the two objects seem to be indistinguishable from one another. How can I tell which of them is a mere accident and which is a work of art? Andy Warhol’s Campbell’s Soup Can (or, if you prefer, Marcel Duchamp’s Urinal, to take the other favorite example of philosophers of art) seems to be an example of an artifact indistinguishable from one which I might encounter in my own home. But if I cannot tell what distinguishes what is my home from what is in the museum in such a case, then how can I ever claim to know whether something is really a work of art? The gap that the Cartesian seeks to bridge here is one between objecthood and art – between the mere physical appearance of the work of art and its status as a locus of aesthetic value, as something that invites and merits the sorts of interest and assessment appropriate to works of art. So this version of the Cartesian skeptic asks: How can I get beyond the mere objecthood of the work (whose characteristics are describable in purely physical, aesthetically neutral terms) to the art?

III. Five Examples of Kantian Skepticism Again, I begin with the classic case of the genre: (what I call) Kantian skepticism about perception. Again, as before, the first task here is to come to appreciate how the case given below represents just one instance of a far broader genre: 1. Kantian skepticism about perception. How can my senses so much as present things as being a certain way? How can my experience so much as be intelligibly of an external world? The Kantian problematic in philosophy of perception is focused on the problem how the senses must be so as to be able to furnish testimony. An outer object’s impinging on the senses would appear, as such, to be a mere transaction in nature, and, taken in and of itself, not to be the sort of item that is “about” anything, let alone the sort that ought to provide anyone with a reason for believing anything. What sort of unity must an episode of sensory experience possess in order to be able to present an appearance about which the question could arise “Shall I

Two Varieties of Skepticism

15

endorse it”? The Kantian paradox lies in its coming to seem a mystery how what impinges on my senses could so much as appear to be revelatory of the world. How am I so much as able to enjoy an experience that possesses a determinate world-directed content (e. g. that I am here in my dressing-gown, sitting by the fire)? The gap the Kantian seeks to overcome is from sensory blindness to sensory consciousness – from a form of sensibility upon which things merely causally impinge to one upon which things impress themselves as being thus and so. Here what was previously taken for granted by the Cartesian, namely, that there is a determinate way which things appear to me to be in perception, has come under question. The question now is how such an appearance is so much as possible. If we think of the Cartesian question in the philosophy of perception as focusing on the topic of how I can know which of my perceptions are veridical, then it can help to see the issue in Kantian skepticism as turning on the topic of how one of my perceptions can so much as be non-veridical. The very possibility of non-veridicality or falsity in thought, judgment, experience, requires the possibility of an intentional relation between the thought, judgment, experience and reality; the Kantian question is how such a relation is so much as possible. Under the pressure of a parallel Kantian question, each of the further following four counterparts to the examples of Cartesian skepticism which we encountered above can be identified in the four further areas of philosophy we considered above: 2. The problem of other minds. This version of the Kantian skeptic is preoccupied by the following question: How can the human body so much as seem to express a mental state? A human body is, as such, nothing but so much bio-matter – i. e., a kind of substance whose possible states, taken in isolation, are without psychological purport. So how can any particular state of such a body itself possess the sort of significance we attribute to it when we take ourselves to see that someone is in pain, or happy, or angry. The Kantian paradox here lies in its coming to seem a mystery how an expanse of fleshy matter could so much as appear to be revelatory of an inner life. The Kantian problematic here is focused on the question: How does the human body even seem to furnish a picture of the human soul? The gap the Kantian seeks to overcome here is from an inexpressive physical entity to an animated field of human expression – from a

16

James Conant

psychologically-neutral locus of bodily movements to the communicative body of a palpably suffering, desiring, pondering human being. 3. Philosophy of language. This version of the Kantian skeptic is preoccupied by the following question: How can a sequence of marks or noises so much as seem to mean something? Marks and noises are, as such, mere physical entities, devoid of any semantic content. How could such entities, in and of themselves, ever be the sort of things that mean something? The Kantian paradox here lies in its coming to seem a mystery how a mere sequence of dead signs could so much as appear to be alive with significance. The Kantian problematic here is focused on the question: How does a linguistic performance acquire the physiognomy of meaning? What sort of unity must a linguistic performance possess in order to appear to be the sort of thing about which the question could arise “Is this what it means”? The gap the Kantian seeks to overcome here is from meaningless sequences of marks and noises to determinate expressions of thought – from a semantically neutral concatenation of scratches or sounds to a legible field of intelligible meanings. 4. The problem of intentional action. This version of the Kantian skeptic is preoccupied by the following question: How can a mere bodily movement, qua merely physiological event, ever be the expression of an intention? The bodily movement is an event in the natural world and, as such, the effect of purely physiological causes. How can there be any sort of internal relation between such a transaction in the natural world (which is, taken in itself, explicable in purely non-intentional terms) and something that supposedly happens in the mind (an event of an intrinsically intentional character)? The Kantian paradox here lies in its coming to seem a mystery how mere bodily movements could so much as possibly seem to be expressions of genuine full-blooded agency. The Kantian problematic here is focused on the question: How does the execution of a sequence of such movements acquire the physiognomy of an intention? What sort of unity must such a sequence possess in order to appear to be the sort of thing about which the question could arise “Is this what he intended to do”? The gap the Kantian seeks to overcome here is from the movements of muscles, tissues, and limbs to determinate expressions of human willing – from an inten-

Two Varieties of Skepticism

17

tionally neutral concatenation of jangling bodily appendages to a legible field of purposeful human action. 5. The philosophy of art. This version of the Kantian skeptic is preoccupied by the following question: How can a mere configuration of clay or bronze, or paint on canvass, or sequences of tones, ever so much as seem to be a structure of aesthetic significance and value? Are not such configurations of matter or sound, as such, mere physical shapes or noises which are, considered in and of themselves, devoid of aesthetic value? However fantastically improbable it might be, couldn’t, in principle, any such configuration come into existence simply as the consequence of a bizarre sequence of physical events? And, as such, wouldn’t such a mere effect of natural causes, absent some further intervening moment of human activity (which aspired to confer significance upon the configuration in question), remain something less than a work of art? How can such mere configurations of matter and sound ever be the sorts of thing that are properly supposed to possess the features we take ourselves genuinely to be able to discover in works of art? The Kantian paradox here lies in its coming to seem a mystery how a mere sequence of shapes and sounds could so much as appear to be alive with aesthetic significance. The Kantian problematic here is focused on the question: How does an object acquire the physiognomy of aesthetic depth? What sort of unity must it possess in order to appear to be the sort of object about which the question could arise “Is this an original work or merely derivative?”, “Does it move me (or speak to me or provoke me) or just leave me cold (or chagrined, or disgusted)?”, or even just “Does this object merit this sort of critical interest and attention”? The gap the Kantian seeks to overcome here is from aesthetically inert constellations of matter and sound to objects that are properly taken to invite and support forms of aesthetic appreciation and criticism – from artistically neutral chunks of stuff or noise to delightful or provocative exemplars of beauty and sublimity. Each of the five varieties of skepticism above involves a problematic of a similar form – (what I call) a Kantian form. Identifying a particular variant of a skeptical problematic as being of this form is a matter of recognizing it as sharing a structure with each of the variants of Kantian skepticism presented above. Beginning with the immense amount of attention paid to the third case – Kantian skepticism about language –

18

James Conant

which gained prominence in the 1970 s and 1980 s, over the past few decades, this has increasingly become the form which so-called “skeptical” problems in philosophy now tend to exhibit.

IV. Homology of Structure of the Variants within a Variety of Skepticism I could go on and multiply pairs of examples of these varieties of skepticism from other areas of philosophy. I take it that variants of both Cartesian and Kantian skepticism arise in ethics, aesthetics, philosophy of law, philosophy of mind, philosophy of science, philosophy of mathematics, etc.14 The foregoing five pairs of cases should suffice, however, for the purpose of furnishing a preliminary overview of the twin problematics that I am seeking to isolate here. This brings us to the first form of philosophical insight that a perspicuous overview of various kinds of skepticism affords. It allows one to command a clearer view of the sorts of relations of symmetry and asymmetry that obtain among variants within a single variety of skepticism. If one commands a clear view of the homology of structure exhibited, for example, across the three variants of Cartesian or Kantian skepticism mentioned above, then one puts oneself in a position to bring re14 I say “variants” of each of these varieties of skepticism arise in each of these “areas” of philosophy (rather than “a variant” of each arises) because in “areas” such as ethics and philosophy of science – “areas”, that is, that involve a tangle of different sorts of philosophical problem – a multiplicity of variants of each variety are to be found. Indeed, in both ethics and philosophy of science, for example, a version of each of the philosophy of perception variants and each of the philosophy of language variants of skeptical problematic (along with a great many others) are to be found. This furnishes yet a further reason why one ought to be suspicious of the idea that one should be able to limn the skeleton of philosophy at its joints merely by effecting a division into areas based solely on differences in subject-matter – ethics, science, mathematics, etc. An extended discussion of the topic of the unity of philosophy is well beyond the scope of this paper. It is, nonetheless, a topic that should be on the mind of any reader who thinks that there is little intellectual cost to the increasing specialization of philosophy into supposedly distinct sub-areas (such as philosophy of science, philosophy of language, moral philosophy, etc.) – especially one who thinks that each of these areas requires a specialized form of professional expertise that can be largely acquired in relative ignorance of what goes on in other areas of philosophy.

Two Varieties of Skepticism

19

sources derived in the consideration of one of these variants to bear on the consideration of others. It is quite remarkable, for example, how many a philosopher today is clear that, whatever else she wants to do in philosophy, she wants to find a way to resist the Cartesian assumption in the philosophy of perception that all perception of external objects requires an inference from how things seem to how things are. She wants to avoid such an assumption because she realizes she will then be saddled with a Cartesian gap (leaving herself sealed inside her own mind, unable to claw her way back out to an unobstructed glimpse of the external world). Yet this same philosopher, when she turns, say, to the philosophy of language yields to the corresponding Cartesian temptation without a pang – finding utterly innocent the assumption that all understanding presupposes interpretation – thus saddling herself with a gap of a homologous sort (leaving herself sealed within a horizon of interpretations, unable way to claw her way back out to an unobstructed grasp of the meaning of an expression). The first step in learning to attain an overview of the problems in this “area” of philosophy known as “the problem of skepticism” is to learn to achieve a sensitivity to the forms of homology in structure that supposedly distinct philosophical problems in supposedly distinct areas of philosophy exhibit. One must learn to discern when and how the form of the problem has gone unchanged, even though the clothing in which the problem is dressed up has been changed.

V. Is Kantian Skepticism Really a Form of Skepticism? I purported, in originally furnishing ways one might unpack the difference between Cartesian and Kantian skepticism, to be offering various ways of unpacking the contrast between a problematic centered on knowledge and one centered on the conditions of knowledge. This way of putting the contrast makes it seem as if what were at issue were two forms of epistemological worry. As, however, should already be evident from the foregoing, this is quite misleading. It is, indeed, constitutive of the Cartesian problematic that it be clothed in epistemological form. The focus in the Cartesian problematic is on knowledge claims. The Cartesian problematic brings the relation between our knowledge claims and reality into question. What makes someone such-and-such a sort of Cartesian skeptic is that he is exercised by a doubt regarding whether we can have knowledge of such-and-such a sort. Given the in-

20

James Conant

ternal relation between the concepts of doubt and knowledge, the Cartesian skeptic is quite aptly characterized as someone who doubts. What figures in the Cartesian problematic as a worry about the relation between knowledge claims and reality comes to look, however, from the vantage point of the Kantian problematic, like only an instance of a more general worry, a worry about the relation between any claim (true, false or fantastic) and reality. In making a claim at all, whether or not one thereby takes oneself to be knowledgeable, one makes oneself answerable to how things are. The Kantian asks: How is one able to accomplish this feat? The Kantian problematic is therefore only optionally clothed in epistemological form. It is, at its root, of a more general nature. The worry that exercises such a skeptic is misunderstood if it is taken to turn exclusively on matters having to do with knowledge. What comes into view in a Kantian problematic under the initial heading of “the conditions of the possibility of knowledge” are the conditions of the possibility of mindedness as such. Kant himself, with some frequency, alternates between characterizing the sorts of conditions at issue here as conditions of the possibility of knowledge and as conditions of the possibility of experience. (And he implies that they might equally aptly be characterized as, among other things, conditions of the possibility of objectively valid judgment, and conditions of the possibility of sensory consciousness of an object.) The Kantian skeptic is therefore only inaptly characterized as someone who suffers from a doubt. And this will seem to some philosophers to constitute a sufficient ground for insisting that what I am calling “a Kantian skeptic” is not a kind of skeptic at all. So let me be clear on the following point: My aim here is never to legislate how the word ‘skeptic’ should be used, but only to illuminate some of the diverse ways in which philosophers, in fact, often use it – with the eventual aim of permitting the formulation of the following question: how are these various problematics (each of which is often called one of “skepticism”) related? 15 15 Since it will seem to some philosophers simply perverse of me to use the term ‘skepticism’ in connection with what I am calling “Kantian skepticism”, given my present insistence that what is at issue au fond in Kantian skepticism is not an epistemological worry, perhaps I should briefly say, by way of defense, in anticipation of the remainder of this paper, the following: if one insists upon restricting the application of the term ‘skepticism’ to overtly epistemological contexts, one is going to be unable to track much that is at issue when the term is employed by Kant, by Wittgenstein, and by many contemporary philosophers. Consider, for example, the debate about the rule-following considerations and

Two Varieties of Skepticism

21

Despite having just now acknowledged a perfectly good reason for wanting to resist the invitation to call (what I have been calling) “Kantian skepticism” a variety of skepticism at all, I will continue thus to refer to it. As I go along, I will touch upon some of the reasons why some philosophers have been drawn to project the term ‘skepticism’ into contexts exhibiting the earmarks of a Kantian problematic and why I think there is a point to following their lead. As will become clear, whether a given philosopher will be inclined to look favorably upon such an extension of the term is not a matter that can be adjudicated on philosophically neutral grounds: which fork in the terminological road one finds less philosophically obfuscating here will depend in large part upon one’s collateral philosophical commitments.16 Indeed, the question of whether Cartesian and Kantian skepticism, as I have been allowing myself to call them, are properly conceived as the threat of so-called meaning skepticism. What is often at issue in that debate is not just how one can know what something (or someone) means, but how it is so much as possible to mean anything at all. The term ‘skepticism’ in that debate often names the paradox that ensues if we cease to be able to make sense of the phenomenon in question as something that is so much as possible. What threatens to lapse here is not just our epistemic access to meaning but the very being of meaning. In reply to this, someone might want to insist that, if that is so, then that is a good reason for not employing the term ‘skepticism’ in the context of that debate. Perhaps so. My point, at the moment, is simply that if one wants to track how the term is often used by many philosophers, one needs to see that it sometimes ranges over philosophical contexts wider than the merely epistemological. Admittedly, this oscillation between a narrowly epistemological and a broader Kantian use of the term makes for many confusions; and it is natural to think the short way to avoid such confusions is to initiate an act of linguistic legislation that restricts the permissible use of the term. But it is too early in our inquiry to adjudicate how the use of the term ought to be restricted, if at all. In order to see how the term is best used, first we need to see more clearly how it is used and why it has come to admit of the variety of uses it presently does. 16 To be more precise: I do not think any particular collateral philosophical commitments need be in place in order to distinguish between those philosophical problematics that bear (what we can call – leaving the term ‘skepticism’ to one side for a moment) “Cartesian features” and those that bear “Kantian features” respectively. A philosopher’s willingness to look upon these as equally instances of something that is properly termed a skeptical problematic, however, may turn on such commitments, especially if what is at issue is the following more pointed question: To what degree is the term ‘skepticism’ across its employment in Cartesian and Kantian contexts to be understood univocally (rather than as a mere homonym)?

22

James Conant

two species of a broader philosophical genus of problem (so that the term “skepticism” therefore denotes some fundamental dimension of philosophical concern of which they both partake) or whether the Cartesian and Kantian case, as described above, have little to do with one another as forms of philosophical problem (so that the term “skepticism” is really just a homonym when employed in these two cases) is itself a philosophical question. One of the aims of this essay is to attract interest in this question. But we will only be in a position to pose this question, regarding the relation between these two philosophical problematics, once we have each properly in view. It is therefore in order that we first introduce some further conceptual tools which permit us to identify when we are faced with the one problematic and when with the other, and when perhaps with both or neither.

VI. Some Features of the Cartesian Genre of Skepticism In order to allow for a more fine-grained discrimination of these varieties of skepticism, I will now proceed to distinguish nine generic features of Cartesian and Kantian skepticism respectively.17 But before I do this, let me caution that a considerable element of idealization is involved in the idea of a philosophical problematic characterized by either of the sets of features adumbrated below.18 There are two independent dimensions of idealization at work here. The first dimension lies in the idea that a fully realized variant of, say, Cartesian skepticism will exhibit all nine of the mentioned features. It is an idealization because most 17 The point of specifying such features is to furnish criteria for determining which sorts of philosophical rehearsals count, according to the taxonomy introduced here, as cases of one or the other of these varieties of skepticism. In distinguishing these nine features, I do not mean to be doing more than offering a rough and ready characterization of the differences between these two varieties of skepticism. In particular, I do not mean to claim either that each of these features is equally distinct from the other eight or that this set of nine features exhausts the significant features that characterize either of these varieties of skepticism. (Indeed, I think both of these claims are false.) But I do hope that these features suffice to allow one to inspect a philosophical discussion of some topic and determine whether it qualifies as participating in one or the other of the two varieties of skeptical problematic at issue here, and, if so, which one. 18 As I employ the terms, ‘Cartesian skepticism” and “Kantian skepticism’ denote ideal types of philosophical problematic that are instantiated with varying degrees of faithfulness in any given actual philosophical discussion.

Two Varieties of Skepticism

23

philosophical discussions that exhibit any one of the features (drawn from one of the two above lists of features) will not necessarily exhibit all of them (though it will probably exhibit many of them). Thus it should be understood that, henceforth, when I speak of “a Cartesian problematic”, I will mean a philosophical discussion that exhibits at least a large number of the Cartesian features specified below, but not necessarily all of them. (The more of these features a given philosophical under discussion exhibits, the more Cartesian the problematic in question is.) And, similarly, when I speak of “a Kantian problematic”, I will mean a philosophical discussion that exhibits at least a large number of the Kantian features specified below. (The more such features a philosophical problematic under discussion exhibits, the more Kantian it is.) 19 The second dimension of idealization concerns the way in which the presentation of these feature-spaces may appear to preclude the possibility of their overlap – to preclude an intermingling of Cartesian and Kantian features within a single philosophical problematic. That is both a desirable and a potentially misleading aspect of the presentation that follows. It is desirable in as much as it is part of my aim to illuminate something about the internal structure of each of these two (admittedly idealized) varieties of philosophical problematic. Each has its own logic. The co-occurrence of such features within a single philosophical discussion is generally (though not necessarily20) a symptom of philosophical confusion on the part of an author; and it is part of my purpose to facilitate the diagnosis and treatment of such forms of confusion. Nevertheless, it is potentially misleading because – even if, in a resolutely executed philosophical inquiry, Cartesian and Kantian features will tend to drive one another out – in the irresoluteness of actual philosophical practice, such features can often be found squashed up against one another.21 Borrowing some terms from Cavell’s theory of genre22, one 19 I, do not mean hereby to rule out cases of philosophical discussion that do perfectly exemplify, without blemishes, either the Cartesian or Kantian format respectively, but only to indicate that they are surprisingly rare. 20 For it not to be a symptom of (at least some degree of) confusion on an author’s part requires, I think, that the author already have thought through – and thus have come to some stable view of his own concerning – the relation between these two varieties of skepticism. There are such authors; but not many. 21 For example, in his book Kripke (1982), Saul Kripke initially presents his motivating example in terms that alternate between a Cartesian worry (“How can I know whether in the present case someone is adding or quadding; and if do not know this, then [given that this would appear to be a best case of knowledge of

24

James Conant

might say that Kantian skepticism represents an adjacent genre of skepticism – one in which each of the features of Cartesian skepticism is displaced 23 in a certain way. And, just as different genres of film (say, a western and a romantic comedy) can be crossed with another – often (though not always) with aesthetically jarring results – so, too, these genres of skepticism can be crossed with another. In philosophy, when this happens, usually (though not always) it is a sign that the author is no longer clear which of these two philosophical problematics he wishes to inhabit. The pair of reasons for calling the first set of features “Cartesian” and the second “Kantian” is perfectly consistent with noting the presence of a problematic in Descartes’s writings exhibiting some of the features of the second sort (and such a problematic is present in Descartes) and with noting the presence of an aspiration in Kant’s writings to address a problematic exhibiting all of the features of the first sort (and such an aspiration forms a crucial part of Kant’s project). I denominate these problematics “Cartesian” and “Kantian” respectively, in order to mark not the point of their earliest philosophical inception (the moment at which the seeds of the problematic first began to blossom philosophically) nor their last philosophical flicker of life (the moment past which addition] how can I ever know if anyone is really adding?”) and a Kantian worry (“How can there ever so much as be a fact of the matter as to whether someone is adding; and, if there cannot then [given that this would appear to be as elementary a case as there can be of someone meaning one thing rather than another] how can anyone ever so much as determinately mean anything?”). This alternation between Cartesian and Kantian idioms has confused some of Kripke’s commentators (who equate skepticism with Cartesian skepticism and therefore cling to the Cartesian formulation because it strikes them as more properly skeptical), thus obscuring their view of the problematic that is at the center of Kripke’s concern and leading them to waffle alternately between the Cartesian and Kantian variants of skepticism in philosophy of language distinguished above, without ever noticing that they thus waffle. Kripke himself, however, is – at least in certain places in his book (see, for example Kripke (1982), 21) – fairly clear that ostensibly Cartesian characterizations of the issue (according to which, as he says: “the problem may appear to be epistemological”) can serve him only as a provisional expository device (thus he says: “the ladder must finally be kicked away”) for leading people into a skeptical paradox of an altogether different and more fundamental variety than that with which we appear to be presented if we restrict ourselves to such Cartesian characterizations. 22 See his Cavell (1981) and his Cavell (1996). 23 Cavell’s term for this is negated.

Two Varieties of Skepticism

25

they cease to have philosophical currency), but rather the historical moment at which their overall philosophical shape first became visible (the moment at which the problematic first reaches full philosophical flower). I denominate the first problematic “Cartesian” because it is in Descartes’s writings that it comes fully into view for the first time as one possessing all nine of the features specified below as Cartesian; and I denominate the second “Kantian” because it is in Kant’s writings that it first comes fully into view as one possessing all nine of the features specified below as Kantian. With these cautionary remarks to be borne in mind, I will henceforth refer to the following nine generic features of the Cartesian genre of skepticism as Cartesian features: (1) The Cartesian investigation begins with and turns on the exploration of a certain sort of example – a best case of knowledge. (2) Such a case is shown to be vulnerable to doubt. (3) The conclusion generalizes – we can move from a conclusion about this particular candidate item of knowledge to a general conclusion about all such items. (4) The investigation thereby issues in a discovery. (5) The investigation ends in a mood of disappointment. (6) The disappointment is born of the impossibility of showing how what we had taken to be possible could be actual. (7) It looks as if there is something we cannot do. (8) Our inability is the consequence of the existence of a Cartesian gap – a gap we seem to be unable to bridge. (9) The skeptical discovery cannot be converted into practice: it is practically unstable – we are obliged to live as if we could bridge the gap in question. First feature. To say that an exploration of the Cartesian problematic begins always with “a best case of knowledge” is to say that it begins always with a carefully selected example – one that possesses (or at least appears able simultaneously to possess) at least the following four characteristics. First, the example must involve a concrete claim to know: a particular person, at a particular time and place must enter the claim. Second, the claim in question must be able to serve as an exemplar of an entire class of claims: it must be sufficiently representative so that each of us can rehearse a version of such a claim in the privacy of our own epistemological closet. Third, the claim must be directed at an unremarkable object – the sort of object which requires no special sort of expertise in

26

James Conant

order to be able to tell one when you see one.24 Thus (what one might be tempted to call) “a kind of object” figures in Cartesian examples – a hand, a tomato, an envelope, a chair (but never a nine iron, an M-16 rifle, a Japanese beetle, a goldfinch, or a bubble chamber). One must begin with the right sort of example, in order to get the Cartesian problematic up and running. (Stanley Cavell suggests we call the sort of object that figures in such examples a generic object. We will return to this suggestion.) Fourth, the object must be encountered under optimal conditions: in good lighting, at short range, for an extended period of time, etc. An important part of exploring the Cartesian problematic lies in exploring the character and the legitimacy of the examples employed to introduce a Cartesian skeptical recital and ascertaining whether they really do simultaneously possess all four of the required characteristics. A preoccupation with such examples is a hallmark of the Cartesian problematic. Second feature. The Cartesian paradox results from the susceptibility of the existence of such an object to doubt. The Cartesian investigation initiates an inquiry into the question what (in our experience of the relevant phenomenon) does and does not admit of the possibility of doubt. It thereby seeks to decompose our experience into two sorts of elements: those which are inherently indubitable and those which involve some (even if perhaps minimal) element of risk. The viability of the ensuing doubt lies in the disclosure of a reliance (on the part of the candidate claim to knowledge) upon those strata of our experience involving this identifiable element of risk. An exploration of what is (and what is not) susceptible to such forms of doubt is a hallmark of the Cartesian problematic. And, under the pressure of the Cartesian investigator’s demand for certainty, it leads to the discovery that very little in our experience is invulnerable to such forms of doubt. Third feature. The possibility of such a discovery depends on the capacity of the initially selected example to generalize. If we don’t know this, then we don’t know anything. That is, in the disclosure that this particular claim to knowledge is vulnerable to doubt, we do not merely 24 The Cartesian skeptical recital always begins by inviting us to direct our attention at a hat, a hand, a chair, a tomato, an envelope – something with regard to which we are all equally experts. As Austin notes, the Cartesian philosopher, like the magician, begins by asking us first to be sure that what we have before is nothing other than a perfectly ordinary hat, a perfectly ordinary chair, etc. (Austin (1979), 87).

Two Varieties of Skepticism

27

take ourselves to learn that we are not able to know a particular thing that we might have thought we did know. (“Oh, O.K., I guess I don’t know that is a goldfinch after all.”) Rather we take ourselves to learn something about knowledge as such – or at least about a whole class of knowledge claims. (“If I don’t know that there is a tomato in front of me right now, then how can I be said to know anything – or at least anything based on the testimony of my senses.”) The vulnerable of our initially selected example to doubt seems, at one and the same time, to disclose the vulnerability of vast portions of our supposed edifice of knowledge. Thus Descartes does not conclude: “Well, then I don’t know that I am sitting here in my dressing gown by the fireplace, after all.” Rather Descartes finds himself threatened with a far more general conclusion of the following sort: “Well, then I can never know that things are as my senses tell me they are.” An apparently irresistible yet extraordinarily precipitous step of this sort – moving from a finding about a particular claim to knowledge to a general conclusion about the very possibility of knowledge of a certain sort – is a mark of entanglement in a variant of the Cartesian problematic. Fourth feature. The possibility of such a doubt seems to yield a discovery. The skeptical discovery takes the form of seeing through the surface of our practices to how they really are. For the results of the Cartesian investigation seem to stand in sharp conflict with our ordinary ways of talking and living. Our practices of entering knowledge-claims seem to be fundamentally at odds with what it is that the investigation shows we are really entitled to claim. So, even if the Cartesian inquirer finds himself unable to do other than to continue to speak with the vulgar and to participate in ordinary ways of speaking and acting, nonetheless, in the light of his discovery, these practices must now seem to him to rest upon a tissue of illusion. To the extent that the Cartesian acquiesces in the conclusion his investigation seems to force upon him, he thereby takes himself to be able to see more clearly and deeply into the true nature of these practices than the majority of his fellows who unreflectively participate in them. The attainment of a discovery that seems to unmask our practices in such a manner is symptomatic of entanglement in a variant of the Cartesian problematic. Fifth feature. The initial thrill of discovery gives way to a mood of disappointment with knowledge. There no longer seems to any way for knowledge (or, within a given variant of Cartesian skepticism: knowledge of a certain sort) to live up to its name. And this gives rise to a mood of disillusionment. One takes oneself to have an understanding

28

James Conant

of what would have had to have been the case for knowledge (or knowledge of a certain sort) to have been possible. But it transpires that, at least for beings such as ourselves, that possibility is not attainable. One is thus left with the feeling that there is something that ought to have been possible but which, as it happens, turns out, at least for us, not to be possible. (The mood of Cartesian skepticism is that of the heroes of Shakespeare’s tragedies: one of disappointment at being fated to live in a world that will necessarily betray one’s trust in it.25) Such a prevailing mood of disappointment or disillusionment as the apparently inescapable response to philosophical inquiry is symptomatic of entanglement in a variant of the Cartesian problematic. Sixth feature. The disappointment lies in our not being actually able to do something that we had always taken to be possible (we are unable to know that things are as they appear). The world we had pre-reflectively taken ourselves to inhabit – a world in which we were capable of attaining knowledge of a certain sort – threatens to turn out not to be the world we actually inhabit. We can formulate thoughts about what such a world would be like and can perhaps still enjoy experiences that purport to be of such a world, but we now are no longer able in our reflective moments to endorse the contents of such thoughts and experiences. Seventh feature. The structure of the situation we inhabit therefore now seems to come into view as one in which we cannot do something we want to be (and pre-reflectively thought of ourselves as fully) able to do. The skeptical discovery seems to disclose a limit to our cognitive abilities – a limit that kicks in at a far earlier point in the cognitive process than we, prior to our philosophical investigation, had any reason to expect. This sense that, in our philosophical inquiry, we run up against such a limit or barrier that we cannot penetrate or circumvent, try as we might, is a mark of entanglement in a variant of the Cartesian problematic. Eighth feature. This inability is taken to be a function of our inability to bridge a certain sort of gap. In the case of philosophy of perception, the gap looms between my inner life and the outer world; in the case of other minds, it opens up between his outer behavior and his inner life; etc. The Cartesian takes himself to have made a genuine discovery in having disclosed the existence of such a gap. Cartesian strategies for 25 The internal relation between Cartesian skepticism and Shakespearean tragedy is explored by Stanley Cavell in the essays collected in his Cavell (1987).

Two Varieties of Skepticism

29

finding a way around the Cartesian paradox involve attempts to find a way to live with this gap. The Cartesian who wants to avoid a skeptical conclusion wants to be able to negotiate his way across the gap. (Descartes himself famously looked to God for some assistance in this matter.) But the gap itself is one he still takes to be just there. As long as one operates within the confines of the Cartesian problematic, the only possible form of “solution” to the problem of skepticism lies in attempting to construct some such sort of bridge across the gap. I will henceforth call the sort of gap at issue here a Cartesian gap. Ninth feature. The Cartesian inquirer will wish to distinguish theoretical doubt from practical doubt. His doubt, he will tell us, is a merely theoretical one. Though, as such, it represents, he will insist, a perfectly intelligible outcome to an intellectual inquiry; nevertheless, he will also insist, conviction in such a conclusion cannot be sustained by someone caught up in the midst of a practical situation. Cartesian skeptical doubt thus possesses an inherent instability due to the impossibility of its realization in practice. Indeed, we are often reassured by philosophers who wish to introduce us to this problematic that the Cartesian inquiry is one that can and should be suspended for practical purposes.26 Such reassurances presuppose the standing possibility of practically abstaining from one’s skeptical surmise and re-embracing the modes of experience, thought and expression thereby brought into question. Such an insistence upon the purely theoretical character of the generality of the form of doubt that is at issue in philosophical inquiry is a hallmark of the Cartesian problematic.

26 For most everyday purposes, we cannot and should not try to do otherwise than to take the perceptual appearances that present themselves to us – say, the sudden appearance of an oncoming car – to constitute genuine instances of perceptual knowledge. The Cartesian skeptic knows this. He knows that we cannot help but take someone writhing in pain to be in pain; we cannot help but take certain ossified habits of interpretation to disclose the meaning of a sign; etc. If he understands the structure of the Cartesian problematic, he will not take such observations to impugn his procedures. Some philosophers have thought – and have thought that later Wittgenstein thought – that such observations themselves could suffice to overturn such forms of skepticism. That is a misunderstanding of Cartesian skepticism – and a misunderstanding of Wittgenstein’s understanding of it. This is not to deny that such observations (or “reminders”, as Wittgenstein preferred to call them) might have a role to play in the treatment of Cartesian skepticism.

30

James Conant

VII. Some Features of the Kantian Genre of Skepticism Kantian skepticism can be differentiated from Cartesian skepticism by noticing that it does not quite share any of the above nine generic features. Kantian genre of skepticism exhibits instead the following nine Kantian features – each of which involves a peculiar displacement or reversal or inversion of the corresponding Cartesian feature: (1) It is constitutive of the sort of investigation into knowledge that it is that it is characterized by a peculiar sort of indifference to the character of the object it takes up as an example. (2) It does not issue in a doubt, but a boggle. (3) The paradox is not the result of moving from a conclusion about a particular object to a general conclusion about all objects of experience, but rather a result of the inability to see how there could so much as be an experience that purports to be of a particular. (4) The investigation climaxes not in a sense of discovery, but one of mystery. (5) This investigation ends in a mood not of disappointment, but of despair. (6) The despair is born not of the impossibility of showing how what we take to be possible could be actual, but of showing how what we take to be actual could be possible. (7) It no longer looks as if there is something we cannot do, now it looks as if there is nothing to do (not even dream) where we had previously thought there was something. (8) The apparent disintegration of this something into a nothing is the consequence of a Kantian gap. (9) The Kantian skeptical surmise is not merely practically unstable, but theoretically unstable qua surmise. First feature. To say that it is constitutive of the Kantian problematic that it is characterized by the absence of a special category of example is to say no more or less than what Kant means to say when he says that transcendental logic, though it does not abstract entirely from objects, is concerned only with the conditions of the possibility of the pure thought of an object: with what it is for thought to be able so much as to be able to have a bearing on the world, what it is for our thought to be related to objects berhaupt. Transcendental logic must abstract from all differences between objects – from what it is to be cognitively related to this rather than that sort of object – but not from relatedness to an

Two Varieties of Skepticism

31

object as such. The two sides of the Kantian problematic, in its classical formulation, turn on how sensibility can yield deliverances that are of objects and how thought can be directed at a (mind-independent) world. The problematic that unfolds here homogenizes the field of possible examples. The questions “What is it to dream that I am in front of a fireplace?” and “What is it to see a fireplace in front of me?” become simultaneously problematized and equally urgent. The questions “What is it to know that I am now in Auburn, Alabama” and “What is it to think of a celestial city?” become equally urgent. The differences between such kinds of example (which play such a crucial role in the context of the Cartesian problematic) cease to be relevant.27 It is no less a problem for the Kantian to understand how we are so much as able to think thoughts that are false than it is to understand how it is that we are able to think thoughts that are true. The examples occurring in explorations of the Kantian problematic therefore often exhibit a curiously schematic character: they lack the concreteness of their Cartesian counterparts – not only in the sense that they come in for less determinate characterization, but also in the comparative lack of specification of the epistemic standing of the claim under investigation or of the cognitive attitude adopted towards it.28 Second feature. Under the pressure of the Kantian question, all our cognitive capacities (the capacity to doubt among them) come to seem equally questionable. Having worked his way far into a particular 27 More precisely: the differences between these sorts of examples cease to be relevant for the bulk of the Kantian inquiry. At a late stage in the inquiry, it will become important to a Kantian inquirer to recover these sorts of differences – differences that underwrite the movement of thought in a Cartesian investigation – but only once the Kantian paradox has been averted and the intelligibility of object-directed thought and experience no longer stands under threat. 28 It does not disrupt the recital of the Kantian problematic – as it would the Cartesian – for the object in question to be of a less than utterly unremarkable sort (a goldfinch can serve the Kantian as well as a tomato) and for no assurances to be forthcoming concerning the epistemic conditions under which the object is encountered. (The lighting, etc., for the Kantian, just has to be good enough to afford some view of the object in question). Early on in the Kantian inquiry, what matters is how we are so much as able to schematize the deliverances of our senses into something that has the determinate character of an appearance – the appearance of, say, a tomato before us. It is a further and comparatively peripheral question, for the Kantian, under what conditions we should or should not endorse the content of such an appearance – and thus judge there to be a tomato before us.

32

James Conant

philosophical dialectic, the Kantian skeptic comes to an impasse: it suddenly no longer seems to him possible that one should be able so much as to frame thoughts that are about the world (or to experience another’s bodily movements as expressions of emotion, or to traffic in forms of words that are replete with meaning, etc.). This sort of skeptic becomes perplexed as to what it is to be experiencing or thinking or meaning things in ways that he also cannot help but take himself to be doing in and through the very act of asking his skeptical question. To move in the direction in which his question leads is apparently to deprive his question (along with the whole of the rest of his “thought”) of the capacity to possess determinate content. And yet he is unable to dismiss his question. It has come to seem intellectually compulsory. So his mind boggles. Such a boggling of the mind, in the face of a looming conclusion that can neither be approached nor avoided – neither fully comprehended nor simply dismissed on the grounds of its incomprehensibility – is a mark of entanglement in a variant of the Kantian problematic. Third feature. What is at issue in the Kantian problematic is the possibility of making claims in general, not knowledge claims in particular. This means that the point of departure for a Kantian investigation is not a particular case of knowledge, but rather the topic of the vulnerability of our thought to reality. This, in a certain sense, reverses the direction of the Cartesian investigation: The Kantian paradox is not the result of moving from a conclusion about the character of our experience of a particular case to a general conclusion about all objects of experience, but rather a result of inability to see how there could so much as be an experience which purports to be of a particular. To respond to the Kantian skeptic is to show how our thinking can have a sort of dependence on the world that allows our activity of making claims to come intelligibly into view as one of claim-making at all. In the absence of an adequate response here, the world threatens to recede from our grasp – to the point where not only do we have reason to fear, with the Cartesian skeptic, that we are unable to know which of our claims about it are true, but rather – to the point where we can not any longer even make sense of the idea that we are able to enter claims about anything of a sufficiently determinate character to be either true or false. This sense of the fading away of the possibility of determinate empirical content in our thought, experience, and discourse is a mark of entanglement in the Kantian problematic. Fourth feature. We can only discover that which we can think. The Cartesian investigation can issue in a discovery, because the Cartesian

Two Varieties of Skepticism

33

takes himself to be able to form a stable conception of that which he discovers we do not have. The Kantian paradox takes the form not of a discovery, but of a mystery. In each of the three variants of Kantian skepticism briefly sketched above, I characterized the Kantian paradox in question as one in which the possibility of a sort of appearance usually taken for granted in the corresponding variant of Cartesian skepticism suddenly comes to seem mysterious. The Kantian paradox regarding perception lies in its coming to seem to me a mystery how what impinges on my senses could so much as appear to be revelatory of the world. The Kantian paradox regarding other minds lies in its coming to seem a mystery to me how the inert fleshy matter comprising his body could so much as appear to be revelatory of his inner life. The Kantian paradox regarding meaning lies in its coming to seem a mystery how a mere sequence of dead signs could so much as appear to be alive with significance. This sense of our ordinary cognitive capacities as being intolerably mysterious – and therefore calling for a philosophical project that relieves our discomfort by providing an account of these capacities that drains them of their mystery – is symptomatic of entanglement in the Kantian problematic.29 Fifth feature. If such a Kantian investigation (into the very possibility of our being able to frame thoughts, enjoy experiences, express meanings, etc.) ends in skeptical paradox, the resulting mood is not one of disappointment – for disappointment (like discovery) is only possible where some glimmering of what it is that one wants (but cannot have) is also available to one. Kant says (concerning what he calls) skepticism, that it is a “way of thinking, in which reason moves against itself with such violence, that it could never have reason arisen except in vçlliger Verzweiflung of achieving satisfaction with respect to reason’s most important aspirations”.30 The violence with which here, in what Kant calls skepticism, reason turns against itself is a violence of the most ex29 Hence, in recent years, the increasingly important relationship between various sorts of philosophical project calling for some sort of naturalistic reduction, on the one hand, and the Kantian problematic in philosophy, on the other. The demand for such reductions in the absence of a plausible program for their execution can, on the one hand, give rise to the relevant sense of mystery; while a philosophically independently fueled sense of Kantian paradox can, on the other hand, fuel the sense that a program of naturalistic reduction is the only possible form of solution to the most urgent philosophical problems that face us today. 30 Kant (1997), 4:271.

34

James Conant

treme possible sort. What reason questions is itself. Our faculty for rational thought arrives at the point where it asks itself (not just how this or that cognitive capacity is possible, but) how it itself is possible, questioning the possibility of the exercise of the very capacity exercised in the framing of such a question. This question is one reason would be driven to pose only if it found itself in a state of vçlliger Verzweiflung – complete desperation or despair – we might say: despair born of desperation. (The mood of Kantian skepticism is that of the heroes of Kafka’s parables: one of bewilderment at the dissolution of the world’s conditions of intelligibility.31) Such a prevailing mood of despair as the outcome of philosophical inquiry is symptomatic of entanglement in a variant of the Kantian problematic. Sixth feature. When reason thus questions its own possibility, the natural history of reason enters a new and radical stage – a skeptical paradox of a different order from the Cartesian is broached. The problem is no longer to understand how something we took to be possible can be actual. The problem is now one that threatens the entire array of cognitive capacities which the Cartesian skeptic takes to be unproblematically available: the capacities to doubt and dream, to feel and think and believe, to enjoy sensory impressions of fireplaces and frame hypotheses about evil demons. The problem now is to understand how something that we take to be actual – for example, the exercise of those cognitive capacities evidently actualized in our philosophical reflections (reflections that themselves seek to address the question of the possibility of such capacities) – can be possible.32 Seventh feature. This collapse of the space of possibilities leaves it looking not – as in the Cartesian case – as if there is something we cannot do. Now it looks as if there is nothing to do (not even to dream) 31 I explore the engagement with a philosophical problematic of this sort in Kafka’s parables in Conant (2001). 32 There is a joke – which I first heard from Dan Dennett – that goes as follows: “Scientists are the sort of people who, once they see that something is possible in principle, want to know whether it is possible in practice. Philosophers are the sort of people who, once they see that something is possible in practice, want to know whether it is possible in principle.” In order to bring out its pertinence to the topic at hand, we might reformulate the joke in even starker terms as follows: “Scientists want to know whether something possible is actual. Philosophers want to know whether something actual is possible.” Dennett intended the joke, I think, as a swipe at a certain sort of philosopher. But the joke neatly sums up an important yet elusive aspect of a certain sort of philosophical problematic – the one at issue in this (sixth) Kantian feature.

Two Varieties of Skepticism

35

where we had previously thought there must be something. The Kantian skeptical discovery, rather than disclosing a boundary which our cognitive abilities run up against, seems to deprive us any territory through which such a boundary might be able to run. This sense that, in our philosophical inquiry, we have found a way to make the Cartesian limit disintegrate (though at the possible cost of ceasing to be able to make sense of our lives) is symptomatic of entanglement in the Kantian problematic. Eighth feature. The Cartesian takes himself to run up against a gap in his philosophizing (a gap between mind and world, between the body of the other and his soul, between interpretation and meaning, etc.). He takes himself to have made a genuine discovery in having disclosed the existence of such a gap. The gap itself is just there. The Kantian also seems to encounter a certain sort of gap in his philosophizing (between sensory blindness and sensory consciousness, between an inexpressive expanse of mere flesh and the animated field of an expressive human body, between meaningless sequences of marks and noises and determinately meaningful expressions of thought, etc.). But it is not clear what it would be acquiesce in the existence of his gap. It must already be bridged (as evidenced by his present ability to exercise his capacities for perception, expression, and thought); and yet, as long as the threat of Kantian paradox has yet to be averted, it also appears that there is no way to bridge the gap. Thus the Kantian problematic tends to be most lucidly adumbrated in the writings of kinds of Kantian who aim to show that where we seem to be confronted with a Kantian gap, we are confronted with only the illusion of a gap. Ninth feature. If, as was said above, the philosophical outcome that now looms is one that threatens the array of cognitive capacities which the Cartesian skeptic takes to be unproblematically available (the capacities to doubt and dream, to feel and think and believe, to enjoy sensory impressions of fireplaces and frame hypotheses about evil demons), then the full import of Kantian skeptical paradox must remain intellectually unschematizable. There can be no such thing as getting “it” fully into focus, for getting things into focus seems precisely to be just a further instance of the sort of thing which we now seem bound to conclude we are unable to do. The Kantian worry is not merely (like the Cartesian one) a form of philosophical perplexity that ceases to be sustainable when the attempt is made to translate it into practice. It cannot even be sustained at the level of theory. The practical possibility of abstaining from the modes of experience and thought and expression

36

James Conant

that are here brought into question is not an option that can present itself even momentarily as a live one. This form of skeptical paradox is therefore not unstable merely in the way the Cartesian one is (i. e., because we cannot sustain our conviction in it, as we leave the closet of our philosophy and immerse ourselves in the practical exigencies of life), the Kant paradoxical surmise already occupies a state of radical instability qua surmise. The Kantian skeptic finds himself drawn to a question he is both unable to hold stably in his mind and yet unable to dismiss.

VIII. The Inflection of Philosophical Vocabulary in Cartesian and Kantian Registers This brings us to the second form of philosophical insight that a perspicuous overview of various kinds of skepticism affords. It allows one to distinguish some of the very different sorts of things philosophers may take themselves to mean when they employ vocabulary such as ‘skepticism’, ‘skeptical paradox’, etc. Moreover, which register – Cartesian or Kantian – a philosopher is operating in will also determine the manner in which a great deal of the rest of his philosophical vocabulary is inflected. Equipped with the preceding overview of Cartesian and Kantian features, it becomes possible to survey the history of twentieth-century analytic philosophy and to notice that philosophers (in their discussions of what they call ‘skepticism’) often tend to be preoccupied with one of these two registers to the exclusion of the other. Most of the mainstream of analytic philosophy has been obsessed with the Cartesian problematic. Moore, Russell, Broad, Ayer, Price, and Chisholm are among the most distinguished members of this Cartesian branch of the analytic tradition. But there is also a branch of the analytic tradition that, though it has fewer members, is equally as distinguished, and occupies itself almost entirely with the Kantian problematic. It includes among its members C.I. Lewis, Wilfrid Sellars, and Peter Strawson. If one then examines how a great deal of philosophical vocabulary (‘experience’, ‘epistemic’, ‘object’, ‘the given’, etc.) is employed, one notices that it acquires a very different inflection in the writings of one of these sets of authors than it acquires in the writings of the other. A preoccupation with a Kantian problematic characterizes the work of a philosopher such as Heidegger no less than it does that of a philosopher such as

Two Varieties of Skepticism

37

Sellars. Thus, armed with the distinction between these two varieties of skepticism, one can also notice points of affinity and divergence across the alleged divide between analytic and Continental philosophy. This can provide philosophically far less superficial categories for sorting philosophers into groups than those we usually rely upon. (One can learn to see past the common philosophical reference points and superficial similarities in philosophical tradition and style that unite thinkers such as Heidegger and Sartre, on the one hand, and Wilfrid Sellars and H.H. Price, on the other, and begin to discern certain fundamental divergences, thus enabling one to notice fundamental philosophical affinities between thinkers from different traditions, allowing one to sort together Heidegger and Sellars, on the one hand, and Sartre and Price, on the other.) Not only individual bits of vocabulary but whole phrases, clauses, sentences and questions acquire a different philosophical valence depending upon whether they occur in the context of the investigation of a Cartesian or a Kantian problematic. Thus, for example, the question “Can our cognitive powers reach all the way to the objects themselves?” can express a Cartesian anxiety about the existence of a gap between our representations of outer objects and those outer objects themselves (the anxiety here is that our cognitive powers always operate at an awkward remove from the objects they represent) or it can express a Kantian anxiety about our capacity to direct our thought at objects (the anxiety here is that our so-called cognitive powers are unable to furnish us with anything which even amounts to a “representation”.) Since the same form of words can, on a particular occasion, express either a Cartesian or a Kantian anxiety, it becomes possible, as we shall soon see, for two philosophers to take themselves to be in agreement with one another when they are not, or to take themselves to be in disagreement with one another when they are not. It thus becomes important, when seeking to identify the sort of philosophical problematic in which a philosopher is entangled, to look beyond the most superficial features of a his work, such as the forms of words that he is drawn to employ when framing his problem. One must look instead to the character of the features that characterize his problematic. One philosopher may adopt another philosopher’s mode of speaking wholesale and yet miss his problematic entirely.33 33 Conversely, one may resolutely avoid a particular philosopher’s ways of speaking as a strategy for avoiding his problems and yet end up mired in precisely the

38

James Conant

Even if one confines one’s survey of varieties of skepticism solely to noting the presence or absence of only the first of the nine generic features mentioned above (i. e., to noting the character of the examples that figure in each of these two varieties of skepticism), that alone ought to suffice to make one begin to wonder whether Price, Ayer and Austin might not, for the most part, be participating in a tradition of thought about skepticism very different from the one that runs through the writings of Heidegger and Lewis and Sellars. (This is, of course, not to deny that Lewis and Sellars ultimately wish to address some of the concerns of a Price or an Ayer, anymore than it is to deny that Kant ultimately wishes to address Descartes.) A pair of passages from Price and Lewis respectively will serve briefly to illustrate the point. In these passages, Price and Lewis are each concerned to isolate a certain stratum or strand of our experience – one which they each call the “given” (thus illustrating that the term “given” functions very differently in the context of a Cartesian and a Kantian investigation). First, here is Price’s tomato: When I see a tomato there is much that I can doubt. I can doubt whether it is a tomato that I am seeing, and not a cleverly painted piece of wax. I can doubt whether there is any material thing there at all. Perhaps what I took for a tomato was really a reflection; perhaps I am even the victim of some hallucination. One thing however I cannot doubt: that there exists a red patch of a round and somewhat bulgy shape, standing out from a background of other colour-patches, and having a certain visual depth, and that this whole field of colour is directly present to my consciousness. What the red patch is, whether a substance, or as a state of a substance, or an event, whether it is physical or psychical or neither, are questions that we may doubt about. But that something is red and round then and there I cannot doubt. Whether the something persists even for a moment before and after it is present to my consciousness, whether other minds can be conscious of it as well as I, may be doubted. But that it now exists, and that I am conscious of it – by me at least who am conscious of it this cannot possibly be doubted. And when I say that it is ‘directly’ present to my consciousness, I mean that my consciousness of it is not reached by inference, nor by any other intellectual process … nor by any passage from sign to significate…. This peculiar and ultimate manner of being present to con-

philosophical problematic one sought to avoid. This is part of the reason why Richard Rorty’s preferred strategies for dissolving philosophical problems tend to be so ineffectual.

Two Varieties of Skepticism

39

sciousness is called being given, and that which is thus present is called a datum. 34

Now here is Lewis’s fountain-pen: At the moment, I have a fountain pen in my hand. When I so describe this item of my present experience, I make use of terms whose meaning I have learned. Correlatively I abstract this item from the total field of my present consciousness and relate it to what is not just now present in ways which I have learned and which reflect modes of action I have acquired. It might happen that I remember my first experience of such a thing. If so, I should find that this sort of presentation did not then mean “fountain pen” to me. I bring to the present moment something which I did not then bring; a relation of this to other actual and possible experiences, and a classification of what is here presented with things which I did not then include in the same group. This present classification depends on that learned relation of this experience to other possible experience and to my action, which the shape, size, etc. of this object was not then a sign of. A savage in New Guinea lacking certain interests and habits of action which are mine, would not so classify it…. In whatever terms I describe this item of my experience, I shall not convey it merely as given, but shall supplement this by a meaning which has to do with relations, and particularly with relation to other experiences which I regard as possible but which are not just now actual…. The infant may see it much as I do, but still it will mean him none of these things I have described it as being, but merely “plaything” or “smooth biteable”. But for any mind whatever, it will be more than what is merely given if it be noted at all.35

One of the things one can say about Price’s tomato can, with equal justice, be said of both Price’s tomato and Lewis’s fountain-pen: Price’s expression, “When I see a tomato” has the force of “whenever I see any object”. But another of the things one can say of Price’s tomato does not hold of Lewis’s fountain-pen: the entire existence of the world is squeezed into that tomato. That is to say, Price’s tomato and Lewis’s fountain-pen each aspire to be in some way representative of the object of knowledge as such. Each of these examples aspires to a sort of generality, but not the same sort. Price’s tomato seeks to represent the object of knowledge as such in virtue of its alleged optimality – in virtue of the supposition that if knowledge fails here it will fail everywhere. Price’s tomato is selected as a best case of knowledge and the point of rolling it on to the scene is to see what in one’s experience of it is susceptible to doubt, what 34 Price (1932), 3. 35 Lewis (1929), 49 f.

40

James Conant

isn’t, and what the consequences of such doubt are. Its capacity for representativeness therefore turns on its being a fully determinate case of knowledge. Its representativeness is a function of the way it allows us to infer something about knowledge in general from a consideration of this particular case of knowledge. (Many of the deeper investigations into the Cartesian paradox therefore probe whether there is a tension in the Cartesian skeptic’s desire to furnish an example of knowledge that is supposed to be at one and the same time fully representative of knowledge as such and yet a fully determinate instance of a knowledge-claim.) Lewis’s fountain-pen seeks to represent the object of knowledge as such in virtue of its exhibiting general characteristics which any instance of empirical knowledge must possess – in virtue of the supposition that all knowledge possesses a common structure and therefore this instance of knowledge must possess that structure. Lewis’s fountain-pen is selected not as a best case of knowledge, but as a coherent candidate for knowledge. It is offered not as an exemplary instance, but merely as a run-of-the-mill specimen of the sort of thing that can figure in a claim to knowledge. What is at issue for Lewis is not the unimpeachable certainty, but rather the bare intelligibility of that with which his experience presents him. His example’s capacity for representativeness therefore turns on its being susceptible not to a certain sort of doubt, but to a certain sort of abstraction – an abstraction to which any determinate case of knowledge can be subjected. Its representativeness is a function of the way conclusions about this particular instance of knowledge flow from a consideration of what is constitutive of knowledge in general. (Many of the deeper investigations into the Kantian paradox therefore probe whether the conditions – e. g., the sensible and intellectual conditions – that Kantian skeptics allege to be constitutive of knowledge as such, and yet which as conceived by the skeptic appear not to be co-satisfiable, are properly conceived by the skeptic.) The existence of the world is not at stake in Lewis’s example of an object, though the possibility of its schematizability – its availability to our thought – is. The point of the example is not to inquire into whether we can know that there is a fountain-pen here, but how it is that we are able to apprehend this particular colligation of sense qualities as a fountain-pen. The point of Lewis’s fountain-pen is to present something (and, for this purpose, he could have chosen a much less generic object) knowledge of which involves both a contribution of sensibility and a contribution of understanding. The point of Lewis’s exploration of his example is not to see what will and what will not withstand doubt, but to isolate the respective

Two Varieties of Skepticism

41

contributions of sensibility and understanding in our apprehension of the object. Now I do not mean here to claim that Lewis’s sort of investigation does not figure in the work of authors oblivious to the possibility of Kantian skepticism. (Descartes’s exploration of his example of the piece of wax figures in The Second Meditation precisely as part of an exercise of isolating the respective contributions of the senses and the intellect.) I mean only to claim that if such an example, treated in such a way, is pressed to the point at which it begins to issue in a skeptical problematic, then the ensuing skeptical problematic will exhibit Kantian features. I have allowed myself, just now, to speak of “Lewis’s sort of example”. But this way of speaking is potentially misleading. What is at issue here is not (what philosophers usually mean when they talk about) “a kind of object”. (Lewis could have chosen a tomato. But he could also have chosen a goldfinch.) Instead of speaking of “Lewis’s sort of example”, I would have done better to speak of Lewis’s manner of taking up an example. Cavell, after introducing the terminology of “generic object” and “specific object” in The Claim of Reason, goes on to caution that these terms should not be taken to mean “that there are two kinds of objects in the world”, rather the point of such ways of speaking is to characterize “the spirit in which an object is under discussion, the kind of problem that has arisen about it, the problem in which it presents itself as a focus of investigation”.36 So, also, when I say that the object at the focus of the Cartesian problematic, call it the Cartesian object, is different from the object at the focus of the Kantian problematic, call it the Kantian object, this is not because there are two different kinds of objects in the world; rather the point is to characterize two different ways of taking up a possible object of experience and investigating the character of the relation that obtains between it and a cognizing subject. The Kantian object is not to be confused with the Cartesian object.37 This 36 Cavell (1979), 55. 37 Contrary to what Cavell suggests in various passing remarks, therefore, the Kantian notion of an “object in general” is radically different from Cavell’s notion of a generic object. The latter is Cavellian shorthand for the sort of object that figures in a Cartesian skeptical recital; the former is Kantian shorthand for what is left of our concept of an object once it has been subjected to a severe degree of abstraction. The Cavellian locution stands for the notion of a certain sort of actual concrete particular that is only very indeterminately characterized – yet, this indeterminacy notwithstanding, it is still meant to be something that can be exhibited in experience. Whereas the Kantian locution stands for the no-

42

James Conant

does not mean that the Cartesian object is an object which, due to its unusual properties, does not fall under the genus of Kantian object, but rather that the sort of concept of an object that is at issue in each of these skeptical problematics – and the sort of generality that attaches to each of these concepts – is different. What the example of the pair passages from Price and Lewis respectively are intended to illustrate is the way in which a whole raft of philosophical terminology – of which the term ‘object’ is meant to serve merely as one representative instance – acquires a new significance when the context of philosophical investigation shifts from the context of a Cartesian to that of a Kantian problematic.

IX. The Kantian Way with Skepticism We now come to the third form of philosophical insight that a perspicuous overview of various kinds of skepticism affords. It allows us to see more clearly some of the ways in which philosophers have had something original and penetrating to contribute to an understanding of the sources and nature of skepticism. One way the work of a philosopher can stand out in this regard is in its appreciation of the symmetries and asymmetries that obtain across the variants of, say, Cartesian skepticism. Another way the work of a philosopher can stand out is in its exploration of the connection between Cartesian and Kantian skepticism. There are two classic figures in the history of philosophy whose writings thus stand out – figures who have been concerned to explore the nature and structure of the Cartesian and Kantian problematics alike and, above all, to explore the relation between them. They are Kant and Wittgenstein. One way of plunging oneself into the heart of their respective philosophies – and of measuring the singular achievement of their respective contributions to the history of thought about skepticism – is tion of a bare abstract concept of a particular – a concept every actual concrete individual equally instantiates and none properly exemplifies. To mistake this Kantian concept of an object for the name of a sort of item that some sort of being could encounter in experience is a mistake Kant himself warns against. The concept that one arrives at through this severe application of the Kantian procedure of abstraction is, as Kant puts it, an “entirely indeterminate concept”; and he warns that fatal confusion results when one “is misled into treating this entirely indeterminate concept … as if it were a determinate concept of an entity that allows of being known in a certain manner” (Kant (1998), B307).

Two Varieties of Skepticism

43

to focus one’s attention on their respective inquiries into the relation that obtains between these two varieties of skepticism. This last observation touches directly on a further respect in which my employment of the labels ‘Cartesian’ and ‘Kantian’ might seem tendentious. Was not Kant as interested in (what I am here calling) “Cartesian skepticism” as anyone? My invocation of Kant, in calling the second of these kinds of skepticism “Kantian”, is not meant to suggest that Kant does not know about or does not care to address Cartesian skepticism. Before turning to the task of clearing up this misunderstanding, it will help to draw a distinction between weaker and stronger senses in which a philosopher might be termed “Kantian”. To be a Kantian in the weak sense – what I will henceforth call mild Kantianism – requires only that one share with Kant a philosophical problem of a certain form; to be a Kantian in a yet stronger sense – which I will henceforth call medium-strength Kantianism – requires that one also share with Kant a conception of the form that the solution to such a problem must assume and what the philosophical implications of such a solution are. I take what I have just said to come to the same thing as this: To be a mild Kantian (the only sense of “Kantian” that has thus far been at issue in this paper) is to be someone who finds himself embroiled in some phase of (what I have been calling) the Kantian problematic; to be a medium-strength Kantian involves not merely finding oneself in the predicament of being thus embroiled, but also wishing to implement a very particular philosophical strategy for liberating oneself from such a predicament – namely, the one that Kant himself sought to implement.38 I remarked earlier (in my elucidation of the eighth Kantian feature) that the Kantian problematic tends to be most lucidly adumbrated in the writings of kinds of Kantian who aim to show that where we seem to be confronted with a Kantian gap, we are confronted with only the illusion of a gap. The strategy of such Kantians is to find a way to make the seeming gap disappear. Part of the motivation for such Kantians in making their gap disappear is that, in the process, they hope to show that the corresponding Cartesian gap can also be made to vanish. I am taking it that Kant himself was such a Kantian. That is to say, where I have been 38 The proper account of what it is to be a medium-strength Kantian therefore ultimately rests on a detailed reading of the most difficult portions of The Critique of Pure Reason. I will make no attempt to defend such a reading here; but I will, in various ways, presuppose such a reading.

44

James Conant

distinguishing between two problematics, Kant (as I read him), and certain Kantians, will insist that there is only one extended philosophical problematic. I earlier canvassed a reason why someone might not want to call (what I have been calling) “Kantian skepticism” a variety of skepticism at all. We now see a reason why some philosophers will consider what is at issue here a philosophical problematic that is properly subsumed under the rubric of “skepticism”: namely, because they think that what is at issue in the Kantian skeptical problematic is a philosophically more perspicuous and resolute understanding of single philosophical anxiety that, as it comes to understand itself better, comes to assume an increasingly radical form – a philosophical anxiety that (both historically and systematically) comes into view (and achieves a comparatively less clear and distinct form of expression) in the initial guise of Cartesian skepticism. A variant of what I have been calling “Cartesian skepticism” comes in for special treatment in the Critique of Pure Reason under the rubric of idealism. So the first thing to notice here is that when Kant is restricting his attention to (what I have been calling) Cartesian skepticism he does not employ the term ‘skepticism’. Idealism, for Kant, centers around a doubt about the existence of mind-independent objects or as Kant sometimes calls them “material objects”. (Thus Kant, employing a label that has not found favor with posterity, also refers to this philosophical position as material idealism.) Kant writes: “Idealism … is the theory which declares the existence of objects in space outside us to be either merely doubtful and indemonstrable, or else false and impossible”.39 Skeptical doubt here can be directed at the demonstrability of the existence of such objects (and can thus issue in agnosticism about their existence); or it can be directed at whether we can so much as make sense of the idea of such objects (and can thus issue into the material-object counterpart of atheism). Thus Kant distinguishes two species of idealism (he calls them “problematic idealism” and “dogmatic idealism”. (He attributes the former to Descartes and the latter to Berkeley.) The former sort of idealist doubts that we are able to attain knowledge of mind-independent objects. The latter sort of idealist doubts that there are any such objects. The version of idealism that Kant attributes to Descartes, and with which Kant is focally concerned in “The Refutation of Idealism”, turns crucially, Kant thinks, on a certain way of drawing the distinction be39 Kant (1998), B274.

Two Varieties of Skepticism

45

tween inner and outer sense. Descartes does not doubt that we have inner sense – that is that we have experiences of objects of some sort. There is a kind of experience that is immune to doubt. The objects regarding which we have such an indubitable variety of experience are all modes of mental substance, res cogitans. Whereas outer objects – modes of corporeal substance, res extensa – figure in Descartes’s philosophy as radically dissimilar to inner objects in that their existence is inherently open to skeptical doubt and consequently in need of philosophical demonstration. The inherent vulnerability of Cartesian outer experiences to doubt, according to Kant, is to be traced to the manner in which experiences of outer objects are conceived by Descartes. Outer objects are conceived by Descartes in such a way that they themselves never fall within the immediate purview of experience. They are never experienced immediately, but only indirectly. The only objects of immediate experience are inner experiences. If one wishes to speak of “outer experiences” then, on a Cartesian conception, the only thing that one can mean by such a phrase is a species of inner experience – something that happens in the mind – where the mental happening in question is one that purports to be about what is outside the mind. So-called “outer experiences” are happenings within the mind which involve representations of what is happening outside the mind. A so-called “outer experience” is thus conceived by Descartes to be an inner experience purporting to be both of something outer and caused by the object it is of. Thus, on Kant’s reading of him, Descartes takes the possibility of experience of how things seem outside of us – the possibility of an experience’s being thus “about” or “of” something outer – to be a comparatively unproblematic phenomenon. The skeptical problem for Descartes – what Kant calls the problem of idealism – only begins when we turn to the question of how we can reason reliably from how things seem to us to how things actually are outside of us. Thus Kant stresses that the apparent coherence of the idealist’s position rests on the presupposition of the apparent coherence of an inference from how things are within the mind to how things are without the mind: “Idealism assume[s] that the only immediate experience is inner experience, and that from it we can only infer outer things – and this, moreover, only in an untrustworthy manner, as in all cases where we are inferring from given effects to determinate causes.”40 Kant characterizes his argument-strategy in “The Refutation of Idealism” as one in which “the 40 Ibid. B276.

46

James Conant

game played by the idealist is turned against itself”.41 Kant aims to show that without what the idealist wants to infer to there would be nothing to infer from. To turn idealism’s own game against itself is, for Kant, to show “that we have experience, and not merely imagination of outer things”; and Kant thinks that the way to show this is to show that “even our inner experience, which for Descartes is indubitable, is possible only on the assumption of outer experience”.42 What Descartes takes to be a separable and self-contained realm of experience – the sphere of inner life (the domain of res cogitans) – Kant wants to show to be not in the least self-standing in the way that Descartes imagines it to be. Kant wants to show that the capacity for inner sense is only possible for a being which also already has the capacity to enjoy immediate experiences of outer objects, and, moreover, that these two capacities are not separable in the way Descartes imagines inner and outer sense to be. Rather they are equally constitutive aspects of the unitary consciousness of any being capable of experience.43 Kant’s argument in “The Refutation of Idealism” itself presupposes the results of the preceding pages of the Transcendental Analytic; and the aim of the Transcendental Analytic, as a whole, is to try to get the Kantian paradox into view and to show how to make it go away. That is to say, Kant’s treatment of the Cartesian problematic presupposes his treatment of (what I have been calling) the “Kantian problematic”. The question that the Transcendental Analytic seeks to address is: what sort of unity must characterize a play of representations for them so much as seem to be of objects. The decisive step in the Transcendental Deduction lies in showing that a play of representations, for it to have objective purport, must be in accord with the unity that the categories prescribe. In showing that anything worthy of the title of experience must accord with these conditions of unity, Kant takes himself to be addressing someone whom he calls “the skeptic”. When, in such contexts, 41 Ibid. 42 Ibid. B275. 43 The argument that inner and outer sense mutually depend upon one another comes in two steps. The argument that all objects – and hence also the objects of outer sense – depend upon the form of inner sense is presented in the Transcendental Aesthetic. This shows that outer experience is only possible on the assumption of inner experience. The Refutation of Idealism shows that inner experience is only possible on the assumption of outer experience. These two arguments jointly show that neither of these two sorts of experience is possible except as part of a larger package which also includes the other.

Two Varieties of Skepticism

47

we find Kant using the word ‘skepticism’ what is at issue is a (Kantian) boggle, not a (Cartesian) doubt. Here is an example of such a context: The concept of cause, for instance, which expresses the necessity of an event under a presupposed condition, would be false if it rested only on an arbitrary subjective necessity, implanted in us, of connecting certain empirical representations according to the rule of causal relation. I would not then be able to say that the effect is connected with the cause in the object, that is to say, necessarily, but only that I am so constituted that I cannot think this representation otherwise than as thus connected. This is exactly what the skeptic most desires. For if this be the situation, all our insight, resting on the supposed objective validity of our judgements, is nothing but sheer illusion.44

What is at issue in this passage is not a Cartesian worry (that we might be subject to illusions of knowledge), but rather that we might be subject to an even more radical form of illusion: an illusion of objective purport. In order to understand better who this skeptic is and how he comes to be called a “skeptic”, it helps further to distinguish a particular sub-species of the Kantian problematic – which one might call the Humean problematic with regard to causation. One then needs to see how a certain sort of radicalization of the Humean problematic, from the category of cause to the other categories, issues directly into the Kantian problematic. A proper discussion of this topic would require considerably more space and take us well beyond what is required for the purposes of the schematic distinction here at issue between varieties of Kantianism of various strengths. It will help now to discriminate more finely the possible strengths of Kantianism and to distinguish between medium-strength and (what I shall call) maximum-strength Kantianism. I earlier defined the former grade of Kantianism by saying that it requires that one share with Kant a conception of the form that the solution to the Kantian problematic must assume and what the philosophical implications of such a solution are. To be a maximum-strength Kantian requires not only that one share Kant’s philosophical aim and strategy for liberating oneself from the Kantian skeptical predicament, but also to have mastered the requisite nuances of such a strategy to be able to implement it successfully. Since the application of the term ‘Kantian” in this strongest possible sense thus depends upon the successful execution of a philosophical endeavor, it can remain a philosophically open question whether it is so 44 Kant (1998), B168.

48

James Conant

much as possible to be a maximum-strength Kantian. If one takes it that it is possible (that is, if one is a medium-strength Kantian), but if one’s philosophical endeavor fails to live up to its own aspirations (thus placing in jeopardy the possibility of being a maximum-strength Kantian), then one is threatened with a skeptical outcome of a different order from that which seems to threaten the unsuccessful execution of a Cartesian inquiry. Historically, the possibility of such an outcome first comes fully into view with the possibility of a Kantian response to Kant’s own radicalization of the Humean problematic. This opens up the possibility of an investigation climaxing in a form of skeptical paradox of a sort apparently quite distinct from that of the Cartesian paradox. In order to gain a firmer sense of how a distinctively Kantian form of skeptical paradox can come to seem a philosophically urgent matter, it is worth returning for a moment to the case of C.I. Lewis.45 Building on his example of the fountain pen, consider how Lewis goes on to argue for his thesis that we must acknowledge a given element in perceptual experience. Lewis’s opening moves are unimpeachably Kantian. Lewis starts from the thought that conceptual activity is a form of activity, and as such freely undertaken. Here Lewis is following Kant in taking it that all cognition must involve an element of spontaneity – that element in knowledge which we, as knowers, must freely bring to the world in order for knowledge of it to be a possible achievement for us. But, according to Lewis, this requires that we also recognize a given element in experience – something that operates as a constraint from outside on the freedom of conceptual activity – the element in knowledge which we, as knowers, must unresistingly take in from the world in order for our “knowledge” of “the world” to be something more than the contemplation 45 This case is discussed by McDowell in his paper “Putnam on Natural Realism”. (McDowell’s paper was written and first read by me in or around 1998. It has been forthcoming ever since in the Library of Living Philosophers volume on The Philosophy of Hilary Putnam – a volume which apparently is now scheduled to be published in 2012.) The ensuing discussion of Lewis simply summarizes in McDowell’s own language McDowell’s treatment of Lewis. It seeks to do so in a manner which brings out as vividly as possible how the climax of Lewis’s own Kantian skeptical problematic is of such a sort as to issue in a form of paradox quite different from the Cartesian one. That McDowell’s own way of summarizing the upshot of Lewis’s Kantianism should be able to bring this point out so vividly itself illustrates how very clear McDowell himself is regarding the differences between Cartesian and Kantian skepticism which form the main topic of this paper.

Two Varieties of Skepticism

49

of our own reflection, for it to be of something outside us. Here Lewis is following Kant in taking it that all cognition must therefore equally involve an element of passivity. 46 Without some contribution of spontaneity, the deliverances of our senses would be mere impacts – as Kant says: blind – less even than a mere dream. But, on its own, the mere activity of thinking is empty – as Kant says: ohne Sinn oder Bedeutung – as long as its play remains unconstrained from without. What Lewis calls “the given” is that which is thus delivered up to the mind, furnishing it with content upon which to operate, thus rescuing its operations from emptiness and arbitrariness. Lewis takes us to have no choice but to acknowledge such a given element in experience, if we wish to frame a conception of what it is to have an empirical world-view according to which (that which we wish to be able to look upon as) our “world-view” amounts to something more than wholly unconstrained fabrication. Indeed, without the constraint of the given, what we are left with is less even than a mere fabrication. For there can only be fabrication where there can be something more; and without any constraint on the operations of our spontaneity, the distinction between mere fabrication and something better lapses. There is no longer any basis for distinguishing between those of our constructions that afford a glimpse of how things are and those that are mere caprice. Without the possibility of drawing a distinction between mere caprice and genuinely world-guided belief, there is no longer anything in our so-called “world-view” that is recognizable as world-directed thought. Lewis, however, construes the second condition on knowledge in a very particular way and takes himself, in thus construing it, to be following Kant (thereby participating in a long-tradition of “followers” of Kant): he assumes that the freedom of conceptual activity must be seen as constrained from outside its own sphere, if we do not wish to lose our grip on thought’s possession of content. Hence Lewis declares not only something Kant would be happy to declare: “If there be no datum given to the mind then knowledge must be contentless and arbitrary; there would be nothing which it must be true to.”47 But he also 46 It is constitutive of any robust (i. e., non-mild) form of Kantianism that it postulates both of these elements and seeks to illuminate and render intelligible the possibility of their productive cooperation with one another. The difference between medium-strength and maximum-strength Kantianisms tends, above all, to turn on the relative degree of failure or success in rendering these apparently independent conditions on knowledge productively co-satisfiable. 47 Lewis (1929), 38 f.

50

James Conant

goes on to declare: “The pure concept and the content of the given are mutually independent; neither limits the other.”48 Lewis insists that the given, qua merely given, must remain utterly uncorrupted by the concepts which we bring to it, on pain of our falling into a vicious form of idealism, in which we are no longer able to see our conceptual activity as constrained by anything from outside its own sphere. Lewis, however, also needs find a way to allow that the data given to the mind be more than bare and brute. They must be able to “guide” belief. He thus needs also to be able to conceive the postulated given element in experience as such that it be able to direct and underwrite our conceptual activity. So he here finds himself in a bind: the given must be able to “guide” experience without in any way “limiting” it. The given must constrain conceptual activity, while exercising its constraining influence from wholly outside the sphere of conceptual activity. The bind that Lewis finds himself in is that these two requirements on the given are not obviously co-satisfiable. He must locate the given as far enough “outside” the sphere of the conceptual to allow it to serve its function as “external constraint” (keeping our cognitive wheels from spinning in a frictionless vacuum), while placing it close enough “inside” that sphere to permit it to serve its “guiding” function (warranting the application of some of our concepts, while refusing that of others). Lewis is thus faced with the following challenge: How is the given able to offer guidance to our concepts while remaining itself always conceptually unarticulated in nature? Lewis heroically strives to tell a story which allows him to walk this tightrope – a story according to which the given can serve its warranting function in spite of itself being that which by its very nature cannot be captured by concepts. It is to this end that he attempts to draw the distinction between being an instance of a quale and being an instance of a concept. He wants to be entitled to see the former sorts of instances as possessing many of the features that instances of the latter sort possess without themselves being instances of the latter sort. He says (in the fountain-pen passage quoted above) of that which is given in the diverse possible experiences of the fountain pen (i. e., those of the infant, the professor, and the savage) that, qua given, what is “presented” is “a colligation of sense-qualities”.49 He wants to be able to say that the same (in some non-conceptual sense of “same”) colligation of sense-qualities is presented in each such 48 Ibid. 37. 49 Ibid. 49.

Two Varieties of Skepticism

51

case; and that it is this that in each guides the use of concepts. This requires that he be able to make sense of the idea that one can recognize (in some sense of “recognize”) such colligations as the same without the mediation of concepts. In order to cash out this idea of a non-conceptual mode of recognition, he needs to be able to exploit the implication of universality in the term “qualities” while conceiving the sort of universality here at issue as utterly distinct from the universality of concepts (on pain of having the supposedly “external” constraint collapse into the sphere of conceptual activity). But Lewis is unable to pull off the trick, he keeps falling off the tightrope, with the consequence that the argument of Mind and the World Order, utterly contrary to Lewis’s own intention, threatens to collapse into a distinctively Kantian variety of skeptical paradox – one which leaves us unable to make sense of ourselves as so much as in possession of a “world-view”. Lewis is the sort of a medium-strength Kantian who has helped to convince many that maximum-strength Kantianism is unattainable and thus that Kantianism represents nothing more than a philosophical dead-end. It is crucial to a proper understanding of Kant’s own way of avoiding the outcome threatening Lewis’s inquiry (and thus to an understanding of “Kantianism” in the maximum-strength sense of the term) to see that it is absolutely central to the aim of the Transcendental Analytic not only to show (as Lewis himself seeks to do) that there are these two dimensions of knowledge – sensibility and understanding – – each of which is a necessary and neither of which is a sufficient condition of knowledge, but also to show precisely what Lewis imagines he must avoid (on pain of sacrificing the external constraint on knowledge that saves it from being contentless and arbitrary): namely, that these twin aspects of our cognitive constitution – our capacity to receive a sensory impression that things are thus and so and our capacity to direct our thought at an object (real or imaginary) are in no way independently satisfiable conditions on the possibility of knowledge.50 On my understanding of what maximum-strength Kantianism entails, what is original in Kant’s understanding of the relation between the two varieties of skepticism distinguished above is that he thinks that it is only once you see the way in which the intelligibility of each of these conditions presupposes the other – that is, that it is only once you see (1) what is transcendentally required in order to exercise the capacity to enjoy sense 50 For more on this point, see Thomas Land’s contribution in the companion volume to this volume, Rethinking Epistemology vol. 1 (Land (2012)).

52

James Conant

impressions that purport to be of something and (2) what is transcendentally required in order to exercise the capacity to entertain thoughts that purport to be about something, and (3) why you can only have one of these capacities if you have both – that you are able to equip yourself with the resources to dispose of the Cartesian skeptic. If the Cartesian skeptic grants, as he must, that he has the resources to enjoy impressions and entertain thoughts – resources required in order so much as to frame his doubt – then, having thought your way through the dialectic of the Kantian problematic, you will be in a position to show him that the gap he seeks to bridge is only the illusion of a gap. Thus, for Kant, the way to handle Cartesian skepticism is not to answer it head-on, but to think through the implications of an even more radical skeptical problematic. In coming to see what is incoherent in Kantian skepticism, we are to come to see what is incoherent in Cartesian skepticism as well. I will call this two-phased approach to the twin problematics of Cartesian and Kantian skepticism the Kantian way with skepticism. Most readers of Kant utterly misunderstand The Critique of Pure Reason by taking its response to Cartesian skepticism to begin much earlier in the book than it does. And such a misreading of the book is inevitable if one fails to see that the Cartesian problematic represents for Kant only a special case of a more general problematic. Kant’s generic term for this more general problematic – of which Cartesian skepticism is eventually to be recognized as only a limiting case – is “skepticism”. This furnishes yet another reason why, even if you yourself wish to reserve the term ‘skepticism’ for a problematic exhibiting characteristically Cartesian features, it is important to be able to realize that in speaking of “skepticism” a philosopher may have (what I have been calling) “Kantian skepticism” in mind: namely, because, unless you allow for this possibility, it will become impossible to see what Kant, and certain Kantians, are up to in their treatments of (what they call) “skepticism”. If one groups together the writings of philosophers who practice something akin to the Kantian way with skepticism, it induces an interesting canon—one which includes, among others, writings as diverse as T.H. Green’s critiques of Locke and Hume, Frege’s critique of psychologism, early Wittgenstein’s attempt in the Tractatus to show that realism and idealism strictly thought through can be seen to collapse into one another, and more recently John McDowell’s criticism of Davidson in Mind and World. The positive touchstone of this way with skepticism is a radical following through of the implicit assumptions of a skeptical position up to the point at which the position founders in incoherence.

Two Varieties of Skepticism

53

The Kantian way with skepticism involves the ascent up a dialectical ladder that one eventually recognizes as one that is to be thrown away.

X. A Case of Apparent Agreement: Putnam and McDowell We now come to a fourth form of philosophical insight that a perspicuous overview of the various kinds of philosophical response to skepticism can afford. A number of the finest contemporary philosophers writing about skepticism today misunderstand one another’s writings by trying to shoehorn a skeptical paradox bearing Kantian features into a problematic of Cartesian form and vice versa. As a first example of such a misunderstanding, let us consider a recent misencounter between Hilary Putnam and John McDowell.51 In his 1994 Dewey Lectures 52, Hilary Putnam attempts to identify a widespread assumption that he claims runs throughout early modern philosophy – he dubs the assumption in question “the interface conception” – and he argues that, three centuries later, it continues to be responsible for many of the difficulties that plague contemporary philosophy. Putnam characterizes the consequences of the continuing hold of this assumption on the philosophical imagination of our time as nothing short of a “disaster”. His summary statement of how contemporary philosophy managed to place itself in its present disastrous position runs as follows: [T]he key assumption responsible for the disaster is the idea that there has to be an interface between our cognitive powers and the external world – or, to put the same point differently, the idea that our cognitive powers cannot reach all the way to the objects themselves.53 51 My discussion of this example is indebted throughout to John McDowell’s paper “Putnam on Natural Realism” mentioned above. This section of my paper contains nothing original that is not to be found in McDowell’s beyond the way in which it seeks to show how the disagreement between McDowell and Putnam at issue is usefully reframed in terms of the two varieties of skepticism distinguished above and therefore how the disagreement here is au fond one regarding the relative priority of these two varieties of skepticism. 52 Putnam’s Dewey Lectures were given at Columbia University in March of 1994 and first published in Vol. XCI, No. 9 of the Journal of Philosophy. They are reprinted in Putnam’s book The Threefold Cord (1999) and all references to them will be to this latter publication. 53 Putnam (1999), 10.

54

James Conant

Putnam here glosses the crucial assumption as follows: “the idea that our cognitive powers cannot reach all the way to the objects themselves”. If we could overcome the assumption expressed by these words, Putnam suggests, we would be in a position to embrace with a sound philosophical conscience what Putnam (following William James) calls “the natural realism of the common man”. This locution – “natural realism” – as Putnam deploys it, is not meant to be a label for an alternative philosophical position; rather it is meant to denote something both more familiar and more elusive: our own pre-philosophical understanding of the character of our cognitive relation to the world, prior to its corruption by certain forms of philosophizing that have now come to seem to be forms of post-scientific common sense. Thus Putnam is able to describe what he seeks to recommend in the Dewey Lectures as the cultivation of a kind of second naiveté about the objects of perception. But our philosophical consciences are troubled. Putnam knows this, and thus knows that, in issuing his call for a return to a lost state of epistemological innocence, he is bound to appear to many of his colleagues to be merely the most recent incarnation of the proverbial philosophical ostrich burying his head in the sands of our everyday ways of talking and thinking. What makes it inevitable that things will so appear to many of his colleagues, according to Putnam, is the interface conception: it is what makes it look as if the recommended species of naiveté cannot be anything other than mere naiveté. Putnam credits John McDowell, in his book Mind and World, with having identified (what Putnam calls in the passage above) the “key assumption”.54 But Putnam

54 The context of the previous quotation makes this evident: Let us now ask just why realism about “the external world” came to seem problematical. Early modern philosophers assumed that the immediate objects of perception were mental, and that mental objects were nonphysical…. What is more, even their materialist opponents often put forward accounts of perception that closely paralleled these “Cartesian” accounts… In his 1991 John Locke Lectures [published in Mind and World] John McDowell argues persuasively that this picture, whether in its classical version or in its modern materialist version, is disastrous for just about every part of metaphysics and epistemology. In McDowell’s view the key assumption responsible for the disaster is the idea that there has to be an interface between our cognitive powers and the external world – or, to put the same point differently, the idea that our cognitive powers cannot reach all the way to the objects themselves (Putnam (1999), 10).

Two Varieties of Skepticism

55

is here misreading McDowell;55 and in order to pinpoint how such a misreading of McDowell comes about, it helps to be able to see, as we already have, how a phrase such as “our cognitive powers cannot reach all the way to the objects themselves” can hover – and does hover on Putnam’s pages – between two different sorts of philosophical problematic: a Cartesian and a Kantian one. To see how Putnam is here misreading McDowell, we need to see how Putnam manages to change the significance of McDowell’s prose while often faithfully paraphrasing or quoting McDowell’s exact words. We need to see how Putnam’s employment of stretches of philosophical prose – prose which he lifts directly from the pages of McDowell’s book – does not leave their sense unchanged. By transplanting McDowell’s locutions into the context of Putnam’s own attack on the interface conception, Putnam confers a sense upon McDowell’s locutions different from that which they possessed in the context of McDowell’s own investigation. As the context in which the relevant stretches of McDowell’s prose figure shifts from the exploration of a Cartesian to that of a Kantian problematic, so does their sense – in ways parallel to the ways in which the central locutions common to both Price’s and Lewis’s passages above (“object” or “thing”, “consciousness”, “object”, “the given”, etc.) shift from the one passage to the next. A central aim of McDowell’s book is to make room for the following truism (one that philosophy can easily seem to place out of reach) which Wittgenstein expresses, in McDowell’s favorite quotation from Wittgenstein, as follows: “When we say, and mean, that such-andsuch is the case we – and our meaning – do not stop anywhere short of the fact.”56 This sounds very much like things Putnam himself wants to say, for example, in wishing to affirm that “our cognitive powers can reach all to the objects themselves”. In affirming this, Putnam represents himself, in the Dewey Lectures, as spelling out some of the basic ideas behind McDowell’s strategy for making room for truisms of the above Wittgensteinian sort. And, of course, the negation of the claim which figures in the last phrase in the above quotation from Put55 I do not mean to deny that McDowell is concerned to criticize the interface conception in some of his writings, most notably in his essay “Criteria, Defeasibility, and Knowledge”, but only that McDowell would not identify the target of that essay with “the key” confusion that he seeks to exorcize in Mind and World. 56 Wittgenstein (1973), section 95. The passage is quoted by McDowell (1994), 27.

56

James Conant

nam – “that our cognitive powers cannot reach all the way to the objects themselves” – can be taken as merely paraphrasing the same truism that figures in McDowell’s favorite quotation from Wittgenstein. It is clear that Putnam, in aligning himself with McDowell, wishes us to take it that way. But it is equally clear that he also wishes us to take the negation of the claim expressed by that phrase as the expression of the repudiation of the disastrous assumption forced on us by the interface conception. It is in wishing to be able to mean words such as these in both of these ways at once that Putnam gets entangled both in his understanding of the structure and aim of the argument of McDowell’s Mind and World and in his proposals for a way out of the problems that currently plague contemporary philosophy. One sign that some slippage has taken place is that the Wittgenstein passage – and McDowell’s book generally – is concerned with the possibility of meaning (with the possibility of our being able to mean something – rather than nothing – by our words), whereas much of the Dewey Lectures are focally concerned with the possibility of knowledge (with how it is that we can know things are as they seem). In his opening remarks, Putnam expresses the thought he wants to vindicate in his lectures as follows: “there is a way to do justice to our sense that knowledge claims are responsible to reality without recoiling into metaphysical fantasy”.57 But from McDowell’s point of view, the focus here on knowledge claims (as the paradigm for understanding the kind of responsibility to reality which philosophy brings into question) must count as at best misleading, and at worst misguided. For it invites a misidentification of the strand of philosophy that McDowell is focally concerned to treat: it invites the substitution of a Cartesian for a Kantian problematic. McDowell would, of course, not want to deny that we should view knowledge claims as responsible to reality. But he would want to insist that, for the purposes of his investigation, they constitute only a special case of a more general (Kantian) problematic concerning the possibility of the vulnerability of any claim to how things are. The sort of vulnerability to reality under investigation in his book is exhibited in false claims equally fully as in true ones. The kind of vulnerability to reality at issue in his investigation figures equally in our capacity for thinking (indeed, dreaming) as it does in our capacity for knowing. McDowell’s investigation does not single out those of our claims which are knowledgeable because it is concerned with what must be the case for any sort 57 Putnam (1999), 4.

Two Varieties of Skepticism

57

of human intellectual activity to be intelligible as one of claim-making at all. It is the possibility of making claims in general, not knowledge claims in particular, that comes to seem at risk when our entitlement to the truism in Wittgenstein’s remark seems to come under threat. As he proceeds in his lectures, as we shall see in a moment, Putnam himself describes the issue with which he is concerned in terms that make no particular reference to knowledge, for instance as the “how does language hook on to the world” issue. So it may seem an uncharitable quibble to raise worries about his opening formulation of the issue in terms of knowledge. But the narrowness of the epistemological manner in which he introduces his topic is symptomatic of a disabling unclarity that pervades his lectures – an unclarity regarding the character of the philosophical problematic which he seeks to address. As we saw in the previous section of this paper, the phrase “our cognitive powers cannot reach all the way to the objects themselves” can be inflected in either a Cartesian or a Kantian register. The phrase can be taken to denote the idea of either a Cartesian or a Kantian gap. Inflected in a Cartesian register, the meaning of the phrase can happily be glossed as Putnam glosses it: as “the idea that there has to be an interface between our cognitive powers and the external world”. But it is only in its Kantian inflection that the phrase tracks the problematic that stands at the center of McDowell’s investigation in Mind and World, and, thus inflected, it cannot simply be identified – as Putnam throughout wishes to identify it – with the Cartesian assumption that there has to be “an interface between our cognitive powers and the external world”. It is only through this conflation of the kinds of “responsibility to reality” at issue in Cartesian and Kantian skepticism respectively that Putnam is able to take the diagnosis that McDowell aims to advance in Mind and World to be coincident with the one he himself seeks to advance in the Dewey Lectures. To see this more clearly, consider the following sequence of passages from Putnam’s Dewey Lectures: Let us now ask just why realism about “the external world” came to seem problematical. Early modern philosophers assumed that the immediate objects of perception were mental, and that mental objects were nonphysical…. What is more, even their materialist opponents often put forward accounts of perception that closely paralleled these “Cartesian” accounts. Even in contemporary cognitive science, for example, it is the fashion to hypothesize the existence of “representations” in the cerebral computer. If one assumes that the mind is an organ, and one goes on to identify the

58

James Conant

mind with the brain, it will then become irresistible to (1) think of some of the “representations” as analogous to the classical theorist’s “impressions” … and (2) to think that those “representations” are linked to objects in the organism’s environment only causally, and not cognitively…. … I agree with James, as well as with McDowell, that the false belief that perception must be so analyzed is at the root of all the problems with the view of perception that, in one form or another, has dominated Western philosophy since the seventeenth century…. The tendency in the last thirty years to repress what continues to puzzle us in the philosophy of perception obstructs the possibility of progress with respect to the broader epistemological and metaphysical issues that do preoccupy us…. … How could the question “how does language hook on the world?” even appear to pose a difficulty, unless the retort: “How can there be a problem about talking about, say, houses and tress when we see them all the time?” had not already been rejected in advance as question-begging or “hopelessly naïve”. The “how does language hook on to the world” issue is, at bottom, a replay of the old “how does perception hook on to the world” issue. And is it any wonder if, after thirty years of virtually ignoring … the task of challenging the view of perception that has been received since the seventeenth century …, the very idea that thought and language do connect with reality has come to seem more and more problematical? Is it any wonder that one can’t see how thought and language hook on to the world if one never mentions perception? 58

In the above sequence of remarks, Putnam wishes to identify the difficulty posed in the question “how does language hook on the world?” with the difficulty made urgent by the interface conception. Elsewhere in his writings, Putnam is extremely sensitive in his treatment of philosophical problems that bear the earmarks of a Kantian problematic, and in showing how putative solutions to those problems, if strictly thought through, can be seen to collapse into variants of Kantian skepticism.59 But in the above sequence, Putnam insists on privileging a Cartesian problematic as holding the key to a diagnosis of philosophy’s most fundamental problems. I do not mean to suggest that Putnam is entirely clear or resolute in his Dewey Lectures in his desire to privilege a Cartesian problematic over a Kantian one. On the contrary: as the Dewey Lectures proceed, it is clear that Putnam also wishes to identify the difficulty posed in the question “how does language hook on the world?” with 58 Ibid. 9 – 13. 59 Putnam’s criticisms of attempts to naturalize meaning are an example of this. He sees such views as degenerating into a variant of Kantian skepticism. Putnam is here influenced by certain arguments of Frege’s. For a brief discussion of the strain of Putnam’s thought that I have in mind here, see my “Introduction” to Putnam (1994), xlii-xlvi.

Two Varieties of Skepticism

59

the difficulty (at the heart of much contemporary philosophy of mind and language) which McDowell seeks to address in Mind and World – one that, as we saw from our brief discussion of C.I. Lewis, does not fit into a Cartesian mould. It is precisely this irresoluteness on Putnam’s part that I am seeking to bring into focus: Putnam wants what is at issue throughout to be the same “how does language hook on the world?” issue. But what I said above about the phrase “our cognitive powers cannot reach all the way to the objects themselves” applies equally to the question “how does language hook on the world?”. In the pages of Putnam’s Dewey Lectures, these forms of words hover unstably between the expression of a Cartesian and a Kantian worry – and that instability can be resolved only at the cost of depriving Putnam’s preferred diagnosis of the source of contemporary philosophy’s ills of its intended generality. Putnam’s “key assumption” cannot unlock the problems that McDowell seeks to address. Large stretches of the Dewey Lectures make sense only if the worry expressed in questions such as “Can our cognitive powers reach all the way to the objects themselves?” and “How does language hook on the world?” is taken to be of a Cartesian variety. This is equally true of the entire sequence of passages quoted above; and if that sequence is taken as offering – as it clearly intends to – a diagnosis of most of what ails contemporary philosophy, and if the question at issue is taken to express both the worry made urgent by the interface conception and yet somehow at the same time the one that animates the sorts of philosophical worry McDowell seeks to address in Mind and World, then Putnam may not claim McDowell as an ally. For what Putnam then says in that sequence of passages must be fundamentally at odds with the diagnosis McDowell seeks to put forward in Mind and World. What McDowell will take to be unsatisfactory in Putnam’s diagnosis can be put as follows: Putnam, in effect, suggests that the various forms of Kantian skepticism that have come to seem so urgent in recent philosophy can be exorcised simply through the treatment of Cartesian skepticism. McDowell would want to press the following question: Is Putnam right in claiming that the “how does language hook on to the world?” issue is, at bottom, simply a replay of the old “how does perception hook on to the world?” issue? The correct answer to this question is: it depends on which “how does language hook on to the world” issue is at issue. (As I have indicated above, there is a Cartesian and a Kantian version of the issue.) McDowell would contend that Putnam’s claim is

60

James Conant

not right about the version of the problem that is, for the most part, the one that figures at the center of concern in the writings of, for example, Lewis or Sellars or Kripkenstein or Crispin Wright. The Kantian difficulty (concerning the very idea of thought’s or language’s capacity to be vulnerable to reality) with which each of the aforementioned authors struggles does not, for the most part, have its source in a Cartesian difficulty (concerning the indirect character of all perceptual contact with the “external” world). None of these authors are evidently concerned to urge a version of (what the Dewey Lectures calls) the “received” view of perception; and Putnam’s efforts to highlight the dubious assumptions underlying the “received” view do not evidently bear on their problems. A fourfold worry about the recent turn that Putnam’s thought has taken in the Dewey Lectures emerges here: (1) that Putnam in the Dewey Lectures tends to conflate the Cartesian and Kantian varieties of problematic potentially at issue in his discussion, (2) that the problematic that is increasingly pressing in contemporary philosophy is the Kantian one and that Putnam does not seem to be sufficiently cognizant of this, (3) that the difficulties to which that problematic gives rise will not be made to vanish simply by exposing what is dubious in the assumptions of the Cartesian skeptic, and (4) that a satisfying treatment of Cartesian skepticism itself requires that we also think through what is confused in Kantian skepticism. This last claim is central both to Kant’s and to McDowell’s own Kant-inspired arguments. Once such a fourfold worry about Putnam’s procedure in the Dewey Lectures comes into focus, it also becomes possible to see what is misplaced in Putnam’s effort to generously credit John McDowell’s Mind and World with having furnished the outline of the diagnosis Putman himself seeks to offer in the Dewey Lectures. Although it is true that McDowell, like Putnam, is concerned to vindicate (what Putnam, following James, calls) “the natural realism of the common man”, it is not an exaggeration to say that McDowell’s diagnosis of the sources of the perplexities with which contemporary philosophy struggles is precisely the reverse of Putnam’s most recent one: where Putnam argues that it is only if we think through what is confused in the Cartesian skeptic’s question “Do we really see houses and chairs?” that will we free ourselves from the confusions visited upon us by the analytical-Kantian skeptical question “how does language hook on to the world?”, McDowell thinks that it is only once we think through what is hopeless about the sort of Kantian bind in which someone like C. I. Lewis threatens to place

Two Varieties of Skepticism

61

himself – one in which it is no longer possible to see how thought can so much as be answerable to reality – that we will be able fully to free ourselves from Cartesian worries about the character of our perceptual relatedness to houses and chairs. For Putnam, in the Dewey Lectures, the Cartesian paradox is the fundamental source of (what Putnam calls) “the broader epistemological and metaphysical issues” that continue to “preoccupy us” in contemporary philosophy.60 For McDowell, in Mind and World, the Cartesian worry is an intelligible, though inept, response to an inchoate form of philosophical anxiety that achieves comparatively fuller expression in the sort of Kantian paradox that, in his view, haunts the projects of authors such as Lewis, Sellars, Davidson and Brandom. This misencounter between Putnam and McDowell is not atypical. The fourth form of philosophical insight that a perspicuous overview of varieties of skepticism can afford is to help us identify such occasions – occasions on which philosophers systematically misunderstand one another’s writings: sometimes paying one another undeserved compliments, and at other times going to battle against one another when their views in no way disagree. In Putnam’s misappropriation of McDowell’s diagnosis of the skeptical paradox that haunts contemporary philosophy, we have a case of merely apparent agreement between two of the finest philosophers of our time. Now let us consider a complementary case – one of merely apparent disagreement.

XI. An Apparent Disagreement: Cavell and Kripke Stanley Cavell’s The Claim of Reason presents a brilliant account of certain aspects of Wittgenstein’s treatment of Cartesian skepticism. Saul Kripke’s book Wittgenstein on Rules and Private Language explores certain aspects of Wittgenstein’s presentation of Kantian skepticism. Each of these books has a hold of a portion of philosophical territory Wittgenstein is concerned to explore that the other misses. Cavell, however, declares himself unable to recognize the problematic that Kripke calls one of “skepticism” to be a variety of skepticism at all, let alone one in which Wittgenstein should interest himself. One can take Kripke to have correctly identified the problem which Wittgenstein means to be grappling with the in the sections which run 60 Putnam (1999), 11.

62

James Conant

from §§185 to 201 of the Philosophical Investigations without having to endorse Kripke’s suggestions for how to understand Wittgenstein’s preferred response to the (so-called) “rule-following paradox” and especially not his attribution to Wittgenstein of a “skeptical solution” to that (or any other) skeptical paradox. In saying that what Kripke calls “skepticism about meaning” is a species of skepticism which Wittgenstein seeks to treat, I mean to affirm nothing more than that the paradox which Kripke finds in Wittgenstein is one that Wittgenstein seeks to address, that it is only one instance of a broader genre of philosophical paradox, that Wittgenstein throughout his writings is repeatedly concerned to formulate and address paradoxes of this genre, and that such paradoxes are ones that belong to that broader genus of philosophical perplexity that Wittgenstein himself often designates – as did Kant before him – through his application of the term “skepticism”. (The scope of the term “skepticism” as employed by Wittgenstein is sufficiently capacious so as to encompass both Cartesian and Kantian varieties of skepticism.) The paradox at which Wittgenstein arrives in section 201 of Philosophical Investigations bears all the earmarks of (what I have been calling) a Kantian skeptical paradox; and the dialectic which threads its way through the preceding sections of the Philosophical Investigations is mired in a Kantian skeptical problematic. Indeed, it is the third variant of the Kantian problematic mentioned early on in this paper – the one that centers on the question “How can a sequence of marks or noises so much as seem to mean something?” Cavell says, in his discussion of Kripke in Conditions Handsome and Unhandsome, that he wishes to “question whether Kripke’s examples illustrate skepticism”, and Cavell evidently feels that something important about his own reading of Wittgenstein would be threatened by the concession that Kripke’s examples illustrate (something that could be called) “skepticism”. But this perception on Cavell’s part stems from a failure to command a clear overview of the shape of the philosophical terrain here. Armed with such an overview, it becomes possible to see that nothing in Cavell’s admirable corpus of work on Wittgenstein would be threatened by such a concession. The term “skepticism” in Cavell’s work – and in his writings on Wittgenstein – exclusively denotes a problematic of the Cartesian variety. To concede, however, that Wittgenstein is interested in variants of the Kantian paradox does not gainsay his equally ubiquitous fascination with their Cartesian counterparts. Cavell, in his discussion of Kripke’s reading, begins with a surely sound observation – to wit: that what Kripke calls Wittgenstein’s “skep-

Two Varieties of Skepticism

63

tical solution” is not anything Wittgenstein would countenance as a “solution” to a philosophical problem. But Cavell moves precipitously from this observation to the conclusion that if Kripke’s solution is not Wittgenstein’s then “the problem to which Kripke offers the solution is not (quite) Wittgenstein’s either”.61 This is a non-sequitur. The reason I think Cavell feels bound to take this step is because he is quite properly unable to recognize the skeptical paradox of section 201 of Philosophical Investigations to be one that bears (what I have called) Cartesian features. Hence Cavell goes on to remark: Kripke reports that sometimes, contemplating the situation of discovery that one may mean nothing at all, he has had “something of an eerie feeling” (21), and that “the entire idea of meaning vanishes into thin air” (22). Is this, I ask myself, like the feelings I have had, under a skeptical surmise, of the world vanishing (as it were behind its appearances), or my self vanishing (as it were behind or inside my body)? These feelings have been touchstones for me of skeptical paradox, of conclusions I cannot, yet become compelled to, believe…. I would like to say that when the entire idea of meaning vanishes into thin air what vanishes was already air, revealing no scene of destruction.62

One can hear Cavell in this passage – and in much of the rest of his subsequent questioning of “whether Kripke’s examples illustrate skepticism” – quite rightly pointing out that Kripke’s alleged skeptical paradox bears none of the Cartesian features: it does not begin with a best case of knowledge, the investigation does not issue in a discovery to which one is unable to accommodate oneself, the conclusions that Kripke draws about his examples do not generalize in accordance with the logic of the Cartesian format, etc. I therefore find myself agreeing with everything that Cavell has to say about this, except his conclusion – that is, I agree that the paradox of section 201 is not a Cartesian paradox, but not that it is not a skeptical paradox.

XII. The Wittgensteinian Way with Skepticism I said earlier that the positive touchstone of the Kantian way with skepticism is a radical following through of the implicit assumptions of a skeptical position up to the point at which the position founders in in61 Cavell (1991), 69. 62 Ibid. 80.

64

James Conant

coherence. The negative touchstone of the Kantian way with skepticism is that it seeks to find a way to respond to the Cartesian skeptic that bypasses the task of having to enter into the details of Cartesian examples, exploring how they are motivated, and considering how they differ from ordinary examples of knowledge. We can contrast the Kantian way with skepticism with what we might call the Wittgensteinian way with skepticism. The Wittgensteinian way is not an alternative to, but rather a supplementation of the Kantian way. The difference between the two ways points up what I take to be utterly original in Wittgenstein’s later treatment of skepticism and what I take to be utterly absent from Kant’s treatment of skepticism. The Wittgensteinian way incorporates a further movement, pushing the skeptic in the opposite direction from the one in which Kant seeks to push him: not only following the skeptic’s presuppositions out to their ultimate consequences, but also examining the initial steps in the Cartesian skeptic’s progress towards doubt, identifying how the skeptic passes from ordinary to philosophical doubt, from a claim to a non-claim context, pinpointing the decisive movement in the philosophical conjuring trick and diagnosing why it is the one that is bound to seem most innocent. Thus we might say, the Kantian way drives the skeptic forward in his doubt, seeking to propel the skeptic to grace by forcing him to pass through utter despair, whereas the Wittgensteinian way supplements this prospective movement with a retrospective one, leading the skeptic back to the point of entry into his problematic, returning him to the lost innocence of the everyday. The Kantian way compels the skeptic to progress further and further forward, further and further from the ordinary, and deeper and deeper into philosophical perplexity, to an ever more violent form of questioning, to the point at which the skeptic’s question consumes itself. The Wittgensteinian way adds to this pressure an additional one that seeks to bring the skeptic back to the place where he started, where he already is and never left, but in such a way that he is able to recognize it for the first time. These two movements that Wittgenstein’s philosophical practice alternately seeks to execute are nicely summarized in the following two remarks: (1) My aim is: to teach you to pass from a piece of disguised nonsense to a piece of undisguised nonsense. (Philosophical Investigations, section 464)

Two Varieties of Skepticism

65

(2) What we do is to bring words back from their metaphysical to their everyday use. (Philosophical Investigations, section 116) I will henceforth refer to these two movements that Wittgenstein seeks to execute as the movement up the dialectical ladder (towards nonsense) and the movement down the dialectical ladder (towards the ordinary). This way of putting things helps to bring out both a fundamental moment of continuity and a fundamental moment of discontinuity between Wittgenstein’s early and later philosophical practice. The former of these two remarks could serve equally aptly as a characterization of the aim of the author the Tractatus. The latter could not. What I am here calling “the Wittgensteinian way” is therefore meant to designate a way with skepticism that we first find only when we turn to Wittgenstein’s later writings. We come therefore now to a further aspect of the fourth form of philosophical insight that a perspicuous overview of the various kinds of philosophical response to skepticism can afford. A number of the finest contemporary philosophers writing about skepticism today misunderstand one another’s writings by specifically failing to appreciate the complementarity of these two movements in later Wittgenstein’s writings, thereby approaching those writings with oppositely occluded blind-spots. H.H. Price and C.I. Lewis – as we saw earlier – are philosophers who are usefully categorized as a Cartesian and Kantian respectively. Price’s thinking moves exclusively within the dialectical space of the Cartesian problematic; Lewis seeks to short-circuit the Cartesian problematic through a prior exploration of the Kantian problematic. John McDowell and Stanley Cavell are, initial appearances to the contrary, philosophers who are not quite so easily categorized. We saw, while reviewing Putnam’s misencounter with McDowell, how McDowell wished to privilege the Kantian problematic over the Cartesian one; and we saw in Cavell’s misencounter with Kripke how Cavell seemed to want to privilege the Cartesian problematic over the Kantian one. Thus, based on the evidence reviewed thus far, the following conclusion might seem tempting: McDowell is primarily concerned to explore the Kantian problematic and Cavell the Cartesian. And that conclusion, in turn, might pave the way for another, possibly equally tempting one, with regard to how each of these philosophers read Wittgenstein: Cavell and McDowell have complimentary blind-spots in their readings of Wittgenstein – each is relatively sensitive to those stretches of Wittgen-

66

James Conant

stein’s writings in which a philosophical problematic bearing one of these sets of features comes in for extended discussion in its own right, and each is relatively insensitive to those stretches of Wittgenstein’s writings in which a philosophical problematic bearing the other comes in for extended discussion in its own right. And there is certainly some truth to this. Cavell is masterful in his exposition of Wittgenstein’s diagnosis and treatment of the ways in which the Cartesian skeptic is led to speak “outside language-games”, but has, for example, little to say that is instructive about Wittgenstein’s treatments of the (Kantian) problematics of rule-following and intentionality. McDowell is masterful in his exposition of the winding dialectic of sections 185 – 201 of Philosophical Investigations and its bearing on the later treatment of, for example, the relation between an expectation and its fulfillment, but has almost nothing to say about the route by means of which the Cartesian skeptic enters his problematic and thus about any of the countless moments Wittgenstein seeks to isolate in which, in our philosophizing, we first begin to lapse into those forms of verbal inflection in which language goes on holiday. If we look more closely at more of Cavell’s and McDowell’s respective writings, however, it becomes less tempting to say that each of them is peculiarly sensitive to only one of these problematics at the expense of the other, either in their own work or in their respective readings of Wittgenstein. Each of them is far too attentive a reader of later Wittgenstein to allow a pervasive focus of philosophical concern in Wittgenstein’s corpus to drop entirely out of view in their own rendering of Wittgenstein’s thought. What happens rather, in their respective writings about Wittgenstein, is that Wittgenstein’s exploration of one of these two skeptical problematics tends alternately to come into focus while the other recedes into the background. And, indeed, I think it is almost inevitable that, in working on later Wittgenstein, one should find this happening to one in one’s writing about him. To illustrate the point, I will place side by side some further passages from Cavell and McDowell – passages in which their roles have been reversed – now McDowell will appear to be the one more preoccupied by the Cartesian dimension and Cavell by the Kantian dimension of Wittgenstein’s concerns.

Two Varieties of Skepticism

67

XIII. A Second Apparent Disagreement: Cavell and McDowell John McDowell, in “Criteria, Defeasibility and Knowledge” argues that (what Wittgenstein calls) criteria are internally related to the justification of claims to knowledge in the following way: if a claim to know that such-and-such turns out not to be justified then the criteria for claiming such-and-such were only apparently satisfied. Thus, for example, if you claim that someone is in pain, and it turns out that that person is only pretending to be in pain, then the criteria for pain were only apparently satisfied. Here is McDowell: Commentators [on Wittgenstein] often take it that the possibility of pretence shows that criteria are defeasible. This requires the assumption that in successful deception one brings it about that criteria for something “internal” are satisfied, although the ascription for which they are criteria would be false. But is the assumption obligatory? Here is a possible alternative; in pretending, one causes it to appear that criteria for something “internal” are satisfied (that is, one causes it to appear that someone else could know, by what one says and does, that one is in say, some “inner” state); but the criteria are not really satisfied (that is, the knowledge is not really available).63

Stanley Cavell, in The Claim of Reason, is concerned to challenge almost exactly the same interpretation of Wittgenstein on criteria that McDowell is concerned to challenge and for many of the same reasons. Yet Cavell comes to (what is at least verbally) precisely the opposite conclusion with regard to how to employ the concept of a criterion in connection with the very sorts of examples that McDowell discusses. Thus Cavell concludes that even if someone is only pretending to be in pain, if it is pain that he is pretending to be in, then his behavior satisfies the criteria for pain. Here is Cavell: [O]nly certain eventualities will count as [someone’s] not being in pain… Circumstances, namely …. in which we will say (he will be) feigning, rehearsing, hoaxing, etc. Why such circumstances? What differentiates such circumstances from those in which he is (said to be) clearing his throat, responding to a joke, etc.? Just that for “He’s rehearsing” or “feigning”, or “It’s a hoax”, etc. to satisfy us as explanations for his not being in pain … what he is feigning must be precisely pain, what he is rehearsing must be the part of a man in pain, the hoax depends on his simulating pain, etc. These circumstances are ones in appealing to which, in describing 63 McDowell (1998), 380.

68

James Conant

which, we retain the concept (here, of pain) whose application these criteria determine. And this means to me: In all such circumstances he has satisfied the criteria we use for applying the concept of pain to others. It is because of that satisfaction that we know that he is feigning pain (i. e., that it is pain he is feigning), and that he knows what to do to feign pain. Criteria are “criteria for something’s being so”, not in the sense that they tell us of thing’s existence, but of something like its identity, not of its being so, but of its being so. 64

The first thing to notice is that McDowell and Cavell appear, at first blush, simply to disagree: in cases of pretending to be in pain, McDowell says the criteria for pain are not satisfied, Cavell says they are. The second thing to notice is that their roles now seem, as promised, to be reversed. McDowell takes Wittgensteinian criteria to operate at a Cartesian level. The question that is settled, according to him, if criteria are satisfied, is one concerning the truth of a claim, the existence of the pain, the reality of the phenomenon. Cavell takes Wittgensteinian criteria to operate at a Kantian level. The question that is settled, according to Cavell, if criteria are satisfied, is not one concerning the truth of a claim but one concerning its purport, not one concerning the existence of something which falls under a concept but the applicability of the concept itself, not one concerning the reality of a phenomenon but one concerning its possibility. Whichever of these two readers of Wittgenstein you take to be on the right track, it is worth noticing that this is the structure of the disagreement here, and that the level at which you take criteria to operate, e. g., in connection with phenomena such as pain, will have decisive consequences for the sort of response to skepticism that will issue from an appeal to “criteria”. Now I myself take it that Cavell has got the merely exegetical question right (about what question is settled if Wittgensteinian criteria are satisfied) and McDowell has got it wrong. But I will not argue that point here. (This does not, however, mean that there is any substantive philosophical disagreement between McDowell and Cavell here.65) What interests me here is the 64 Cavell (1979), 45. 65 Cavell, if he were brought to see how McDowell is employing the term ‘criterion”, could concede, without harm to any of his philosophical commitments, something along the following lines: “Well, that is not how Wittgenstein uses the term; but if you are determined to use the term in this (un-Wittgensteinian) way, then the right (i. e., philosophically Wittgensteinian) thing to go on and say, so using it, is just what you say (e. g., that in cases of pretending the criteria for pain are only apparently satisfied).” And McDowell, equally, if he were taught Cavell’s understanding of the term (criteria are not criteria for

Two Varieties of Skepticism

69

very fact that their disagreement should, at this juncture, have this particular structure. This has a two-fold irony, stemming from the ways in which each of them seems to have given up his previous role in the two misencounters canvassed above. The first irony is to be found in the fact that in much of his writing about Wittgenstein McDowell reads Wittgenstein as seeking to address Kantian skepticism and in Mind and World follows his Wittgenstein in seeking to take an exclusively Kantian way with skepticism. McDowell tends to see the Cartesian craving for epistemic security as an intelligible, though inept, response to an inchoate form of the philosophical anxiety that only arrives at clear expression when it is posed as a Kantian skeptical paradox. Thus, despite his Cartesian construal of the grammar of the concept of a criterion, McDowell tends to see the treatment of philosophical skepticism as requiring attention only to the Kantian problematic. He assumes, in the manner of Kant himself, not only that the Cartesian skeptical paradox can be shown to be merely a special case of a more general worry, but more importantly that, once this is shown, Cartesianism will be robbed of all its force and will wither away of its own accord without requiring any additional form of specialized treatment. The second irony to be noted in connection with the structure of McDowell’s and Cavell’s disagreement about Wittgenstein’s concept of a criterion lies in the fact that Cavell, though he is far too acute a reader of Wittgenstein to fail to appreciate that the Philosophical Investigations is frequently operating at a Kantian level, nonetheless, never sees that for Wittgenstein, as for Kant, the term “skepticism” ranges over far something’s being so, but for its being so) would have no reason not to concede that the right thing to say, so using the term, is just what Cavell says (i. e., that in cases of pretending, if it is pain you are pretending to be in, then the criteria for pain are satisfied). The disagreement between Cavell and McDowell about criteria is a further instance of a merely apparent disagreement in which the appearance of disagreement is engendered through an unacknowledged transition from a Cartesian to a Kantian problematic. Both McDowell and Cavell are in profound disagreement with their respective and very similar sets of interlocutors (Baker, Wright, Albritton, Malcolm, etc.), but they are disagreeing with them about different things. (McDowell’s point has to do with how justification and knowledge are internally – and not merely externally – related to one another; Cavell’s has to do with how an appeal to criteria cannot do the work that Wittgenstein calls upon it to do, if criteria are understood in the manner of such commentators.) Their respective philosophical motivations for disagreeing with such commentators are perfectly compatible.

70

James Conant

more than a merely Cartesian skeptical problematic – thus over far more varieties of philosophical perplexity than a merely Cartesian inflection of the term is able to encompass. Nonetheless, Cavell, through carefully following out the inner movement of the dialectic traced in Wittgenstein’s investigations, often finds himself fetching up in the terrain of the Kantian problematic. Here is a characteristic passage that may stand for a thousand others: If you do not know the (non-grammatical) criteria of an Austinian object (can’t identify it, name it) then you lack a piece of information, a bit of knowledge, and you can be told its name, told what it is, told what it is (officially) called. But if you do not know the grammatical criteria of Wittgensteinian objects, then you lack, as it were, not only a piece of information or knowledge, but the possibility of acquiring any information about such objects berhaupt; you cannot be told the name of that object, because there is as yet no object of that kind for you to attach a forthcoming name to.66

What is threatened with the loss of (what Cavell calls) Wittgenstein criteria is not merely the possibility of isolating an unimpeachable item of knowledge, but the possibility of so much as turning up a candidate for knowledge. At one point in The Claim of Reason, Cavell quotes the following four passages from Philosophical Investigations in rapid succession: … [O]nly of a living human being and what resembles (behaves like) a living human being can one say: it has sensations; it sees; is blind; hears; is deaf; is conscious or unconscious. (section 281) What has to be accepted, the given, is – so one could say – forms of life. (section 226) What gives us so much as the idea that living beings, things, can feel? (section 283) The human body is the best picture of the human soul. (section 178)

These are passages in which Wittgenstein presents what Cavell says he understands “as the background against which our criteria do their work; even make sense”67. They are also quintessential examples of moments in his work in which Wittgenstein’s investigations move (from the Cartesian) to the Kantian level. And, in commenting on these passages, Cavell (taking the problem of other minds here as his example of a skeptical problematic) expresses what he takes to be the significance of such passages in Wittgenstein’s work: 66 Cavell (1979), 77. 67 Ibid., 83.

Two Varieties of Skepticism

71

To withhold, or hedge, our concepts of psychological states from a given creature, on the ground that our criteria cannot reach to the inner life of the creature, is specifically to withhold the source of my idea that living beings are things that feel; it is to withhold myself, to reject my response to anything a living being; to blank so much as my idea of anything as having a body. To describe this condition as one in which I do not know (am not certain) of the existence of other minds is empty. There is now nothing there of the right kind, to be known. There is nothing to read from that body, nothing the body is of; it does not go beyond itself, it expresses nothing; it does not so much as behave. There is no body left to manifest consciousness (or unconsciousness). It is not dead, but inanimate; it hides nothing…. My problem is no longer that my words can’t get past his body to him. There is nothing for them to get to; they can’t even reach as far as my body…. The signs are dead; merely working them out loud doesn’t breathe life into them; even dogs can speak more effectively.68

I take the presence of a Kantian problematic in this passage to be selfevident. Though Cavell in such passages (and there are many such passages in The Claim of Reason) sees that the transition to a Kantian problematic plays an essential role in Wittgenstein’s treatment of skepticism, he seems only able to interest himself in this transition to the extent that it forms part of a response to Cartesian skepticism; hence his view of the scope of this problematic in Wittgenstein’s writing is artificially blinkered. Central issues – concerning, for example, the nature of the accord between a rule and its application (not only, as in certain employments of the mathematical case, as a trope for learning a word, but as a perplexing instance of the Kantian problematic in its own right), between an expectation or wish and its fulfillment, etc – recede into the background of Cavell’s reading of Wittgenstein. This blind-spot in Cavell’s reading of Wittgenstein (and in his reading of Kant) comes perhaps most visibly to the surface in his discussion of Kripke’s reading of Wittgenstein. Although he sees that Wittgenstein is concerned to explore the Kantian problematic, he does not see (and feels that he must not allow) that it can issue in a distinctive (and, as I have tried to show, distinctively Kantian) variety of philosophical paradox – one that Wittgenstein, like Kant, views as a variety of skepticism. This leaves Cavell’s account of Wittgenstein’s treatment of skepticism essentially incomplete. Only a reading able to accommodate, both exegetically and philosophically, the insights contained in both Cavell’s and McDowell’s respective readings of Wittgenstein – that is, only one that understands why, in 68 Ibid. 83 f.

72

James Conant

Wittgenstein’s treatment of the extended philosophical dialectic of which they each form a part, neither variety of skepticism is to be privileged over the other – will be complete.

XIV. Conclusion The aim of this paper has been to suggest that a perspicuous overview of various kinds of skepticism and the kinds of response they engender affords a number of different kinds of philosophical benefit: it allows one to command a clearer view of the sorts of relations that obtain across apparently distinct areas of philosophy, it allows one to distinguish the very different sorts of things philosophers mean when they employ (what is apparently) the same philosophical vocabulary (and especially when they employ the word ‘skepticism’), it allows one to see more clearly what is distinctive about Kant’s and Wittgenstein’s respective contributions to the history of thought about skepticism, and it allows one to see more clearly why these authors are often misread and why their best commentators often misread one another and misunderstand themselves – taking themselves to disagree with one another when they do not, taking themselves to agree with one another when they do not, and taking themselves to be in agreement with themselves when they are not.69

References Austin (1979): John L. Austin, Philosophical Papers, Oxford. Cavell (1979): Stanley Cavell, The Claim of Reason, Oxford. Cavell (1981): Stanley Cavell, Pursuits of Happiness, Cambridge/MA. Cavell (1987): Stanley Cavell, Disowning Knowledge, Cambridge/MA. Cavell (1991): Stanley Cavell, Conditions Handsome and Unhandsome, Chicago. Cavell (1996): Stanley Cavell, Contesting Tears, Chicago. Conant (1992): James Conant, “The Search for Logically Alien Thought: Descartes, Kant, Frege and the Tractatus”, in: The Philosophy of Hilary Putnam, Philosophical Topics 20 (1), 115 – 180.

69 As should be evident to anyone familiar with their work, this paper is pervasively indebted to the writings of Stanley Cavell, John McDowell, and Hilary Putnam. A much shorter version of this paper was published in Wittgenstein and Skepticism, edited by Denis McManus.

Two Varieties of Skepticism

73

Conant (2001): James Conant, “In the Electoral Colony: Kafka in Florida”, in: Critical Enquiry 27 (4), 662 – 702. Descartes (1986): René Descartes, Meditationes de Prima Philosophia, Stuttgart. Kant (1997): Immanuel Kant, Prolegomena to Any Future Metaphysics, Cambridge. Kant (1998): Immanuel Kant, Critique of Pure Reason, Cambridge. Kripke (1982): Saul Kripke, Wittgenstein on Rules and Private Language, Cambridge/MA. Land (2012): Thomas Land, “Kantian Conceptualism”, in: G. Abel/J. Conant (eds.), Rethinking Epistemology, Vol. 1, Berlin/New York, 197 – 239. Lewis (1929): Clarence Irving Lewis, Mind and the World Order, New York. McDowell (1994): John McDowell, Mind and World, Cambridge/MA. McDowell (1998): John McDowell, “Criteria, Defeasibility, and Knowledge”, in: J. McDowell, Meaning, Knowledge, & Reality, Cambridge/MA, 369 – 394. Price (1932): Henry Habberley Price, Perception, London. Putnam (1994): Hilary Putnam, Words and Life, Cambridge/MA. Putnam (1999): Hilary Putnam, The Threefold Cord, New York. Williams (2001): Michael Williams, Problems of Knowledge, Oxford. Wittgenstein (1973): Ludwig Wittgenstein, Philosophical Investigations, New Jersey.

I. First-, Second-, and Third-Personal Knowledge

Knowing What You Believe* Quassim Cassam I A familiar thesis is that knowledge of our own thoughts, beliefs and other attitudes is normally immediate, that is to say, not normally based on observation, evidence or inference.1 Many philosophers who endorse this immediacy thesis take it to be obviously correct. They think that it is the immediacy of self-knowledge rather than its supposed infallibility which makes it epistemologically distinctive, and that the hard question in this area is not whether self-knowledge can be immediate but how it can be so. The immediacy of self-knowledge can seem puzzling because knowledge of our own thoughts and beliefs is surely knowledge of contingent facts, and the usual presumption is that knowledge of contingent matters must be based on observation, evidence or inference.2 So if it is just a contingent fact about me that I believe that P, and yet I know without observation, inference or evidence that I have this belief, then it needs to be explained how this is possible. One explanation, given by Richard Moran among others, appeals to the notion of transparency. The idea is that “a person answers the question whether he believes that P in the same way he would address himself to the question whether P itself”.3 The question “Do you believe that P?” is, in this sense, transparent to the question “Is it the case that P?”, which means that I can answer the former question “by consider* 1 2

3

Originally published in Proceedings of the Aristotelian Society CXI (2011), 1 – 23. Reprinted by courtesy of the Editor of the Aristotelian Society: © 2011. See, for example, Davidson (1994) and Moran (2001). As Boghossian points out, there are contingent judgements which one may be justified in making “even in the absence of any empirical evidence” (Boghossian (1998), 166). An example is the judgement I am here now. Immediate knowledge of one’s own beliefs is, however, not relevantly similar to such examples, and cannot be explained in the way that my immediate knowledge that I am here now can be explained. Moran (2004), 457.

78

Quassim Cassam

ation of the reasons in favour of P itself”.4 The proposal is that this way of knowing one’s beliefs is non-observational, non-inferential and nonevidential, and that knowledge of the admittedly contingent fact that one believes that P is immediate because the question “Do you believe that P?” is transparent to a corresponding question about the world. So the thesis is that transparency explains immediacy (TEI).5 Knowledge of one’s beliefs and other standing attitudes is not the only supposedly immediate self-knowledge. There is also knowledge of one’s sensations and passing thoughts, that is, thoughts that just occur to one.6 These are occurrent states, and TEI does not purport to explain how immediate knowledge of them is possible. Why not? Here is one suggestion: the states of mind that fall within the purview of TEI are propositional attitudes. Since sensations are not propositional attitudes it is not surprising that TEI has no bearing on them. But this can’t possibly be the whole story because many of our thoughts, including passing thoughts, are propositional attitudes. So it is not clear why TEI doesn’t apply to passing thoughts if it applies to beliefs. And if TEI doesn’t explain how immediate knowledge of our passing thoughts is possible then what does explain it? I have three main questions: (1) Is the immediacy thesis correct? (2) Is TEI correct? (3) How can we have immediate knowledge of our passing thoughts? I’m going to take these questions in reverse order. The first thing to figure out in relation to (3) is why the transparency procedure can’t account for our immediate knowledge of our own passing thoughts. My proposal is that among the relevant factors is what I’m going to refer to as the passivity of passing thoughts. Passing thoughts are passive in the sense that they are (i) not necessarily responsive to reason and (ii) states from which one can distance or dissociate oneself. Between them (i) and (ii) help to explain why the transparency procedure can’t deliver 4 5 6

Moran (2003), 405. For a clear statement of this thesis see Moran (2004), 457. Moran writes: “There are two basic categories of psychological state to which the ordinary assumption of ‘privileged access’ is meant to apply: occurrent states such as sensations and passing thoughts, and various standing attitudes of the person, such as beliefs, emotional attitudes, and intentions” (Moran (2001), 9). He is explicit that his account of self-knowledge is not intended to apply to sensations.

Knowing What You Believe

79

immediate knowledge of one’s passing thoughts. My main positive suggestion in connection with (3) builds on Ryle’s insight that ‘much of our ordinary thinking is conducted in internal monologue or silent soliloquy’.7 I claim that this remark points to a promising response to (3): I know that I am having the thought that P by being conscious of saying to myself that P in my internal monologue.8 With regard to (2), there is a problem with TEI. Even if I can answer the question “Do I believe that P?” by consideration of the reasons in favour of P the resulting knowledge is not immediate. In fact, there are two notions of immediacy in play, one epistemic and the other psychological. My knowledge that P is epistemically immediate only if my justification for believing that P does not come, even in part, from my having justification to believe other, supporting propositions.9 My knowledge that P is psychologically immediate only if it is not acquired by conscious reasoning or inference. If I come to know that I believe that P by employing the transparency procedure then my knowledge does not appear to be immediate in either sense. Since it is knowledge acquired by conscious reasoning it cannot possibly be psychologically immediate. It can’t be epistemically immediate either because, as we will see, my justification for believing that I believe that P does come, at least in part, from my having justification to believe other propositions. The transparency procedure gives me inferential selfknowledge and inferential knowledge isn’t immediate. It will turn out that some of the reasons for thinking that the transparency procedure doesn’t give us immediate knowledge of our passing thoughts are also reasons for thinking that it doesn’t give us immediate knowledge of our beliefs. If passing thoughts are passive the same is true of some beliefs. The latter are not necessarily responsive to reason and there are also beliefs from which it is possible to distance oneself. The point is not that the passivity of belief is exactly parallel to the passivity of passing thoughts. Given what it is for a mental state to be a belief there are much tighter restrictions on the passivity of belief. All the same, to the extent that belief can be unresponsive to reason this puts extra pressure on the idea that I can come to know whether I believe that P by considering the reasons in favour of P. 7 8 9

Ryle (1949), 28. Ryle’s view is also discussed in Byrne (2011), but Byrne wrongly insists that there is no such thing as inner speech. This is essentially the account of immediate justification given in Pryor (2005).

80

Quassim Cassam

That leaves (1). Is it obvious that self-knowledge is immediate? Perhaps the thought is this: if asked whether I believe that P I can usually answer straight off. I don’t generally need to work out whether I believe that P, I just know. But this only shows that my self-knowledge is psychologically immediate, and this does not entail epistemic immediacy. In my view, it is not a datum that knowledge of our own beliefs is epistemically immediate, or that accounts of self-knowledge that fail to secure its epistemic immediacy are necessarily flawed. Still, the fact remains that if the object of the exercise is to explain how knowledge of our own beliefs could be epistemically immediate then the transparency account doesn’t work.

II Consider the following passage from Harry Frankfurt’s paper ‘Identification and Externality’: In our intellectual processes, we may be either active or passive. Turning one’s mind in a certain direction, or deliberating systematically about a problem, are activities in which a person engages. But to some of the thoughts that occur in our minds…. we are passive bystanders. Thus there are obsessional thoughts, whose provenance may be obscure and of which we cannot rid ourselves; thoughts that strike us unexpectedly out of the blue; and thoughts that run willy nilly through our heads. The thoughts that beset us in these ways do not occur by our own active doing. It is tempting, indeed, to suggest that they are not thoughts that we think at all, but rather thoughts that we find occurring within us. This would express our sense that, although these thoughts are events in the histories of our minds, we do not participate actively in their occurrence…. It is not incoherent, despite the air of paradox, to say that a thought that occurs in my mind may or may not be something that I think. 10

What Frankfurt describes here is a perfectly familiar phenomenon.11 The successive thoughts with respect to which we are passive bystanders are what I am calling passing thoughts but passing thoughts come in differ10 Frankfurt (1998), 59. 11 It is also one that is noticed by Sellars. He points out that “thinking is often a deliberate action, as in thinking about (i. e. attempting to solve) a problem” but that “there is a sense of ‘thought’ in which thoughts just occur to one. We say ‘it suddenly occurred to me that….,’ and, we can often add ‘for no reason’” (Sellars (1975), Lecture 1, #29). It is in cases of the latter kind that we begin to see the full force of the Lichtenbergian suggestion that Descartes should have said ‘There is thinking’ rather than ‘I think’.

Knowing What You Believe

81

ent varieties. Obsessional thoughts are one variety. A jilted lover finds herself thinking obsessively about her ex and cannot concentrate on anything else. For all his assurances that no one else is involved, and despite the absence of convincing contrary evidence, she can’t get rid of the thought that he is seeing someone else. In contrast, the thoughts that strike us out of the blue need not be obsessive. As I write these words it suddenly occurs to me that today is the first of the month. This thought is unconnected with the deliberative thinking in which I am currently engaged; it just comes to me, for no apparent reason, but I am hardly obsessed with today’s date. The final category of passing thoughts, those that Frankfurt describes as running willy nilly through our heads, might include the thoughts that run through one’s head as one gazes absent mindedly out of the window on a train journey or just before one falls asleep.12 The sense in which such thoughts run through our heads willy nilly is that they are not the product of deliberation and are not in any obvious way rationally ordered. If we can be said to be thinking these thoughts at all, we do so ‘not by concentrating on anything in particular’ but by ‘moving from one idea to the next in an endless chain of associations’.13 How do we know our passing thoughts? Suppose that one of my passing thoughts is the thought that P. How do I know that I am thinking that P or having the thought that P? Not by employing the transparency method. The jilted lover does not know of the occurrence of the thought that her former lover is seeing someone else by asking “Is he seeing someone else?”, or by considering the reasons in favour of the proposition that he is seeing someone else. She might realize that the reasons in favour of this proposition are weak but this doesn’t alter the sad fact that the thought that he is with someone else keeps her awake at night. In contrast, there might be good reasons for thinking that today is the first of the month but it is still implausible that I come to know that this is what I am thinking by consideration of such reasons. The same goes for the random thoughts that run through my head willy nilly on a train journey or as I am falling asleep. I don’t come to know that I am having these thoughts by consideration of the reasons in favour of them or their contents. 12 Richard Moran suggested (in conversation) that the thoughts that run through one’s head just as one is about to fall asleep are the best example of ‘passing thoughts’. 13 Bollas (2009), 6.

82

Quassim Cassam

The point is this: consideration of the reasons in favour of P only reveals whether I am thinking that P on the assumption that what I am thinking is somehow determined by those reasons or by my reflection on them.14 Such an assumption would be implausible in relation to passing thoughts. This is especially clear in relation to obsessional thoughts since part of what makes them obsessional is their unresponsiveness to reason. I can find myself thinking or worrying that P even though I realize that there is no reason to think or worry that P. If I realize that there is no good reason for thinking that P then asking whether it is true that P will not tell me that I am thinking that P. Ex hypothesi, there is no reason to think that P is true but this doesn’t alter the fact that the thought that P keeps running through my head. In the case of nonobsessional passing thoughts the point is not that they are not responsive to reason but that they aren’t necessarily responses to reason. For example, the thought that today is the first of the month might be extinguished by evidence to the contrary. It is in this sense responsive to reason, and yet it did not occur to me that today is the first of the month because I gave the matter any thought. The thought came to me for no apparent reason, and if the thought that P is not the product of rational reflection then it is hard to see why reflection on the reasons in favour of P should have any bearing on whether one is in fact thinking it. Another way of making this point would be to note that one can distance oneself from one’s passing thoughts. Distancing oneself from a thought means recognizing that it is not well-founded or is in some other way inappropriate. Sitting on a plane, “the thought that it is going to crash comes to me, and perhaps I find that I cannot help thinking that it will crash”.15 Here it is a “datum of consciousness”16 that I am thinking that the plane will crash even though I am well aware that there is no real evidence that it will crash. In this case, I am effectively alienated from one of my own occurrent thoughts. I do not treat the question whether I am thinking that P as equivalent to the question whether P is true. By the same token, I don’t come to know that I am thinking that P by asking myself whether P is true. 14 The point I am making here is analogous to one that Moran makes when he is discussing the conditions under which consideration of the reasons in favour of P can tell me whether I believe that P. See Moran (2003), 405, and section III below for further discussion. 15 Hampshire (1965), 101 f. 16 Hampshire (1965), 101.

Knowing What You Believe

83

To say that passing thoughts are not necessarily responsive, or responses, to reason and that one can distance oneself from one’s passing thoughts is to say that such thoughts are ones with respect to which one is passive. What I am suggesting is that it is the passivity of passing thoughts that partly explains why they fall outside the purview of TEI. It is not just that the transparency account fails to explain how knowledge of our passing thoughts can be immediate. The deeper worry about this account is that it fails as an account of knowledge of our passing thoughts. A different account is needed, one that does justice to the passivity of passing thoughts and to the sense in which they are conscious occurrences. In a series of papers Tyler Burge develops what looks like a promising alternative to the transparency account of self-knowledge. Imagine I am asked what I am thinking and that I respond by judging: I am thinking that writing requires concentration. This judgement is self-verifying in this sense: judging that I am thinking that writing requires concentration makes it true that I am thinking that writing requires concentration. It makes it true because “the cognitive content that I am making a judgement about is self-referentially fixed by the judgement itself”.17 My judgement is based on nothing else and so is my knowledge of what I am thinking. In particular, I don’t come to know what I am thinking by considering the reasons in favour of the proposition that writing requires concentration. How does this approach deal with examples like the fearful flyer? Take the case in which the thought that the plane is going to crash occurs to me at t1. At that precise moment you happen to ask me what I am thinking. A moment later, at t2, I tell you: I am thinking the plane is going to crash. What I have told you is true but what makes it true? Burge’s account implies that what makes it true is my judgement at t2. If I judge at t2 that I am thinking that the plane is going to crash I thereby make it true at t2 that I am thinking that the plane is going to crash. However, I do not thereby make it true that at t1 I was thinking that the plane is going to crash; the judgement “I have just been thinking that the plane is going to crash” is not self-verifying, unlike the judgement “I am thinking that the plane is going to crash”.18 17 Burge (1998), 120. 18 As Boghossian points out in a helpful discussion of Burge’s views. See Boghossian (1998).

84

Quassim Cassam

This is a problem for Burge because when someone asks me at t1 what I am thinking they presumably want to know what I am thinking at that time, or just before t1. That is why, in response to the question “What are you thinking?”, it is facetious to say “I’m thinking about your question”. By the same token, what makes it appropriate for me to judge at t2 that I am thinking that the plane is going to crash is the fact that I was thinking that the plane is going to crash when I was asked the question at t1. What makes my answer appropriate is not my judging at t2 that I am thinking that the plane is going to crash. It is in this sense that I would be mistaken if at t2 I were to judge “I am thinking that the plane won’t crash”. In the act of judging this I would be making it true that I am thinking that the plane won’t crash but I wouldn’t make it true that this is what I have been thinking. What I have just been thinking is that the plane will crash. Moreover, I know that this is what I have been thinking by being aware of my own thoughts. If this is right then the way to understand how we know our passing thoughts is not to focus on the fact that judgements like “I am thinking that the plane will crash” are self-verifying but to try to understand the distinctive way in which we are aware of our passing thoughts. A convincing account of the distinctive way in which we are aware of our passing thoughts will need to be grounded in a convincing account of what it is for one to have such thoughts in the first place. For example, in what sense do I find myself thinking that the plane is going to crash? What is it for this thought suddenly to occur to me? Perhaps I have mental images of the plane crashing but it is not clear that this amounts to thinking, as distinct from merely imagining, that the plane is going to crash. But now suppose that as the plane picks up speed down the runway I find myself saying to myself that the plane is going to crash. I might say this out loud but consider the case in which the saying is a purely inner saying, a saying in inner speech or in what Ryle calls ‘silent soliloquy’.19 Here are three suggestions: (a) Inwardly saying to myself that P is, or is closely related to, thinking that P. (b) In being aware of saying to myself that P I am aware of having the thought that P. 19 Sellars is someone else who purports to “take very seriously the view that a thought, in the sense in which thoughts occur to one, is the occurrence in the mind of sentences in the language of ‘inner speech’” (Sellars (1975), Lecture 1, #31).

Knowing What You Believe

85

(c) I know that I am having the thought that P by being aware of saying that P. With regard to (a), the suggestion is not that all thinking is like this but that some is. If there is no such thing as inner speech then (a) is a nonstarter. I take it as obvious that there is inner speech but will not attempt to analyse or explain this notion any further here.20 One sense in which thoughts occur to one is that sentences in the language of inner speech occur to one; to find oneself saying to oneself that P is to find oneself thinking that P. This could either be because saying to oneself that P is or constitutes thinking that P or because, even though it isn’t the same as thinking that P, it discloses that this is what one is thinking. Either way, inwardly saying to oneself that P is closely related to thinking that P. This leads to (b). If saying that P is, or is very closely related to, thinking that P then it is plausible that in being aware of saying to myself that P I am aware of having the thought that P. Indeed, part of what it is for thinking that P and inwardly saying that P to be closely related is for awareness of inwardly saying that P to amount to awareness of thinking that P. The hard question is: in what sense is one ‘aware’ of saying to oneself that P if the saying is a saying in inner speech? Auditory metaphors are virtually inescapable. The sense in which one is aware of inwardly saying to oneself that P is that one “hears” oneself saying to oneself that P. This is hearing with the mind’s ear rather than with the ears attached to one’s skull.21 No doubt much more needs to be said about this form of awareness but to deny its existence is to deny the existence of something which certainly seems phenomenologically real. That leaves (c). If I am aware of saying to myself that P, and thereby aware of myself thinking that P, or having the thought that P, that enables me to know that I am thinking that P. This is what Ryle is getting at when he observes that our internal monologues ‘disclose’ our frames of mind, and that eavesdropping on our internal monologues enables us 20 See Carruthers for further discussion. As he points out, “many of us are inclined to report, on introspective grounds, that at least some of our conscious propositional thinking is conducted in imaged natural-language sentences” (Carruthers (2005), 116). In studies, all subjects report at least some instances of inner speech. So “the existence of inner speech, itself, isn’t – or shouldn’t be – in doubt” (ibid.). According to the cognitive conception of language which Carruthers wants to defend, inner speech is partly constitutive of thinking. Inner speech is not merely expressive of thought, or merely what gives us access to our thoughts. 21 I owe the expression “the mind’s ear” to Alex Byrne. See Byrne (2011).

86

Quassim Cassam

to describe our frames of mind. Is the resulting knowledge immediate? If immediate knowledge must not be based on observation or perception then one issue is whether “hearing” oneself think is a form of perception. That depends on how generously the notion of perception is construed but it is not clear, in any case, why perceptual knowledge cannot be immediate.22 A far more pressing question is whether knowledge of what one is thinking is a form of inferential or evidence-based knowledge. Not if inwardly saying that P is thinking that P. Rather, direct awareness of one’s inner speech would amount to direct awareness of one’s thoughts. Even on the view that our internal monologues disclose our thoughts without being our thoughts there is still no reason to classify knowledge of one’s thoughts as mediated. To regard one’s inner speech as disclosing one’s thoughts is not to be committed to the view that one infers one’s thoughts from one’s inner speech, or that inwardly saying to oneself that P is merely a reliable sign or good evidence that one is having the thought that P. The connection between inner saying and thinking is more intimate than that. The passivity of passing thoughts is now easily explained. Inwardly saying to oneself that the plane is going to crash need not be the product of reflection on the reasons in favour of the proposition that the plane is going to crash and may well be entirely unresponsive to such reflection. We can distance ourselves from our passing thoughts in the sense that we can distance ourselves from what we find ourselves saying. We may recognize that some of our inner utterances are ill-founded but their occurrence is still not in question. As for the idea that it is a datum of consciousness, in Hampshire’s example, that I am thinking that the plane is going to crash, the sense in which this is so is that it is a datum of consciousness that I am saying to myself that the plane is going to crash. The latter is a datum of consciousness in the sense that I can either literally or metaphorically hear my own utterances. We now have an account of the nature of some of our passing thoughts, the distinctive way in which we are aware of them, and how this awareness grounds our knowledge of our passing thoughts. This account respects the passivity of passing thoughts and their distinc22 Indeed one might think that perceptual knowledge is the paradigm of immediate knowledge. There might be good reasons for denying that self-knowledge is perceptual but it is not the immediacy of self-knowledge that counts against its being perceptual unless one thinks that perceptual knowledge is itself inferential.

Knowing What You Believe

87

tive phenomenology. It explains how immediate knowledge of our passing thoughts is possible and does so without appealing to the notion of transparency. The suggestion is not just that immediate knowledge of our passing thoughts can be explained without appealing to the transparency procedure but that immediate knowledge of our passing thoughts cannot be explained by the transparency procedure.

III The next question is: is TEI correct? Remember that this thesis is limited in scope. It is only concerned with immediate knowledge of our own beliefs and other standing attitudes so the fact that immediate knowledge of our passing thoughts can’t be explained by reference to transparency is neither here nor there as far as TEI is concerned. Having said that, it is not clear that TEI is successful even on its own terms. There are reasons for thinking that the transparency procedure does not explain the supposed immediacy with which we know our own beliefs. Furthermore, despite all the differences between having the passing thought that P and believing that P, these reasons are not unrelated to the reasons for thinking that the transparency procedure can’t explain how we can have immediate knowledge of our passing thoughts. We can clarify these concerns about TEI by focusing on two specific problems facing the transparency explanation of immediate self-knowledge. I will refer to the first of these as the two questions problem and to the second as the sticking problem. The former is one to which Moran himself draws attention. The challenge is to understand how the question “Do you believe that P?” can be transparent to the question “Is it the case that P?”. The first of these questions is inward-directed while the second is outward-directed. These questions have quite different subjectmatters, and it is conceivable both that I believe that P when P is false, and that I don’t believe that P when P is true. In that case, as Moran asks, “what right have I to think that my reflection on the reasons in favour of P (which is one subject-matter) has anything to do with the question of what my actual belief about P is (which is a quite different subject-matter)?”23

23 Moran (2003), 405.

88

Quassim Cassam

Moran gives his response to the two questions problem in this passage, in which he is discussing the relationship between the questions “Is it raining?” and “Do you believe that it is raining?”: And then my thought at this point is: I would have a right to assume that my reflection on the reasons in favour of rain provided an answer to the question of what my belief about rain is, if I could assume that what my belief here is was something determined by the conclusion of my reflection on those reasons. An assumption of this sort would provide just the right link between the two questions. And now, let’s ask, don’t I make just this assumption, whenever I’m in the process of thinking my way to a conclusion about some subject-matter? 24

The conclusion of my reflection on the reasons in favour of P is a judgement. Assuming that my belief concerning P is determined by the conclusion of my reflection on the reasons in favour of P is therefore equivalent to assuming that my belief concerning P is determined by whether I judge that P. Call this the linking assumption (LA). Moran represents the linking assumption as one which I actually make, and am entitled to make, when I am in the process of thinking my way to some conclusion. The specific role of this assumption is to connect the inward-directed and the outward-directed questions in such a way as make it intelligible that I am entitled to answer the former by answering the latter. Insofar as I rely on LA in coming to know that I believe that P is my knowledge of what I believe still immediate? This divides into two questions, one concerning psychological immediacy and the other concerning epistemic immediacy. Starting with the former, there is a straightforward reason for thinking that the transparency doesn’t give us psychologically immediate self-knowledge: to come to know whether I believe that P by consideration of the reasons in favour of P itself is to come to know whether I know that P by reasoning. Since the reasoning is conscious it follows immediately that the resulting self-knowledge is not immediate in the psychological sense.25 Another way of making the point would be to emphasize that, on Moran’s view, self-knowledge is arrived at by deliberation. But to deliberate is to reason, and psycho24 Moran (2003), 405. 25 Dorit Bar-On argues that the transparency method is “epistemically rather indirect” to the extent that it implies that self-judgements “are arrived at on the basis of consideration of wordly items” (Bar-On (2004), 113). In my terms, the indirectness implied by this characterization of the transparency method is primarily psychological.

Knowing What You Believe

89

logically immediate knowledge is knowledge that is not arrived at by conscious reasoning or inference. As for whether transparency delivers epistemically immediate knowledge we need to go back to the definition of epistemic immediacy. A person’s knowledge that P is immediate in this sense only if his justification for believing that P does not come, even in part, from his having justification to believe other, supporting propositions. When his justification comes from his having justification to believe other, supporting propositions then his knowledge is inferential. Now consider the following simple argument: I come to know that I believe that P by following the transparency method only if doing so gives me a justification for believing that I believe that P. But when I rely on transparency in coming to know what I believe my justification for believing that I believe that P comes in part from my having justification to believe at least one other proposition, namely, the linking assumption. So my knowledge that I believe that P is epistemically inferential, and therefore not immediate, when it is based on the transparency procedure. There are several ways of trying to block this argument, none of them convincing. For example, there is the thought that for the purposes of connecting up the inward-directed and outward-directed questions LA isn’t an assumption that the self-knower actually has to make; it is enough that it is an assumption to which, as a rational being, he is entitled to make. But if LA is not an assumption I actually make, at least implicitly, then how am I supposed to find it intelligible that I can answer the question whether I believe that P by considering the reasons in favour of P itself ? It is because the inward- and outward-directed questions have different subject-matters that something is needed to link them, and the something had better not be beyond my ken if it is to make it intelligible to me, the knower, and not just the theorist of self-knowledge, that I can answer the inward-directed question by answering the outward-directed question. So it won’t do to represent LA as an assumption that somehow links the two questions without at the same time mediating my justification for believing that I believe that P. Another move might be to argue along the following lines: all knowledge has general background assumptions or presuppositions but this is not to say that all knowledge is, in any interesting sense, inferential.26 It all depends on the precise nature and role of the back26 Thanks to Crispin Wright for suggesting this response.

90

Quassim Cassam

ground assumptions. Even non-inferential knowledge can have presuppositions if they are what one might call structural presuppositions. Such presuppositions are, for example, presuppositions of thought or rational deliberation as such whose role is to enable knowledge or reasoning without also serving as mediating premises that contribute to the justification of one’s beliefs. So if, on the transparency account, LA is only a structural presupposition of self-knowledge it doesn’t follow that this account makes self-knowledge inferential in any interesting sense. Is LA a presupposition of rational deliberation as such? It is one thing to agree that if I am rational then whether or not I believe that P will generally be determined by my reflection on the reasons in favour of P. It is another matter entirely whether, as a rational being, I must believe that this is how my beliefs are determined. It is questionable whether a commitment to LA in this sense is a condition of rational deliberation and yet this is what the transparency account seems to require. In addition, it is very hard not to read the transparency account as implying that LA is a mediating premise that contributes to the justification of one’s beliefs about one’s beliefs. At any rate, if LA is not a mediating premise then, as argued above, it is obscure how it is supposed to make it intelligible to one that it is possible to answer an inward-directed question by answering a corresponding outward-directed question.27 So there remains the strong suspicion that the self-knowledge that the transparency procedure makes available is not epistemically immediate. This suspicion is confirmed by the sticking problem. The following example from Peacocke illustrates the problem: Someone may judge that undergraduate degrees from countries other than their own are of an equal standard to her own, and excellent reasons may be operative in her assertions to that effect. All the same, it may be quite clear, in decisions she makes on hiring, or in making recommendations, that she does not really have this belief at all.28

The point is this: the conclusion of my reflection on the reasons in favour of P is a judgement but concluding or judging that P is not the same as believing that P. Normally when I judge that P I also believe that P but I can judge that P without believing or coming to believe that P. This is what happens in examples such as Peacocke’s. In such cases the belief that P fails to stick despite the acknowledged presence of good reasons in favour of P, reasons which lead one to judge that 27 Sydney Shoemaker also makes this point. See Shoemaker (2003), 401. 28 Peacocke (1998), 90.

Knowing What You Believe

91

P.29 The reverse of this is also possible: the belief that P sticks or perseveres despite one’s unwillingness to judge that P in view of the weakness of the reasons in favour of P. Notice that such cases are also a problem for the linking assumption since they suggest a weaker link between what one judges and what one believes than that assumption implies. If I judge that P how can it be true that I nevertheless fail to believe that P? Suppose that P is the proposition that undergraduate degrees from countries other than my own are of an equal standard to my own. Should we not then insist, in response to Peacocke’s example, that I either fail to judge that P or that I do in fact believe that P, at least at the point at which I make the judgement? 30 To see why Peacocke’s example cannot plausibly be dealt with in this way it is important to be clear about the relationship between believing that P and concluding or judging that P in response to consideration of the reasons in favour of P. To judge that P is to affirm that P with the aim of getting the truth value of P right.31 Judging that P should lead one to believe that P but is not guaranteed to do so because belief can be influenced by evidentially irrelevant factors.32 In Peacocke’s example, the influence of prejudice prevents the judgement that undergraduate degrees from countries other than my own are of an equal standard to my own 29 As Shah and Velleman observe, “arriving at the judgement that P doesn’t necessarily settle the question whether one now believes it, since one may find oneself as yet unconvinced by one’s own judgement” (Shah/Velleman (2005), 507). 30 This will sometimes be the right response to putative sticking scenarios. Shah and Vellaman give the example of someone reasoning his way to the conclusion that his plane is not going to crash but still believing that it will. In this example it might be plausible to insist that the person has the obsessive thought that the plane will crash, or fears that it will crash, but does not literally believe that it will crash. Peacocke’s example is different. The thought here is that what I really believe comes out in what I do. Judging is a kind of doing. If I judge that all undergraduate degrees are equal that is evidence that I believe that all undergraduate degrees but this evidence is trumped by the fact that my letters of recommendation and hiring decisions only make sense on the assumption that I don’t believe that all undergraduate degrees are equal. Unlike the plane crash example, this is not a case of fear or obsession. 31 My account of judgement is essentially the one given in Shah/Velleman (2005). For further discussion of the relationship between judging, believing and thinking see Cassam (2010). 32 As Shah and Velleman point out. See Shah/Velleman (2005), 500. They include phobias and wishful thinking among the evidentially irrelevant factors that can influence belief.

92

Quassim Cassam

from leading, as it should, to formation of the belief that undergraduate degrees from countries other than my own are of an equal standard to my own. The possibility of belief and judgement coming apart in this way should come as no surprise given that judging and believing belong in different ontological categories. Judging is a mental action whereas belief is a mental state. The stative character of belief is marked by the impropriety of progressive tenses: “I am believing that P” is deviant in a way that “I am judging that P” is not.33 When I judge that P I do not occurrently believe that P because there is no such thing as occurrently believing.34 Belief is a dispositional state that is regulated for truth. Specifically, the belief that P “tends to be formed in response to evidence of P’s truth, to be reinforced by additional evidence of it, and to be extinguished by evidence against it”.35 What happens when I judge that P but don’t believe that P is that the act of judging that P does not result in the formation or acquisition of the appropriate dispositions. The point of judging is to make a mark on one’s beliefs but sometimes judging that P doesn’t make a mark on one’s beliefs; it doesn’t result in one’s believing that P. Just because I judge that P in response to what I recognize as good evidence for P that doesn’t guarantee that I will end up in a mental state that is regulated for truth in the way that belief is regulated for truth. To admit that belief can be influenced by evidentially irrelevant factors is to accept that beliefs are not necessarily responsive to reason. This is one respect in which beliefs, like passing thoughts, can be passive. Just as one can find oneself having the thought that P so one can find oneself believing that P. And just as one can distance oneself from one’s passing thoughts one can distance oneself from one’s beliefs.36 Martin gives a nice example of this. A father comes to realize, when engaged in second-order enquiry, that he has the conviction that his son is a great painter. Yet he may feel “forced to distance himself from what he recognizes is one of his own strongly held convictions” since “he believes 33 Williamson uses this argument to show that ‘know’ and ‘believe’ denote states rather than processes. See Williamson (2000), 35. 34 Tim Crane also makes this point. He points out that “occurrent belief” is a myth. See Crane (2001), 108. 35 Shah/Velleman (2005), 500. 36 So Hampshire is mistaken when he says that “a belief is a thought from which a man cannot dissociate himself” (Hampshire (1965), 98).

Knowing What You Believe

93

something on insufficient evidence, even by his own lights”.37 To distance oneself from the conviction that P is not necessarily to cease to believe that P. Distancing might lead to the abandonment of the belief but then again it might not. Some beliefs may be able to survive the recognition that they are not rationally grounded.38 All of this strengthens the case for thinking that the transparency procedure can only deliver self-knowledge that is not immediate. To see how, consider the following response to examples such as Peacocke’s: what such examples show is that a person’s belief concerning P is not invariably determined by his consideration of the reasons for or against believing that P. They do not show that they are not normally so determined. For part of what it is to be a rational agent “is to be able to subject one’s attitudes to review in a way that makes a difference to what one’s attitude is”.39 The goal of deliberation is conviction, and it would not be possible to regard a person as deliberating if he does not presume, at least implicitly, that when he reasons his way to the conclusion that P he also ends up believing that P as a result. The fact that this is a defeasible presumption does not make it any less dispensable. This strengthens the case for thinking that the transparency procedure can only deliver self-knowledge that is not immediate because it implies that that a person’s judging that P is no more than defeasible evidence that he believes that P. For even if my concluding or judging that P in response to consideration of the reasons in favour of P is presumed to result in my believing that P, the fact remains that my judging that P does not entail that I believe that P. What it does is to raise the probability that I believe that P. My judging that P makes it likely, given that I am rational, that I believe that P and is, in this sense, a reliable indicator that I believe that P. But this is just what it is for one thing to be evidence for another.40 Furthermore, it is not just that my judging that P is evidence that I believe that P. It is also evidence I have, to the extent that 37 38 39 40

Martin (1998), 115. Perhaps religious beliefs are a case in point. Moran (2001), 64. As Williamson remarks, what is required for e to be evidence for the hypothesis h is that “e should speak in favour of h” and should itself have “some kind of creditable standing” (Williamson (2000), 186). In probabilistic terms, e speaks in favour of h if it raises the probability of h. Kelly points out that “the notion of evidence is that of something which serves as a reliable sign, symptom, or mark of that which it is evidence of” (Kelly (2006)).

94

Quassim Cassam

I am aware of judging that P and understand that if I judge that P then it is highly likely that I believe that P. Finally, my knowledge that I believe that P is based on evidence in my possession if I know I believe that P because I know that I judge that P. But if I know that I believe that P on the basis of evidence then my knowledge is, by definition, not epistemically immediate.

IV The story so far is that the transparency procedure fails to secure either the epistemic or the psychological immediacy of self-knowledge. So what? That depends on whether the immediacy thesis is correct. If it is a given that self-knowledge is immediate, and the aim is to account for its immediacy, then the transparency account fails. But should we endorse the immediacy thesis? It looks like a datum that knowledge of our own beliefs is, by and large, psychologically immediate so it is certainly an objection to the transparency approach that it fails to account for, or even accommodate, this datum. But epistemic immediacy is another matter. It is not obvious that knowledge of our own beliefs is typically immediate in this sense, and it is only taken as obvious because epistemic immediacy is confused with psychological immediacy. If one is sceptical about the assumption that knowledge of our own beliefs is normally epistemically immediate then it is not necessarily an objection to a theory of self-knowledge that it implies that we do not have epistemically immediate access to our own beliefs. The only genuine epistemological datum in this area is that self-knowledge is not normally based on behavioural evidence, and this is something that the transparency account certainly does account for.41 My judging that P may be evidence that I believe that P but it is not behavioural evidence. By the same token, if I know that I believe that P on the basis of my awareness of judging or mentally affirming that P this is not self-knowledge based on behavioural evidence. All of this points to the need for a distinction between different kinds of epistemic immediacy and correspondingly different versions of the immediacy thesis. My knowledge that I believe that P is weakly 41 Moran sometimes gives the impression that all he means when he says that selfknowledge is immediate is that it is not based on behavioural evidence. But if self-knowledge is based on evidence other than behavioural evidence then it is certainly not “radically nonevidential” (Moran (2001), 68).

Knowing What You Believe

95

epistemically immediate just if it is not based on behavioural evidence. It is strongly epistemically immediate just if it is based on no evidence. The sense that some philosophers have that self-knowledge is epistemologically distinctive is the sense that it is strongly epistemically immediate, and this is what gives rise to the idea that a good theory of selfknowledge should be able to explain how strongly epistemically immediate self-knowledge is possible. Viewed in this light, there are two natural responses to the fact that the transparency approach doesn’t succeed in this task. One would be to look for an alternative account of selfknowledge which does explain the strong epistemic immediacy of some self-knowledge. The other would be to question the assumption that knowledge of our own beliefs is immediate in this sense. The suggestion is not just that it is not obviously correct that knowledge of our own beliefs is not strongly epistemically immediate but that reflection on the nature of belief should lead one to conclude that this kind of self-knowledge cannot be immediate in this sense. The most promising version of the first of these responses is a Monitoring Mechanism (MM) approach to self-knowledge. The idea is this: when I come to believe that P what happens is that the representation that P enters my Belief Box. The question “Do I believe that P?” calls for a search of my Belief Box. This search is not carried out by me, the subject, but by one of my sub-personal monitoring mechanisms, that is, “a distinct mechanism that is specialized for detecting one’s own mental states”.42 If the belief that P is found in my Belief Box this leads to the formation of the second-order belief that I believe that P, and this second-order belief constitutes knowledge as long as the monitoring mechanism is reliable and produces beliefs about my beliefs which could not easily be false. This knowledge is both psychologically and epistemically immediate. It is not produced by conscious reasoning and is not based on behavioural or any other evidence.43 42 Nichols/Stich (2003), 163. 43 As Stich and Nichols point out, a good theory of self-awareness needs to be able to explain the fact that “when normal adults believe that P, they can quickly and accurately form the belief I believe that P” (Nichols/Stich (2003), 160). In order to implement this ability, “all that is required is that there be a Monitoring Mechanism (MM) that, when activated, takes the representation P in the Belief Box as input and the representation I believe that P as output” (Nichols/ Stich (2003), 160 f.). The Monitoring Mechanism simply has to copy representations from the Belief Box and embed copies of them in a schema of the form ‘I believe that…’. Stich and Nichols do not draw attention to the consequences

96

Quassim Cassam

This account of self-knowledge has its attractions but presupposes an externalism about knowledge which some may find hard to swallow. Others may have difficulty with the idea that beliefs are the kind of thing that can be stored and monitored in the way the MM theory implies. While these and other problems with the MM theory might not be insuperable they do raise fundamental questions about the nature of belief and the assumption that we can have epistemically immediate knowledge of our own beliefs. Suppose we think of belief as a form of acceptance, so that to believe that P is to accept that P or, in other words, to regard P as true. The challenge is to say what distinguishes believing from other modes of acceptance such as supposing and assuming, and it is in the course of responding to this challenge that the limitations of the immediacy thesis become apparent. For when we think about the various respects in which belief is different from other modes of acceptance it becomes hard to see how knowledge of one’s own beliefs could be strongly epistemically immediate. Suppose that what distinguishes belief from other modes of acceptance is that there is a distinctive way in which beliefs are regulated, that is, formed, revised and extinguished. Belief is regulated for truth in a way that other modes of acceptance are not, and being regulated for truth is a broadly dispositional property of beliefs. Whether a given mental state is the state of believing that P is therefore partly a matter of what dispositions the state has, and one epistemological consequence of this is that one is not always in a position to know whether one believes that P since one is not always in a position to know that one is in a mental state with the relevant dispositions.44 It is also unclear, on the

of their view for the issue of immediacy, but it seems obvious that if a Monitoring Mechanism is sufficiently reliable to produce knowledge of one’s own beliefs then the knowledge to which it gives rise is both psychologically and epistemically immediate. It was Timothy Williamson who first drew my attention to the possibility of exploiting something like the MM theory to explain the immediacy of self-knowledge. He does not, however, endorse the present approach to self-knowledge. 44 In Williamson’s terminology (which is different from Moran’s) ‘transparency’ is the thesis that “for every mental state S, whenever one is suitably alert and conceptually sophisticated, one is in a position to know whether one is in S” (Williamson (2000), 24). He goes on to argue that transparency fails for the state of believing since “the difference between believing P and merely fancying P depends in part on one’s dispositions to practical reasoning and action manifested only in counterfactual circumstances” (Williamson (2000), 24). In effect, Wil-

Knowing What You Believe

97

present account, how one’s knowledge that one believes that P could be immediate. I only believe that P if I am in a mental state which is regulated for truth but how can I know, other than on the basis of evidence, that I am in such a state? How, for example, can I know without evidence, not only that I accept that P but also that my acceptance of P is such that it would be extinguished by evidence against the truth of P? In fact, matters are even more complicated than this. Even if I find myself continuing to accept that P in the face of what I recognize as evidence against P it still does not follow that I don’t really believe that P. Genuine beliefs can be influenced by non-rational factors, and this makes them even harder to detect. The true epistemological significance of the distinction between occurrent states and standing attitudes is now apparent. The emerging picture is that occurrent mental states and mental actions are on the surface of the mind, and that is why they can be known immediately by their subject. I do not need to dig deep in order to know what I am thinking or judging or feeling at the time when I am thinking or judging or feeling. But standing attitudes are not surface phenomena. Beliefs, for example, are psychological states that in some sense underlie one’s occurrent mental states. Judging that P might manifest the belief that P but it is natural to think that the belief is not as directly accessible as its manifestations. While the distinction between what is on the surface of our minds and what lies beneath is obviously metaphorical it is an important element of the naive conception of the mental. Perhaps, on reflection, even what is below the surface is immediately knowable, but it is not obvious that the presumption is one to which we are committed. Once we have dispensed with implausible, philosophically inspired versions of the immediacy thesis, and also taken on board the extent to which belief-formation is not a rational process, we can start to think realistically about how we are able to know our own beliefs.45 liamson’s point is that the dispositional dimension of believing makes trouble for what he calls transparency. My point is that it makes trouble for immediacy. 45 I gave earlier versions of this paper at the University of Bonn, the University of Chicago and at the 2010 Oslo conference on Self-Knowledge and Rational Agency. I thank Frank Barel, who was my commentator in Oslo. For many other helpful comments and questions I thank Elke Brendel, Jason Bridges, James Conant, Fred Dretske, Ciara Fairley, David Finkelstein, Adrian Haddock, Jonathan Lear, Anna-Sara Malmgren, Conor McHugh, Richard Moran, Anders Nes, Julia Peters, Robert Stalnaker, Josef Stern and Crispin Wright.

98

Quassim Cassam

References Bar-On (2004): Dorit Bar-On, Speaking My Mind: Expression and Self-Knowledge, Oxford. Boghossian (1998): Paul Boghossian, “Content and Self-Knowledge”, in: P. Ludlow/N. Martin (eds.), Externalism and Self-Knowledge, Stanford, 149 – 173. Bollas (2009): Christopher Bollas, The Evocative Object World, London. Burge (1998): Tyler Burge, “Individualism and Self-Knowledge”, in: P. Ludlow/N. Martin (eds.), Externalism and Self-Knowledge, Stanford, 111 – 127. Byrne (2011): Alex Byrne, “Knowing That I am Thinking”, in: A. Hatzimoysis (ed.), Self-Knowledge, Oxford, 105 – 124. Carruthers (2005): Peter Carruthers, “Conscious Thinking: Language or Elimination?”, in: P. Carruthers, Consciousness: Essays from a Higher-Order Perspective, Oxford, 115 – 133. Cassam (2010): Quassim Cassam, “Judging, Believing and Thinking”, in: Philosophical Issues 20, 80 – 95. Crane (2001): Tim Crane, Elements of Mind: An Introduction to the Philosophy of Mind, Oxford. Davidson (1994): Donald Davidson, “Knowing One’s Own Mind”, in: Q. Cassam (ed.), Self-Knowledge, Oxford, 43 – 64. Frankfurt (1998): Harry Frankfurt, “Identification and Externality”, in: H. Frankfurt, The Importance of What We Care About, Cambridge, 58 – 68. Hampshire (1965): Stuart Hampshire, Freedom of the Individual, London. Kelly (2006): Thomas Kelly, “Evidence”, in: E. N. Zalta (ed.), The Stanford Encyclopedia of Philosophy, URL: http://plato.stanford.edu/entries/evidence. Martin (1998): Michael Martin, “An Eye Directed Outward”, in: C. Wright/ B. Smith/C. Macdonald (eds.), Knowing Our Own Minds, Oxford, 99 – 121. Moran (2001): Richard Moran, Authority and Estrangement: An Essay on SelfKnowledge, Princeton/Oxford. Moran (2003): Richard Moran, “Responses to O’Brien and Shoemaker”, in: European Journal of Philosophy 11, 402 – 419. Moran (2004): Richard Moran, “Replies to Heal, Reginster, Wilson, and Lear”, in: Philosophy and Phenomenological Research 69, 455 – 472. Nichols/Stich (2003): Shaun Nichols/Steven Stich, Mindreading: An Integrated Account of Pretence, Self-Awareness, and Understanding of Other Minds, Oxford. Peacocke (1998): Christopher Peacocke, “Conscious Attitudes, Attention, and Self-Knowledge”, in: C. Wright/B. Smith/C. Macdonald (eds.), Knowing Our Own Minds, Oxford, 63 – 98. Pryor (2005): James Pryor, “There is Immediate Justification”, in: M. Steup/E. Sosa (eds.), Contemporary Debates in Epistemology, Oxford, 181 – 202. Ryle (1949): Gilbert Ryle, The Concept of Mind, London. Sellars (1975): Wilfrid Sellars, “The Structure of Knowledge”, in: H. Castañeda (ed.), Action, Knowledge and Reality: Studies in Honor of Wilfrid Sellars, Indiana, 295 – 347. Shah/Velleman (2005): Nishi Shah/David Velleman, “Doxastic Deliberation”, in: The Philosophical Review 114, 497 – 534.

Knowing What You Believe

99

Shoemaker (2003): Sydney Shoemaker, “Moran on Self-Knowledge”, in: European Journal of Philosophy 11, 391 – 401. Williamson (2000): Timothy Williamson, Knowledge and its Limits, Oxford.

From Transparency to Expressivism1 David H. Finkelstein 1. At least since John Locke defined “REFLECTION”, as the “Perception of the Operations of our own Minds within us”,2 one influential approach toward understanding the knowledge we have of, and the authority with which we speak about, our own psychological states and events has been what I call “detectivism”. A detectivist is someone who would claim that we know our own mental goings-on thanks to a process by which we perceive or, anyway, detect them and thereby render them conscious. So characterized, “detectivism” names a broad family of positions. While some detectivists claim that self-knowledge comes via an “inner sense” that provides each of us with infallible epistemic access to a peculiarly private, perhaps immaterial, mental realm, others (of a more naturalistic bent) posit a perceptual process that is epistemically on all fours with seeing and hearing, except that it happens to be directed inward, toward the mind/brain rather than outward toward the external world.3 Detectivism as such is controversial; many philosophers and psychologists are committed to one or another version of it, while many others reject it. In the present essay, I won’t be engaging directly with arguments either for or against detectivism. Instead I want to consider a way of answering the following question: If we reject detecti1

2 3

Versions of this paper were presented at the University of Chicago’s Wittgenstein Workshop in March 2010; at a conference called Self and Others in Wittgenstein and Contemporary Analytic Philosophy at the University of Southampton in March 2010; at the Inter-University Workshop on Expression and the Inner, held in Oviedo, Spain, in April 2010; at a conference called Aspects of Self-Knowledge at LMU München in December 2010; at Auburn University’s Department of Philosophy in February 2011; and at Johns Hopkins University’s Department of Philosophy in September 2011. For especially helpful comments, questions, or conversation, I’m grateful to Anita Avramides, Stina Bäckström, Matthew Boyle, Jason Bridges, James Conant, Adrian Haddock, Arata Hamawaki, Jane Heal, Irad Kimhi, Thomas Lockhart, Robert Pippin, William Small, and Barry Stroud. Locke (1689/1975), II, I, 4. For a selective history of detectivism, see ch. 1 of Finkelstein (2003).

102

David H. Finkelstein

vism, how are we supposed to make sense of the fact that people are able to accurately and easily self-ascribe beliefs, desires, fears, hopes, and the like? How can we give up detectivism without this capacity just coming to seem miraculous? The answer that I’ll be discussing in the bulk of what follows is developed in the work of Gareth Evans, Sydney Shoemaker, and Richard Moran. According to (what we might call) the transparency approach toward understanding self-knowledge and firstperson authority, we are able to speak about our own states of mind by addressing questions about the world outside us. Where this is possible, we don’t need to rely on any sort of inward-directed perception or detection in order to say what we ourselves want, fear, hope, or believe. This is not the first time that I have written about the transparency approach. In a postscript to Finkelstein (2003), I criticized it, focusing especially on the innovative version that Moran had defended in his 2001 book, Authority and Estrangement. In a paper published in 2009, Matthew Boyle offered (what struck me as) a sensible reply to my criticism of Moran. When I read Boyle’s paper, I decided that I needed to consider again the question of precisely what, if anything, I take to be wrong with Moran’s position—along with the transparency approach, more generally. The present essay is a result of that reconsideration. 2. Evans writes: [I]n making a self-ascription of belief, one’s eyes are, so to speak, or occasionally literally, directed outward—upon the world. If someone asks me ‘Do you think there is going to be a third world war?’, I must attend, in answering him, to precisely the same outward phenomena as I would attend to if I were answering the question ‘Will there be a third world war?’ I get myself into a position to answer the question whether I believe that p by putting into operation whatever procedures I have for answering the question whether p. 4

We can capsulize Evans’s point in the following claim, which I’ll refer to in what follows as “TC” (for “transparency claim”): TC: The question of whether I believe that p is, for me, transparent to5 the question of whether p is true. This is to say: I can answer the former question by answering the latter.

4 5

Evans (1982), 225. In this use of the phrase “transparent to”, I’m following Moran (2001).

From Transparency to Expressivism

103

Suppose that TC strikes us not only as true, but as showing that we need not posit an inner sense in order to make sense of self-ascriptions of belief. Suppose, moreover, that we would like to extend the thought expressed in TC beyond belief (as it were) and to thereby explain the authority with which we speak about our own desires, fears, hopes, and other attitudes. It’s far from obvious how such an extension might be effected. Shoemaker tries and, it seems to me, fails with various proposals. For example, he writes: The rational agent who wants X and has normal mastery of language will, ceteris paribus, respond affirmatively to the question “Shall I give you X?” And given her mastery of the concept of desire, she will respond affirmatively to the question “Do you want X?” if she will respond affirmatively to the question “Shall I give you X?” So she will, unless she has devious motives, give correct answers to questions about what she wants.6

Here, Shoemaker is suggesting, in effect, that if you were to ask me a question of the form “Do you want X?”, I could answer by acting as if you had asked me a corresponding question of the form “Shall I give you X?”. But this suggestion seems blatantly unsatisfactory. I might want anything—a sandwich, a summer house, a fulfilling career, or an ability to yodel—without wanting you (or perhaps anyone) to give it to me. As I read Moran, he offers another way to extend the kind of explanation of first-person authority that is suggested by TC (to extend it, that is, to attitudes other than belief). And it seems to me that what he provides along these lines is, prima facie, a good deal more appealing than what Shoemaker offers. Moran manages this by shifting his starting point from TC to what I’ll call TC*: TC*: The question of whether I believe that p is, for me, transparent to the question of what I ought rationally to believe—i. e., to the question of whether the reasons require me to believe that p. I can answer the former question by answering the latter.

The difference between TC and TC* lies in TC*’s appeal to reasons.7 TC* says, in effect, that if I am asked whether I believe it’s about to 6 7

Shoemaker (1994), 283. Moran (2003), 405, writes: “[T]here’s something a little misleading in Evans’s original formulation. He speaks of one’s eyes ‘directed outward, upon the world’, and contrasts this outward gaze with the discredited inward one. But it is rarely by simply looking at something that we answer such world-directed questions for ourselves, and when looking is involved it is because in this con-

104

David H. Finkelstein

rain, I can address this question by answering the question “Do the reasons require me to believe that it’s about to rain?”. This shift in starting points has the effect of making TC* straightforwardly extendable to other attitudes. According to Moran, the question of, e. g., whether I want X is transparent to the question “Do the reasons require me to want X?”; I can say what I want by consideration of what I ought to want in light of the appropriate reasons. Or, rather, I can do this on the condition that what I in fact want is, after all, in accord with my assessment of what I ought to want. (Moran calls this the “Transparency Condition”.) When, and only when, this condition is in place, I can say what I want by considering what I ought to want, what I fear by considering what I ought to fear, what I intend by considering what I ought to intend, etc. Moran uses the word “avowal” only where a person’s answer to a question about some attitude of hers is dictated by her assessment of what is rationally required of her. Avowal, in this sense, is for Moran the essence of first-person awareness. He understands “the ability to avow one’s belief as the fundamental form of self-knowledge”.8 He holds that a person has “genuine first-person awareness”9 only when she can avow (in his sense) her attitude—only when she can speak about it by addressing the outward-directed question of what the appropriate reasons require of her.10 In the postscript to my book, I argued text we are counting what we see to provide some reason for or against some possible belief. The more general and central truth in this context is that I answer the external question about the weather or the possibility of war by putting myself in a position to confront and assess the reasons relevant to the truth about the weather or the possibility of war.” 8 Moran (2001), 150. 9 Moran (2001), 107. 10 Boyle (following Moran) sometime speaks as if Moran’s central idea is that I know my beliefs, intentions, fears, etc., by making up my mind, or by deliberating. Thus Boyle writes: “What Moran finds striking about our knowledge of our own attitudes is this: we often seem to be able to know whether we hold them by deliberating about the topics they concern” (Boyle (2009), 136, his emphasis). This way of putting the point is, I think, liable to mislead. Here’s what’s misleading: I often say what my attitude is—e. g., what I believe—when I’m not making my mind, when my mind is already made up, and I don’t now need to deliberate. (If I were asked who I thought the president of the United States was, I could answer without needing to deliberate or make up my mind.) One might conclude that Moran cannot accommodate self-knowledge in such cases. But I don’t think this is a compelling objection to Moran’s view, rightly understood. For his view could be put this way: regardless of whether I’m mak-

From Transparency to Expressivism

105

that this claim about first-person awareness should be rejected. I pointed out that there is a wide range of cases in which a person speaks with first-person authority about her state of mind even though she cannot “avow” it in Moran’s sense. Here is one kind of case: Max has a spider phobia. He is terrified of spiders—spiders in general, and the one on the pillow beside him in particular. Max doesn’t take this fear to be rational; he knows that he suffers from a phobia. Still, when he says, “Get that thing away from me; I’m really afraid of it!”, he is speaking (or shouting) about his own state of mind with first-person authority. In that example, Max’s fear is, by his own lights, irrational. Here is another kind of case, one in which the attitude at issue is not irrational: I sometimes look after my friend Adam’s dog, Sadie, when Adam is out of town. I don’t view it as rationally incumbent upon me to be fond of Sadie. I cannot answer the question of whether I am fond of Sadie by addressing a question about whether I ought rationally to be. Nonetheless, I’m fond of Sadie, and I have no trouble speaking with first-person authority about my fondness for her. In the postscript to my book, I argued that once it becomes apparent that one may speak with first-person authority regardless of whether one is talking about a belief that is dictated by the conclusion of theoretical reflection, an irrational fear, or an attitude that is neither dictated by reasons nor irrational, it is hard to go on being convinced by Moran’s account of first-person authority. 3. Boyle allows that Moran can seem to have overstated his point: There seem plainly to be kinds of mental states of which our knowledge is both non-observational and non-deliberative: not just sensations but, for instance, appetites (i. e., brute, unreasoned desires for things of a certain kind) and what might be called “recalcitrant attitudes” (e. g., feelings of anger that I know to be unjustified but cannot overcome.) … [I]f his [Moran’s] account does not apply to such knowledge, it is not clear how he can justify his claim to have described the fundamental form of selfknowledge, the one that “makes the difference” between first-person awareness and the kind of awareness we might have of the mental states of another person.11 ing up my mind or merely maintaining it (rationally), I answer the question of what my attitude is according to my current assessment of what it ought rationally to be. When I say who I believe the president is, my answer is beholden to the outward-directed reasons, even if I don’t have to deliberate about them or “make up my mind” in order to answer. 11 Boyle (2009), 138 f.

106

David H. Finkelstein

Boyle aims to show that, even so, there is an important sense in which Moran has explained the fundamental form of self-knowledge. The important point, according to Boyle, is not that Moran has managed to explain the only kind of first-person authority that there is. It is, rather, that he’s accounted for the fundamental kind, where this is understood as the kind that the other kind(s) presuppose. According to Boyle, the real value of Moran’s book lies not in extending the transparency point far beyond belief, or beyond belief at all. It lies in allowing us to see something essential to our agency—in helping us to understand a kind of authoritative self-awareness that is tied up with our nature as reflective agents. Boyle accuses me (and others) of simply assuming that first-person authority is a unitary phenomenon—of presupposing that “a satisfactory account of our self-knowledge should be fundamentally uniform, explaining all cases of ‘first-person authority’ in the same basic way”.12 He asks why there shouldn’t be a kind of first-person authority that is in place only when the subject’s attitude is determined by her assessment of what the appropriate reasons require. Why, moreover, shouldn’t we think that Moran has provided the right explanation of this kind of firstperson authority? This, I think, is a good question—a reasonable response to the objection that I raised in the postscript to my book.13 To address it, we need to ask whether Moran offers a satisfactory explanation of self-knowledge in the limited range of cases where a person’s attitude is determined by her assessment of what the reasons require her to believe, desire, or fear. According to Moran, when I come to the conclusion that, e. g., fear is rationally required of me, I am able to, as it were, read it off this conclusion that I am afraid. Putting his point this way suggests the following question: What entitles me to hold, on any particular occasion, that my actual attitude toward some object or proposition is the one that, according to me, is rationally required? After all (as we have seen), I don’t always have the attitude that I think is rationally required of me. Imagine that I think through my current situation and conclude that I ought to be afraid. Am I afraid? If I simply avow the conclusion of my rational deliberation as my state of mind, is my avowal justified? 12 Boyle (2009), 141. 13 Boyle goes on to argue that first-person authority cannot be understood as a unitary phenomenon with a single explanation. I take him to be wrong about this, but I won’t be directly engaging with his argument here.

From Transparency to Expressivism

107

Moran answers (what amounts to) this question in a paper published two years after his book. There he writes: [O]ne of the challenges to the Transparency claim can be put in the following way: What right have I to think that my reflection on the reasons in favor of P (which is one subject-matter) has anything to do with the question of what my actual belief about P is (which is quite another subject matter)? Without a reply to this challenge, I don’t have any right to answer the question that asks what my belief is by reflection on the reasons in favor of an answer concerning the state of the weather. And then my thought at this point is: I would have a right to assume that my reflection on the reasons in favor of rain provided me with an answer to the question of what my belief about the rain is, if I could assume that what my belief here is was something determined by the conclusion of my reflection on those reasons. An assumption of this sort would provide the right sort of link between the two questions.14

Having come to the conclusion that my attitude ought rationally to be v, what right do I have to claim that v is, in fact, my attitude? Moran’s answer is that I am entitled to assume that this is so; I’m entitled to assume that the attitude I in fact have is the one that, by my lights, the reasons call for me to have. In what follows, I shall refer to such an assumption as an RA (for “rationality assumption”). And in what remains of §3, I mean to question the way in which RAs figure in Moran’s picture of self-awareness. Before proceeding, I should perhaps say something about what I don’t mean to be questioning in this area of Moran’s thought. Here, I’d like you to consider the way in which the passage from Moran (2003) that I quoted in the preceding paragraph continues: 14 Moran (2003), 405. Boyle quotes this passage from Moran and glosses it as follows: “If a subject who possessed the concept of belief were entitled to assume that, in reaching the conclusion that p is true, he was coming to believe that p, then it seems that he could justifiably answer the question whether he believed that p by reflecting on ground for taking p to be true” (Boyle (2009), 137 – 8). In a footnote to this gloss, Boyle makes explicit that such assumptions won’t always turn out to be true: “This is not to suggest that my merely taking myself to have a certain belief must make it so. As Moran emphasizes, his view does not demand that a subject be incorrigible about her own attitudes, or that her claims about her own attitudes have special authority no matter what their basis. What is important is that the question of what attitude I hold can often enough be settled by me on the basis of deliberation about whether p, and that—according to Moran—it is fundamental to the very possibility of thought about such attitudes that this should be so” (Boyle (2009), 137).

108

David H. Finkelstein

And now let’s ask, don’t I make just this assumption, whenever I’m in the process thinking my way to a conclusion about some matter? I don’t normally think that my assessment of the reasons in favor of P might have nothing to do with what my actual belief about P is, and it’s hard to imagine what my thinking would be like if I did normally take this to be an open question. And if I did think that my actual belief about the rain might be left quite untouched by my reflections on the weather-related reasons, what do I imagine could possibly close this gap for me? 15

I don’t want to question the claim that, insofar as I take myself to be deliberating, I must assume—as I am “in the process of thinking my way to a conclusion”—that my assessments of the reasons for and against, e. g., believing that it’s about to rain (or wanting to travel in China, or fearing that I’ll lose my job) bear on what my attitude will turn out to be. (Perhaps it’s a transcendental condition on my being a rational agent that I make this assumption.) But Moran is saying more than this. Again, he is claiming that I can and do self-ascribe attitudes by: (1) reasoning about what my attitude ought to be and (2) avowing the conclusion of this reasoning as what my attitude is. And as we’ve seen, a worry arises at this point: Sometimes my actual attitude doesn’t jibe with the conclusion of my reasoning; what gives me the right to make any claim about what my attitude is, if I consider only what it ought to be? What entitles me to answer a question about my actual attitude by addressing a question about what my attitude should be? Moran invokes RAs at this point, to address this worry. When I say what my attitude is, I am assuming—and I’m entitled to assume—that this is not one of those cases in which there is, as it were, a gap between my assessment of what attitude is called for and my actual attitude. Now, there are various concerns one might have about this position. My chief concern is this: It seems to me that insofar as someone needs to make this sort of assumption in order to self-ascribe an attitude, one is alienated from the attitude. This is connected to the fact that we don’t often find ourselves saying such things as: “Given that my current desire is the one that I take to be most reasonable, I don’t want to order any desert”. The problem here is not merely that this would sound odd if I said it to a waitress; it’s that such an utterance would mischaracterize the situation that I’m in (at least ordinarily) when I say what I want. There are situations in which someone assumes he has an attitude that jibes with the reasons as they strike him. Consider the following: 15 Moran (2003), 405 f.

From Transparency to Expressivism

109

Henry is a Navy SEAL, about to embark on a hazardous mission. His wife, who has gathered that he’s being sent off to do something dangerous, asks him, “Are you afraid?”. He answers, in a flat tone of voice, “No”. She says: “Come on, Henry; you can tell me what you’re feeling. We’ve talked about how dangerous your work is. Surely, you feel some fear”. Henry thinks about what he’s liable to be doing for the next few days and says, “Well, I see that there are lots of good reasons for me to be afraid, so I suppose that at some level, I must be”.

Suppose that Henry is not lying to his wife in order to appear brave; he is trying, anyway, to speak honestly. Suppose, moreover, that Henry’s self-ascription is correct; he is afraid at some level. Then, this is a case in which someone is able to correctly ascribe an attitude to himself only thanks to an RA. Insofar as this is Henry’s situation, he is alienated from his own state of mind.16 This is connected to the fact that when Henry finally ascribes fear to himself, he does not speak with (much17) first-person authority. Contrast Henry’s case with this one: It’s 7 pm on a Monday. I’m at my office, working, when a friend phones to see if I want to go with her to a movie. I need only engage in a moment or two of practical reasoning—I think about the class that I have to teach in the morning, the letter of recommendation that’s two days overdue, and the dissertation that I still haven’t finished reading—whereupon I say to my friend, “I’m sorry; I’m sure I’d enjoy the movie, but I want to stay at the office this evening”.

This is the kind of case that Moran’s account should be best suited to explaining, one in which a subject rationally deliberates about what he wants to do, arrives at a single conclusion, and avows it. But now: at the point in the story when I tell my friend that I want to stay at the office, am I assuming that what I, in fact, want is in line with my assessment of what I ought to want? Surely not. At that moment, I know what I, in fact, want; my desire is (unlike Henry’s fear in the last example) fully conscious, and I have no need of any such assumption. Indeed, it would be more plausible to claim that, in the case described, I know that my desire accords with my reasoning in part because I know that 16 Indeed, to the extent that this is his situation, the attitude that Henry self-ascribes is—to that extent—unconscious. (I argue that consciousness and first-person authority should be understood to come in degrees in Finkelstein (2003), §5.5.) 17 See the preceding note.

110

David H. Finkelstein

what I want is to stay at the office, than to claim, with Moran, that I can say what I want thanks to my assuming I have a desire that is dictated by the conclusion of my reasoning. The same point obtains if we consider a case in which a belief, rather than a desire or fear, is self-ascribed: After painstakingly going over the evidence, Jill comes to the conclusion that her business partner, whom she took to be a friend, has been embezzling money from their shared company. She says, “I believe that Joel has been embezzling money from the company”.

Unless we somehow flesh out the story in such a way as to make it plausible that Jill is talking about a belief of hers from which she’s alienated, it’s not going to make sense to claim that her self-ascription depends upon her assuming that she has an attitude that accords with the conclusion of her theoretical reasoning. If the belief that she’s self-ascribing is an ordinary conscious one, then she has no need for such an assumption. (Here too, it would be more plausible to claim that she knows her attitude accords with her reasoning because she knows what she believes about Joel than it is to claim that she can say what she believes about Joel only because she is assuming that she has an attitude that jibes with her theoretical reasoning.) Whether we are talking about fears, desires, or beliefs, if a person needs an RA in order to avow her attitude, then she is alienated from it; she lacks “genuine first-person awareness” of it. And what Moran means to be offering is precisely an account of unalienated self-awareness.18 I’ll sum up. Moran (2001) says that we can answer questions about our own attitudes by addressing outward-directed questions about what ought rationally to be believed, desired, or feared, given our situation. Moran (2003) adds that in so answering a question about his own attitude, a subject must assume that it jibes with his assessment of what his attitude ought rationally to be. I have been arguing that self-awareness

18 A similar objection might be raised in connection with detectivism. We do sometimes detect our own attitudes and emotions via self-observation (or inference or testimony). What the detectivist misses is that insofar as one must selfascribe a mental state on this sort of basis, one is alienated from it. Detectivism goes wrong by representing the best cases of self-knowledge as if they were cases of alienated self-knowledge. I’m claiming that Moran winds up doing this as well.

From Transparency to Expressivism

111

does not ordinarily require such an assumption and that where it does, the subject is alienated from his own state of mind.19 4. The argument against Moran’s position set out in the preceding section does not tell against Evans’s original proposal for how we answer questions about our own beliefs. Moran is driven to represent avowals as depending on RAs because he shifts his starting point from (what I called in §2) TC to TC*. This allows him to extend an account of self-knowledge and first-person authority beyond belief—to desire, fear, intention, etc.—but it comes at the cost of claiming that, even in the best kind of case, a subject can only assume that the attitude he in fact has is one that accords with his assessment of what his attitude ought rationally to be. TC requires no such assumption; it makes no reference to reasons or rationality. According to Evans, I can answer a question about what I believe by addressing a corresponding question about what is true. I needn’t address the question about what’s true in a way that is, even by my own lights, rational. So I am able, e. g., to answer the question of whether I believe that God created the Earth by addressing the question “Did God create the Earth?”—and I can do this regardless of whether I answer the latter question on the basis of what I take to be good reasons. Let’s say that you are convinced by §3’s objection to Moran’s position. So you are not going to accept that (unalienated) self-knowledge and first-person authority depend upon RAs. Let’s say, as well, that you are moved by Boyle’s suggestion that we shouldn’t just assume that 19 It might be possible to work out a variation on Moran’s position that doesn’t require the problematic appeal to RAs—by embracing a kind of disjunctivism about self-knowledge. In what we can think of as “the good sort of case”, when deliberation results in the appropriate rational attitude, there is no gap between the outward-directed question “Ought I to be v ?” and the inward-directed “Am I v ?”. Thus, in a good case, the subject is entitled—without an RA—to answer, e. g., the question, “Am I afraid?” by addressing the question “Ought I to be afraid?”. Here, there is no need for a rationality assumption to bridge a gap between the two questions because in a good case there is no gap. If this sort of position could be made out, then it would be immune to the objection I’ve raised in this section. Although I won’t work through the details, the right response to such a position, would, I believe, be a variation on what I’ll say about Evans in §4. Evans’s position is (already) like a disjunctivist version of Moran’s in that there is no gap between the outward-directed question (“Is p true?”) and the inward-directed question (“Do I believe that p?”).

112

David H. Finkelstein

there is only one kind of first-person authority or (unalienated) selfknowledge. At this point, you might be tempted to draw a conclusion that could be put as follows: “Evans was right about the authority with which we speak about our own beliefs—but this authority turns out to be sui generis. We shouldn’t try to explain our knowledge of our own desires, fears, wishes, etc. in the same way that we explain our knowledge of our own beliefs. We don’t have the same kind of authority about our own desires, fears, and wishes as we do about our beliefs.” In what remains of this essay, I’ll argue that this is not the conclusion you ought to draw. I’ll suggest that, rightly understood, TC leads—not to a picture of first-person authority as a divided phenomenon, but—to a unitary expressivist20 account of the authority with which we speak about our own beliefs, desires, fears, moods, and sensations. Just after the passage from Varieties of Reference that I quoted at the start of §2, Evans writes: “We can encapsulate this procedure for answering questions about what one believes in the following simple rule: whenever you are in a position to assert that p, you are ipso facto in a position to assert ‘I believe that p’”.21 How is it that this “simple rule” is supposed to show that we needn’t posit any sort of inward-directed observation or detection in order to make sense of belief self-ascriptions? An answer to this question could, I take it, be put as follows: We feel no pressure to posit anything like an inner sense in order to understand how it’s possible for someone to address an outward-directed question, e. g., a question like “Is it about to rain?”. Since one need do nothing more in order to answer a question like “Do you believe that it’s about to rain?”, we should feel no pressure to posit an inner sense in order to make sense of self-ascriptions of belief. Now, I’d

20 In earlier writings, I reserved the word “expressivist” for views according to which psychological self-ascriptions express a speaker’s state of mind but are not themselves truth-apt. The position I argued for—according to which a speaker can express his state of mind by asserting truly that he is in it—did not, therefore, count as expressivist. But people were always inclined to describe me as a kind of expressivist, and since Bar-On (2004) distinguished between “simple expressivism” (according to which avowals are expressions that are not truth-apt) and “neo-expressivism” (according to which avowals are truth-apt expressions), I’ve given up resisting the word “expressivist” as a label for, among other things, the sort of position I’ve defended since Finkelstein (1994). 21 Evans (1982), 225.

From Transparency to Expressivism

113

like you to consider an objection to this line of thought. Imagine a detectivist who says the following: Suppose that someone asks you whether you believe that it’s about to rain. You walk over to a nearby window, peer at the sky, and come to some conclusion; you form a judgment concerning whether or not it is about to rain. According to Evans, what is your “procedure” supposed to be at this point? (You haven’t yet answered your interlocutor.) Presumably, the idea is this: If you have judged that it is about to rain (and so are “in a position to assert that p”), then you are to utter the sentence “I believe that it’s about to rain” (or some suitable equivalent). If, on the other hand, you’ve judged that it’s not about to rain, then you’re to say instead, “I don’t believe that it’s about to rain”. But now: how are you supposed to know the content of the judgment you’ve reached concerning whether it’s about to rain? In order to follow Evans’s “simple rule”, you must have some way of finding out what judgment (if any) you have made about the outward-directed question. Thus, we still need to posit an inner sense—one that informs the subject of his own judgments—in order to understand how someone can say what he believes about, e. g., the weather. Really, all that Evans provides is a “procedure” by which a person can learn what she believes about some subject matter, given that she already (somehow) knows what she judges about it. And this is to offer precious little.22

I believe that we can make significant progress toward understanding first-person authority—not only about self-ascriptions of belief, but about a wide range of self-attributions—by thinking about how one might reply to this objection. A first pass at such a reply might go: What Evans means to be suggesting is not that you’re entitled to utter a sentence of the form “I believe that p” just in case you have (somehow) learned a particular fact about yourself, viz., that you judge p to be true. It would be better to say that, according to Evans, we simply learn to use sentences of the form “P” and “I believe that p” interchangeably, at least to a certain extent. Thus, having just looked out your win22 A version of this objection could be raised against Moran’s position. As we saw, according to Moran, I am able to say whether or not I believe that p (or fear x, or…) by coming to a deliberative conclusion about whether the reasons call for me to believe that p (or to fear x, or…). Our imagined detectivist would respond: “How, given this account of self-knowledge, are you supposed to know what deliberative conclusion (if any) you have reached on the question of whether the reasons call for you to believe that p (or to fear x)? Here too, we need to posit an inner sense—one that informs you of your own judgment about whether you ought rationally to believe that p (or to fear x). Without such a mechanism, you will be in no position to say how you have made up your mind—even if you have managed to make it up on the basis of the appropriate reasons.”

114

David H. Finkelstein

dow at a fast-approaching bank of black clouds, you might say either, “It’s about to rain”, or, “I believe that it’s about to rain”. The latter assertion calls for no more inward observation than the former. But suppose that our detectivist is unsatisfied with this reply to his objection. Imagine that he answers it as follows: You’re failing to grasp the seriousness of my objection to Evans. You should think about a case in which a person is unaware of some judgment that he makes. Suppose that I unconsciously judge (perhaps on the basis of good, though painful, evidence) that my mother loves my brother more than she loves me. Thus I come to unconsciously believe that she loves him more. Now, it may be possible for me to gain knowledge of this belief of mine about my mother. But Evans doesn’t suggest a way for me to do it. The interchangeability of “My mother loves my brother more” and “I believe that my mother loves my brother more” does me no good, for I am not in a position to say either. I can take advantage of Evans’s “simple rule” only when my judgment concerning an outward-directed matter is already conscious—so only when my judgment that p is already something that I am (somehow) aware of. The postulation of an inner sense is needed in order to account for this sort of awareness. (So: my judgment that p is conscious just in case I learn about it via the inner sense.) The interchangeability of “P” and “I believe that p” simply does not account for the difference between: (1) a case in which someone can say what he believes, e. g., one in which you consciously believe that it’s about to rain; and (2) a case in which someone is unable to say what he believes, e. g., one in which I unconsciously believe that my mother loves my brother more than she loves me.

Note that our imagined detectivist doesn’t deny that “P” and “I believe that p” are, as it were, interchangeable. But he cannot see how this fact obviates the need for inner sense. (Indeed, what our detectivist is liable to take away from Evans is that inner sense is needed merely to look out a window and say, “It’s raining”.) Now, while there is something to our detectivist’s objection, he is nonetheless missing a point of crucial significance. As a step toward bringing this point into view, we might wonder about the interchangeability of “P” and “I believe that p”. Why, in spite of their having different truth conditions, am I able to use (e. g.) “It’s about to rain” and “I believe it’s about to rain” so interchangeably? The answer could be put this way: It’s because (in most circumstances) I’ll express the same attitude—the same belief—regardless of which of these sentences I utter. If someone wants to know whether I believe that it’s about to rain, I can give him what he wants by uttering either sentence. Notice that one up-

From Transparency to Expressivism

115

shot of this is that “I believe it’s about to rain” may be used to express the very state of mind that it self-ascribes. For all that Evans says, our imagined detectivist cannot see how selfascriptions of belief are supposed to be possible unless the self-ascriber inwardly detects either her own beliefs or the judgments that correspond to them. I want to suggest that the crucial point missed by our detectivist just is that a statement like “I believe it’s about to rain” may be understood as an expression of the speaker’s state of mind—an expression of the speaker’s belief that (and also her judgment that) 23 it is about to rain. Suppose that, upon looking out a window and concluding that it’s about to rain, you feel disappointed and sad. (Perhaps it’s your wedding day, and you’ve arranged for an outdoor reception.) You utter the sentence “I believe that it’s about to rain”, whereupon you collapse onto the floor and begin to cry. We should understand the relation between your belief and your self-ascription of belief as akin to that between your sadness and your crying—so: not as depending on inward detection, but as an expression, a manifestation, of your psychological condition.24 At the start of the present essay, I asked how it might be possible to reject detectivism without making it appear miraculous that we are able to easily and accurately self-ascribe beliefs, desires, fears, hopes, and the like. We have arrived at what I take to be a good answer (or, anyway, 23 I’m not suggesting anything controversial here about how to understand the relation between a belief that p and a judgment that p. The point is, rather, that an utterance like “I believe it’s about to rain” may express various psychological states and goings-on: not only a belief about the weather and its corresponding judgment—however precisely we understand the relation between beliefs and judgments—but (e. g.) disappointment, resignation, or anger. Just as it’s a mistake to hold that if something is an expression, it can’t also be a truth-evaluable assertion, it’s a mistake to hold that if something expresses a belief, it can’t express anything else. 24 Because he doesn’t see this point, our imagined detectivist (also) misunderstands the distinction between conscious judgments/beliefs and unconscious ones. What it means for a belief or a judgment to be conscious, rather than unconscious, is not that the subject is aware of it—or that she is aware of it via inner sense. Rather, a conscious belief or judgment (or desire or fear) is one that the subject is able to express in a particular way—by self-ascribing it. (I defend this claim in Finkelstein (1999) and Finkelstein (2003), §§5.4 – 5.5.) We can, after all, account for the difference between the two cases that our imagined detectivist adduces in his objection to Evans (the case in which someone consciously believes that it’s raining and the case in which someone unconsciously believes that his mother loves his brother more than she loves him) without appealing to an inner sense.

116

David H. Finkelstein

the start of a good answer) to this question. We can reject detectivism— at least about belief self-ascriptions—by recognizing that saying, e. g., “I believe it’s about to rain” is a way in which I can express (rather than report an observation of) my attitude about the weather.25 Evans brings us close to this point by, in effect, claiming (correctly) that I no more need to look inward in order to say, “I believe that p”, than I need to look inward in order to say, “P”. But the significance of this claim comes into focus only when we attend to what explains it—viz., the fact that saying, “P”, and saying, “I believe that p”, are two (amongst various) ways in which I might express my belief that p. Until we get a firm grip on the point that (as I put it a moment ago) Evans only brings us close to, we are likely to react to what he says about belief self-ascriptions in one or the other of two ways. On the one hand, we might think, as our imagined detectivist does, that Evans offers “precious little”—showing merely that once I have found out whether I judge that p is true, I can say whether I believe that p is true without any further inward observation. On the other hand, we might have a sense that Evans is onto something important, but—failing to get the point about expression into focus—fixate on the fact that “P” and “I believe that p” may be used in place of each other across a wide range of contexts. Then we’re liable to think that in order to account for the authority with which we speak about attitudes other than belief, we should try to find outward-directed questions that are, for the subject, transparent to such questions as “Do I want to go to a movie this evening?”. Once we’ve started down this road, we are bound to conclude that first-person authority has no single explanation—if only because we are not going to find outward-directed questions to which questions like “Am I feeling pain?” or “Am I in a good mood?” are transparent. We can, after all, extend a non-detectivist understanding of belief self-ascriptions not only to other attitudes (regardless of whether or not they jibe with our assessments of what’s rationally called for), but to moods and sensations as well. Just as I can express my belief by saying, “I believe…”; I can express my desire by saying, “I want…”; my hope 25 To express one’s state of mind is not to report an observation of it; I take this to be a point of significance for the later Wittgenstein. When a child comes into language and learns, among other things, to express his states of mind by selfascribing them, a great deal changes, but expression doesn’t suddenly call for an epistemic basis. For more on Wittgenstein’s treatment of this point, see Finkelstein (2001) and Finkelstein (2010).

From Transparency to Expressivism

117

by saying, “I hope…”; my elation by saying, “I’m elated”; and the pain I feel in my shoulder by saying, “I have a pain in my shoulder”. Just as I needn’t inwardly detect my belief about the weather in order to express it by saying, “I believe that it’s about to rain”, I needn’t inwardly detect my irrational fear of spiders in order to express it by saying, “I’m afraid of spiders”. In §1, I characterized the “transparency approach”—according to which we are able to speak about our own states of mind by answering questions about the world outside us—as an approach toward understanding self-knowledge and first-person authority. I have, in effect, argued that this approach gets us only partway to its goal. I’ve also sketched what I take to be a more satisfactory way of addressing the question from which I began, the question of how we can reject detectivism without making our capacity to self-ascribe psychological states and events come to seem miraculous. Filling in this sketch would entail saying more than I have here about, among other things, how to understand the notion of expression. But, if you have followed me this far, then you’ll understand why I think the best response to Evans (and Shoemaker and Moran) is to see what’s right in the transparency approach as pointing in the direction of an expressivist one.

References Bar-On (2004): Dorit Bar-On, Speaking My Mind: Expression and Self-Knowledge, Oxford. Boyle (2009): Matthew Boyle, “Two Kinds of Self-Knowledge”, in: Philosophy and Phenomenological Research 78 (1), 133 – 164. Evans (1982): Gareth Evans, The Varieties of Reference, J. McDowell (ed.), Oxford. Finkelstein (1994): David H. Finkelstein, Speaking My Mind: First-Person Authority and Conscious Mentality, Ph.D. Dissertation. University of Pittsburgh, Ann Arbor: ProQuest/UMI. (Publication No. 9521463. Finkelstein (1999): David H. Finkelstein, “On the Distinction between Conscious and Unconscious States of Mind” in: American Philosophical Quarterly 36 (2), 79 – 100. Finkelstein (2001): David H. Finkelstein, “Wittgenstein’s ‘Plan for the Treatment of Psychological Concepts’”, in: T. McCarthy/S. Stidd (eds.), Wittgenstein in America, Oxford, 215 – 236. Finkelstein (2003): David H. Finkelstein, Expression and the Inner, Cambridge/ MA. Finkelstein (2010): David H. Finkelstein, “Expression and Avowal”, in: K. Jolley (ed.), Wittgenstein: Key Concepts, Durham/UK, 185 – 198.

118

David H. Finkelstein

Locke (1689/1975): John Locke, An Essay Concerning Human Understanding, P.H. Nidditch (ed.), Oxford. Moran (2001): Richard Moran, Authority and Estrangement: An Essay on SelfKnowledge, Princeton. Moran (2003): Richard Moran, “Responses to O’Brien and Shoemaker”, in: European Journal of Philosophy 11 (3), 402 – 419. Shoemaker (1994): Sydney Shoemaker, “Self-Knowledge and ‘Inner Sense’, Lecture II: The Broad Perceptual Model”, in: Philosophy and Phenomenological Research 54 (2), 271 – 290.

On Finkelstein’s Account of the Distinction between Conscious and Unconscious States of Mind1 James Doyle 1. Introduction Some of my mental states are conscious – all my sensations, presumably, and many beliefs, hopes, fears, emotions, and so on. Others of my mental states, it seems, and I will assume, are unconscious. I may be consciously angry with my father, but I might be unconsciously so instead. An account of what it is for a mental state of mine to be conscious should make sense of certain marked dissimilarities between my relation to my own conscious states and my relation to other people’s. For whether or not I can be mistaken about my own conscious states, I do not seem to know about them, when I do, on the basis of anything like evidence, and there seems to be something strained and implausible in the idea that I know about them from anything like observation. Whereas evidence and observation are of course very much to the point in my ascriptions of conscious states to other people, and theirs to me. This ‘first-/third-person asymmetry’ has led some philosophers to try to think of (at least some) first-person avowals of conscious mental states as first and foremost expressions of them, so that “I have a headache” expresses my pain in something like the way a groan would. This certainly gives us asymmetry: I can hardly express someone else’s pain in that sort of way. But this sort of account seems to miss out something else important: what the first-person avowal and third-person ascription have in common. For they both seem to be ascriptions: when you tell someone else that I have a headache, you seem to be ascribing the same state to me as I ascribe to myself when I say I have a headache; and both ascriptions seem to be made true by the same state of affairs, 1

I would like to thank Jim Conant, David Finkelstein, James Ladyman and participants in the University of Chicago Wittgenstein Workshop for helpful comments on earlier versions of this paper.

120

James Doyle

namely, my having a headache. But the more closely my avowal is assimilated to a groan, the less sense it makes to suppose that it could be made true by my having a headache, or indeed by any state of affairs at all. This problem with the view of avowals as expressions may incline us, in turn, to think of them as straightforward self-ascriptions on a par with “I have a rash on my arm”; but then we seem to lose sight of the first-/third-person asymmetry. If the self-ascription is made true by the presence of the same conscious state as the corresponding third-person ascription, we are tempted to suppose that the first-person relation to the state is a kind of super-reliable version of the third-person relation, and so we are led to postulate a special observation-like faculty of introspection that threatens to raise many more philosophical problems than we started out with.

2. Finkelstein’s View of the Nature of Conscious States In his book Expression and the Inner, David Finkelstein presents a novel and interesting account of the distinction between conscious and unconscious mental states. He begins by considering what he calls the simple account of consciousness put forward by Colin McGinn: your mental state is conscious if you know that you are in it, and unconscious if you don’t know that you are in it.2 Finkelstein notes that this is shown to be false by the fact that I can be in an unconscious state I nevertheless know about – for example, if I have been rationally persuaded I am in it by an analyst who herself knows about it and whom I know to be reliable. So the ability to self-ascribe a mental state, even from knowledge, is not sufficient for that state’s being conscious; and my relation to an unconscious state I self-ascribe seems to belong on the third-person side of the first-/third-person asymmetry: I gain access to the state through evidence, observation, testimony, or some combination of these. But nor is the ability to express the state sufficient for consciousness. For just as I may come to know about my anger toward my father from my analyst, so may I express that anger by forgetting his birthday; and in neither case does my anger attain to consciousness. Finkelstein’s 2

Finkelstein (2003). Page references to this work will be included parenthetically in the text. See also Finkelstein (1999). McGinn proposes the simple account, at least for the so-called propositional attitudes, in McGinn (1979), 37; cited in Finkelstein (2003), 115.

On Finkelstein’s Account

121

account combines the conditions: first-person avowals of conscious states simultaneously express and self-ascribe those states, and they express them in virtue of self-ascribing them. This account is attractive because it enjoys the advantages of both sorts of view mentioned above, while cancelling out their defects. The account says that first-person avowals are to be understood as truth-apt, and as made true by the same state of affairs as makes true the corresponding third-person avowal. Nevertheless I have a very special relation to the conscious state I self-ascribe. This relation is not a special way I have of knowing about it, but a special capacity I have for expressing it: only I can express the state by self-ascribing it. Furthermore, I only have this expressive ability with regard to those mental states of mine which are conscious, and my having this ability is in fact what it is for the relevant state to be conscious: Someone’s mental state is conscious if he has an ability to express it simply by self-ascribing it. If he lacks such an ability with respect to one of his mental states, it is unconscious.3

So for example, if I am angry with my father consciously, I can express that anger by saying “I am angry with him” (or saying to him “I am angry with you”); but if I am angry with him unconsciously, while I can still ascribe this anger to myself, e. g. because I have learnt about it from a reliable analyst, I cannot express it when I so ascribe it to myself, any more than I can express another’s anger in ascribing it to him.4

3. Finkelstein’s Blasé Line on the Issue of Knowledge McGinn was clearly wrong to think that my knowledge that I am in a mental state is sufficient for its being conscious, but it’s natural to suppose that it’s necessary: isn’t it obvious that I know I’m in those mental states I’m conscious of ? Wittgenstein (as some interpret him) was the first to challenge this idea, at least for the case of sensations,5 roughly on the 3 4 5

Finkelstein (2003), 120. I may consciously express anger toward my father in the self-ascription, but if the anger I intend to ascribe to myself is unconscious and real, that is not the anger I am expressing. See Lear (2004), 448 – 454. Wittgenstein (2001), §246; cf. Anscombe’s extraordinary claim that “being able to say where one feels pain is [not] a case of something known” (Anscombe (2000), §8).

122

James Doyle

grounds that we are too closely and securely related to our conscious states to count as knowing about them. In his own account, Finkelstein doesn’t mention our status as knowers in relation to these states. This is neither because he thinks the necessity of knowledge too obvious to mention in the aftermath of his refutation of McGinn’s simple account, nor from a Wittgensteinian suspicion of the idea that we know that we’re in these states. It turns out that, although Finkelstein does want to deny that our self-ascriptions of (e. g.) pain are ‘epistemically justified’, he does not want to insist that they don’t count as expressions of knowledge. He doesn’t insist either way: he is blas about the issue of knowledge. He considers the case of Hank, who reads in a medical journal that someone who gets a sharp pain in the back of his knee is more likely than the average American to get Green’s disease, and later experiences just such a pain and so infers, and perhaps knows, that he’s more likely than the average American to get Green’s disease. Finkelstein writes (this is the final paragraph of his book): Imagine three philosophers who are talking about Hank’s case. The first says: “Hank has a sharp pain in the back of his knee, but it’s wrong to speak of him as either knowing or not knowing about it. Such talk is, at best, unnatural and misleading and, at worst, sheer nonsense.” The second philosopher says: “Hank has ascribed a pain to himself, and he is entitled to make inferences on the basis of this self-ascription. But such an ascription of pain is not epistemically justified; he doesn’t know that he is in pain.” The third says: “Hank does know that he’s in pain; we can say this about him. The mistake is to think that knowledge always requires epistemic justification. He knows that he’s in pain, but knowing this isn’t like knowing that one has termites in one’s basement.” I don’t see that these philosophers need be disagreeing about anything of genuine philosophical import. Moreover, it seems to me that all three of them might agree with everything I’ve said in this book.6

4. My Claims In this paper, I argue that: (i) Finkelstein cannot afford to be so blas about the issue of knowledge: if he is basically right about the nature of conscious states, these philosophers are disagreeing about something of genuine philosophical import – at least, the first two are disagreeing with the third about something important, viz whether Hank knows he is in pain. (ii) In fact, none of these philosophers can consistently agree 6

Finkelstein (2003), 151 f.

On Finkelstein’s Account

123

with everything Finkelstein says in his book, since (iii) if Finkelstein is basically right about the nature of conscious states, the first two of these three philosophers are entirely wrong, while the third, although less wrong, is still insufficiently hospitable to knowledge-talk in connection with our conscious mental states. In particular (again, if Finkelstein is basically right), knowing that one is in pain must be in at least some important respect like knowing that one has termites in one’s basement. (We will get clearer later on about what this respect is.) In short: If Finkelstein wants to hold onto his account of the nature of conscious mental states, he cannot afford to be blas about the knowledge issue – he must insist that we do know that we’re in those mental states of ours that are conscious –, and so (contraposing), if he wants to be blas about the knowledge issue, he needs a different account of the states.

5. Problems for Finkelstein’s Account of the Nature of Conscious Mental States I shall first make an assumption about what Finkelstein means by an ability to express a state through self-ascription: namely, that if someone does express a state in this way, it follows that she has such an ability. This may seem to follow a fortiori, but my assumption may be challenged, and I will call it into question in section 8 below. But for now it is an assumption; I will call it the Assumption. In a fascinating paper on the Socratic method and psychoanalysis, Jonathan Lear writes: When analysands start to talk explicitly about how to live, they are often using their capacity for intellectual thought to maintain a distance from what is really on their minds. When they talk about some psychological trait they have – “I have this obsessive need to…” – it is usually in the service of staying with pat formulations, and thereby avoiding any genuine insight into what they are like. It may even be that the content of what they are saying is true; the problem is that it does not connect to their lives in a living sort of way. Psychoanalysts call these resistances to genuine analysis. In short, “raising the question of how to live” can be a way of avoiding the question of how to live.7

7

Lear (2009), 447; see also Lear (2003).

124

James Doyle

I am interested in a special subset of this sort of case in which, in a further ironic twist, the analysand avoids the problem by identifying it and self-ascribing it correctly, and the avoidance manifests the problem. These cases show that Finkelstein was wrong to say “while it is possible for someone to ascribe an unconscious belief to himself, he does not thereby express the belief”8. Hence,

Resistance Analysand: I have this obsessive need to label, categorise and intellectualise my problems as a way of distancing myself from them. In such a case the analysand is (correctly, we may assume) ascribing a mental state to herself, a certain obsessive inclination, and in the process of the self-ascription, she is also expressing it: the purported identification of the inclination is itself an instance of distancing intellectualisation. On Finkelstein’s account, it follows that the state in question is conscious. The difficulty, of course, is that it’s not. If it were, the ascription would not count as a kind of resistance whereby, in terms of Finkelstein’s crucial distinction9, the analysand has become conscious (that is, aware) of this intellectualising episode without however making it a case of intellectualising consciously. Call this sort of counterexample a resistance case. Such cases can be multiplied:

Positive Transference Analyst: I sometimes wonder whether you might have an unconscious urge to take any opportunity that presents itself to seize upon anything I say about your condition, even if it’s a rather airy speculation, and treat it as though it were a profound insight of fundamental importance. It’s as if you believed that I was virtually infallible. Analysand: You know something? You’re right! I do have such an urge! It’s amazing – you’ve really put your finger on something there. And in a funny sort of way, I suppose I do think that you’re virtually infallible. 8 9

Finkelstein (2003), 121, fn. 10. Ibid.

On Finkelstein’s Account

125

Negative Transference Analyst: I’m starting to think you have an aversion to contradicting me. Analysand: Rubbish! I learnt from my previous analyst that I do tend to contradict in therapy when it’s my condition that’s at issue.10 As I intend these cases to be understood, in the moment, I think I am simply expressing and endorsing the analyst’s insight; indeed, I am expressing and endorsing it, but I am also expressing the condition this is an insight into, and this escapes me.11

6. The “Adverbial” Defence: Finkelstein’s “Clipped Tone of Voice” Case Finkelstein anticipates and pre-emptively responds to a counter-example that may look similar to these. To the summary of his account quoted above, “Someone’s mental state is conscious if he has an ability to express it merely by self-ascribing it”12, he appends a footnote:

10 These exchanges are a travesty of the psychoanalytic aspiration of avoiding suggestion, and are at best cartoonish representations of transference and resistance; their only point here is to draw attention to a certain kind of psychological possibility. I call them ‘resistance cases’, but actually the relevant kind of possibility can be illustrated by cases in which the state in question isn’t neurotic or otherwise problematic: Judiciousness Analyst: You know, I’m starting to think that you have a strong tendency, whenever I speculate about your condition, to consider the matter very carefully before you come down for or against. Analysand: [long pause] I dare say you’re right about that. Decisiveness Analyst: When I speculate about your condition, your response is always Yes or No, isn’t it? – there’s no grey area. Analysand: Yes. 11 The resistance cases actually provide circumstantial evidence for the centrality of expression-through-self-ascription to the status of mental states as conscious. For this is where a good part of the irony derives from: the states that drive my resistance are that close to being conscious, but my true and justified self-ascription is still an expression of my problem rather than a solution to it. Expression through self-ascription is central to consciousness, but by itself it is not enough. 12 Finkelstein (2003), 120.

126

James Doyle

The word “merely” as I’m using it here is meant to forestall an objection that might be put as follows: “Imagine that you occasionally express your unconscious anger toward your sister by speaking in a peculiar, clipped tone of voice. One day, while speaking in this tone of voice, you say, “My therapist tells me that I’m unconsciously angry with my sister, and I suppose she must be right.” Through your tone of voice, you express your anger toward your sister in a self-ascription of it, even thought the anger is unconscious. What this case shows is that being able to express your state of mind by self-ascribing it is not a sufficient condition for the state’s being conscious.” When I say that someone’s state of mind is conscious if he has an ability to express it merely by ascribing it to himself, I mean this: the sort of ability at issue is one that enables a person to express his state of mind in a self-ascription of it, where what matters – what carries the expressive force – isn’t his tone of voice (or whether he is tapping his foot, or what he is wearing, or to whom he happens to be speaking), but simply the fact that he is giving voice to his sincere judgment about his own state of mind. That I might manage to express my unconscious anger in a self-ascription of it via a clipped tone of voice doesn’t show that I have the relevant sort of expressive ability. When I am consciously angry, I can say in a neutral tone of voice, “I’m angry,” and thereby express my anger.13

Now it is true that my cases include such ‘adverbial’ features. In Positive Transference, for example, I had to make the analysand’s line an exclamation, in order to make it express the affective component of the self-ascribed condition: she has to seize upon the analyst’s suggestion. But my point can be conveyed by cases that lack any such adverbial dimension. Consider Positive Transference II Analyst: You seem to have an inclination to agree with my speculations about your condition. Analysand: You’re right. I do. We may imagine the analysand replying here in an emotionally colourless way, since the case hinges entirely on the content of what she says. Since the unconscious condition in question is an inclination to agree with the analyst, it is enough, for the analysand to express it, that she simply agree with him; and this agreement is entirely a matter of the content of her assertion – that it be the same as his. She expresses her 13 Ibid. fn. 9.

On Finkelstein’s Account

127

condition, in Finkelstein’s words, “merely by ascribing it to herself,… where what matters – what carries the expressive force – isn’t her tone of voice (or whether she is tapping her foot, or what she is wearing, or to whom she happens to be speaking14), but simply the fact that she is giving voice to her sincere judgment about her own state of mind.” Since Positive Transference II eludes the adverbial defence, it will serve as our representative counterexample from now on.

7. The Knowledge Requirement My claim, to repeat, is not that Positive Transference II, or resistance cases generally, discredit the basic idea behind Finkelstein’s account of the nature of conscious states, but that they show that he cannot be blas about the issue of knowledge: his account must include the requirement that, for a subject’s state to count as conscious, she must know that she is in that state – and, in fact, her self-ascription must be from knowledge. I will refer to this as the knowledge requirement. So the revised account reads Someone’s mental state is conscious if he has an ability to express it simply by self-ascribing it from knowledge. If he lacks such an ability with respect to one of his mental states, it is unconscious. If Finkelstein’s account is modified to include the knowledge requirement, Positive Transference II seems no longer to work as a counterexample to it. Why not? The basic thought is this: the idea of speaking from knowledge excludes the possibility that the thoughts one is expressing are simultaneously determined by psychological forces beyond one’s knowledge and control. If the analysand in Positive Transference II is expressing an inclination to agree with her analyst that remains unconscious, the fact that she is expressing the inclination must be likewise unconscious. Since her thoughts are to this extent not under her conscious control, she cannot be speaking from (her utterance cannot be explained as an expression of) knowledge; while, conversely, if she really is speaking from knowledge, then either she is not expressing the unconscious state 14 Well, of course, it does matter that she’s talking to the analyst, but only because (i) she uses the second person (“You’re right”) and (ii) the relevant unconscious condition is an inclination to agree with him. The same condition could be expressed in a self-ascription addressed to a third party in which the third person of the same verb was used: “He said I’m inclined to agree with him. And he’s right.” The condition can still remain unconscious.

128

James Doyle

she knowledgeably self-ascribes (as in the case by appeal to which Finkelstein refuted McGinn’s simple view), or she expresses it in the way that Finkelstein’s original account was intended to capture – in the way, that is, that one expresses a conscious state. The fishy thing about the resistance cases was that the analysand expressed her condition by some extrinsic feature of her self-ascription. She does not express her condition via a self-ascription of it under that description, or as such. Her self-ascription in Positive Transference II, for example, counts as an expression of the ascribed condition, in the first instance, because it is a case of her agreeing with her analyst. Her expression of agreement is only a self-ascription of the inclination to agree because that inclination is what her analyst happens to have just ascribed to her! The expressed condition coincides with the self-ascribed condition only given the radical contingency that that was the analyst’s immediately preceding judgement. Her agreement with any other judgement of the analyst’s (about her condition) would have served just as well, as an expression of the inclination to agree. The fact that the object of her agreement is an ascription of the inclination to agree gives the exchange an extra ironic twist, but it’s only a self-ascription of that inclination, so to say, per accidens. What’s doing the expressive work is the fact of agreement, not the content that’s agreed to. The knowledge requirement blocks the resistance counterexamples because, if the analysand is self-ascribing from knowledge, the content of her judgement, considered as such (and not merely in virtue of some extrinsic feature) is bound to express the self-ascribed condition in just the way that indicates its status as conscious, for the expressibility through knowledgeable self-ascription just is the hallmark of a conscious state. Finkelstein thought that the lesson of McGinn’s failed simple view was that the expressive condition should replace the knowledge condition. But the lesson of the resistance cases is that it should supplement it. The case of learning from my analyst shows that if the state is conscious the self-ascription must express it, and the resistance cases show that this expressive self-ascription must be from knowledge. If the knowledge requirement is in place, then, conscious states turn out to be all and only those the subject can express in (knowledgeable) self-ascription, and the basic shape of Finkelstein’s account is preserved.15 Furthermore, it looks as though only the knowledge require15 Someone might worry that Finkelstein’s account is now altered beyond recognition. That account was distinctively ‘non-epistemic’: it sought to explain

On Finkelstein’s Account

129

ment can block the resistance counterexamples. So long as the subject does not know she is in the state, the possibility will remain that in self-ascribing it she is expressing it in a way that escapes her.

8. Dropping the Assumption The Assumption was, that if the subject expresses the mental state by self-ascribing it, then she ipso facto, and so trivially, has the ability to do so, where this ability, on Finkelstein’s original account, is what guarantees that the state is among the conscious states of the subject. But not all abilities are governed by such an Assumption. My ability to metabolise glucose is so governed: my metabolising it trivially entitles me to claim the ability to do so. But my ability to wiggle my ears, as this is usually understood, is not so governed: if I wiggle them by a fluke, it doesn’t count, because this is an ability that is essentially exercised at will. Some of the things Finkelstein says about the ability suggest that he is thinking of it as an ability that is exercised at will.16 what it is for a state to be conscious without invoking the idea of a ‘special way of knowing’, which has started to look to many like a philosophical dead-end. Am I not re-introducing knowledge by the back door and so compromising the account’s main selling point? No: to insist on the subject’s knowledge of her conscious states as a condition of their being conscious is not to concede anything to the view that a special kind of knowledge on the subject’s part has anything to do with what makes the states conscious. This is an important point: we will see below that the concept of knowledge, as applicable indifferently to conscious states and things observed, must be considered a unity precisely because the revised account does not invoke any such special way of knowing, any more than Finkelstein’s original version did. I would like to record here a residual worry that, even with the knowledge requirement, Finkelstein’s account may still be vulnerable to the resistance counterexamples. My claim that the knowledge requirement enables us to deny that the state expressed in self-ascription in (e. g.) Positive Transference II is conscious depends upon the principle that we cannot be said to know what we are (even partly) unconsciously impelled to affirm. But certain cases of teaching and learning make this principle look insecure: the authority of a teacher may instil an unconscious inclination to believe what he says, and yet he may still exploit this inclination to impart knowledge. See Augustine (1998) and (1992). 16 The potential Harpo counterexample (“Objection 2” in Finkelstein’s paper “On the distinction between conscious and unconscious states of mind” (1999), 94 f.) is dealt with by describing Harpo as possessing the ability to express his anger at his brother by self-ascribing it, but not exercising this ability because

130

James Doyle

If Finkelstein’s account is understood in this way, it is not vulnerable to the resistance counterexamples, which turn out to be irrelevant because my ability in those cases to express my unconscious inclination, being itself exercised unconsciously, is therefore exercised independently of my will. Nevertheless, dropping the Assumption is not a way for Finkelstein’s account to avoid having to incorporate the knowledge requirement. For (i) dropping the Assumption from the account actually amounts to adding in the knowledge requirement and more; and (ii) the more that is added in makes the account implausible, because the conditions it places on a mental state’s being conscious are now far too stringent. I will take these two points in turn. (i) The ability to do something at will (say, wiggle my ears) is the ability to do that thing intentionally. Anscombe argued that if I intentionally ¦ under that description, I must know without observation that the description ¦-ing applies to what I am doing.17 Davidson shows that this isn’t quite right: “A man may be making ten carbon copies as he writes, and this may be intentional; yet he may not know that he is; all he knows is that he is trying.”18 Such a man need not even believe that he is succeeding! And similarly, someone might intentionally express their state through self-ascription, even though all they knew was that that was what they were trying to do.19 But if we transpose our talk from actions to abilities, something like Anscombe’s claim looks right. It is hard to see how we could credit someone with an ability to ¦ at will if they did not know, when they were ¦-ing, that they were ¦-ing. So if the Assumption is dropped, whenever I express a mental state by ascribing it to myself in the way that constitutes that state as conscious, I must at least know that I am expressing it. (It is not enough for me to know merely that I am self-ascribing it. An account with this weaker requirement would once again be vulnerable to the resistance counterexamples it was the point of dropping the Assumption to avoid.) But I cannot express a mental state I am not in, so any doubt he is afraid to; and Finkelstein clearly means that the fear is functioning as a reason and not as (mere) cause. I cannot stop metabolising glucose for fear that something bad may happen if I continue to do so, since I cannot (normally) decide to stop metabolising glucose for any reason at all; contrast wiggling my ears. 17 Anscombe (2000), §6. 18 Davidson (1980), 50. 19 A message in a bottle might be an example of this, or trying to speak a foreign language one has not yet mastered.

On Finkelstein’s Account

131

about my being in the state will invalidate my claim to know that I’m expressing it. It follows that on Finkelstein’s account minus the Assumption, any mental state of mine that is conscious must be one I know I am in when I express it, and which I know I am expressing, and which I therefore express from knowledge. Such an account therefore recognises the knowledge requirement. (ii) The argument of (i) showed that, on Finkelstein’s account minus the Assumption, if I am exercising the right sort of ability to express a state in self-ascription (i. e. an ability exercised intentionally), I must know, not only that I am in the state, but also that I am expressing it in self-ascription. The ability is supposed to confer the status conscious on the state in question, so any conscious state requires an ability whose exercise involves all this knowledge. But this is surely too much. There are plenty of people who enjoy all manner of conscious states, but don’t even grasp the concept of expressing a state, never mind the highly arcane expressing a state through self-ascription. Dropping the Assumption from Finkelstein’s account leaves it with conditions on mental states’ being conscious that are far too stringent.20

9. The Knowledge Requirement Can only Do Its Job if the Relevant Concept of Knowledge Is a Unity Someone who had worries (e. g. deriving from Philosophical Investigations §246) about talking of knowing that one is in those of one’s mental states that are conscious, but (like Finkelstein’s third philosopher) didn’t want to ban such talk, might say: here the concept of knowledge is given application merely homonymously or by courtesy. But someone who held this view could not build knowledge into the account of conscious states, because they could only understand the concept disjunctively: either the relation to the fact about termites, or the relation to the fact about the conscious state. But if the covering-concept of 20 The conditions may be too stringent in another way, too. When one ¦s intentionally, one knows that one is ¦ing without observation if one knows this at all. See Anscombe (2000), §28. I suspect that our best account of intentional action here effectively presupposes the idea of a state (e. g. the belief or knowledge that one is ¦ing or trying to ¦) being conscious. If so, Finkelstein’s account minus the Assumption, invoking as it does the concept of intentional action, would risk triviality. But I cannot pursue this issue here.

132

James Doyle

knowledge is understood as cobbled together in this way it cannot be deployed in an account of conscious states, because the second disjunct presupposes an understanding of our relation to them. This is what I meant when I said that knowing that one is in pain must be in at least some important respect like knowing that one has termites in one’s basement. Yet, so far as I can see, we may still deny, with Finkelstein, that the former kind of knowledge involves justification.

References Anscombe (2000): Gertrude Elisabeth Margaret Anscombe, Intention, Cambridge/MA. Augustine (1992): Augustine, De Utilitate Credendi, ber den Nutzen des Glaubens, N. Brox et al. (eds.), Freiburg. Augustine (1998): Augustine, De Magistro, ber den Lehrer, B. Mojsisch (ed.), Stuttgart. Davidson (1980): Donald Davidson, “Agency”, in: Essays on Actions and Events, Oxford, 43 – 61. Finkelstein (1999): David H. Finkelstein, “On the Distinction between Conscious and Unconscious States of Mind”, in: American Philosophical Quarterly 36 (2), 79 – 100. Finkelstein (2003): David H. Finkelstein, Expression and the Inner, Cambridge/ MA. Lear (2003): Jonathan Lear, Therapeutic Action: an Earnest Plea for Irony, New York. Lear (2004): Jonathan Lear, “Avowal and Unfreedom”, in: Philosophy and Phenomenological Research 69 (2), 448 – 454. Lear (2009): Jonathan Lear, “The Socratic Method and Psychoanalysis”, in: S. Ahbel-Rappe/R. Kamtekar (eds.), A Companion to Socrates, Malden/MA, 442 – 462. McGinn (1979): Colin McGinn, “Action and Its Explanation”, in: N. Bolton (ed.), Philosophical Problems in Psychology, New York, 20 – 42. Wittgenstein (2001): Ludwig Wittgenstein, Philosophical Investigations, Malden/ MA.

Practical Knowledge and the Structure of Action Will Small [T]he will is by nature active; cognitive phenomena, since they must bend to the condition of the world, may be described as passive in relation to the world they represent.1 [F]rom what glimpses I have caught of it I gather that this so-called practical knowledge is as queer a bird as any that ever flew in Plato’s famous aviary. Evasive as it is, I must try to lay hands on it though I doubt if it is a genuine bird at all.2 Can it be that there is something that modern philosophy has blankly misunderstood: namely what ancient and medieval philosophers meant by practical knowledge? Certainly in modern philosophy we have an incorrigibly contemplative conception of knowledge. Knowledge must be something that is judged as such by being in accordance with the facts. The facts, reality, are prior, and dictate what is to be said, if it is knowledge.3

Introduction If practical knowledge has any significance for epistemology, it must be because it is a distinctive form of knowledge, or is acquired via a distinctive way of knowing, and not merely because it is knowledge of a distinctive topic (practice, action) in virtue of which it has significance for those branches of philosophy with a prior interest in such subject matter (philosophy of action, ethics). Its prospects, by these lights, appear limited; practical knowledge surely does have such a distinctive and restricted topic, especially when compared with perceptual, inferential, and testimonial knowledge. Yet renewed interest in G.E.M. Anscombe’s Intention has provided some encouragement: Anscombe argues that if an agent is doing something intentionally, he knows “without observation” that he is doing it. This agent’s knowledge (as I’ll call it) is, Anscombe claims, practical knowledge, the formal character of which 1 2 3

McGinn (1996), 117. Jones (1960), 137. Anscombe (1963), §32.

134

Will Small

she characterizes positively with a slogan borrowed from Aquinas: practical knowledge is “the cause of what it understands”4. But the idea that causing, or acting, or intending, could be a way of knowing somehow like perceiving, inferring, or being told runs counter to deep philosophical intuitions about the nature of knowledge. Knowledge and belief, in Colin McGinn’s phrase, “must bend to the condition of the world”5 ; attaining a suitably non-accidental grip on the way things are anyway is what characterizes the kind of cognitive achievement that is knowledge. By contrast, causing, acting, and intending involve changing the way the world is, or representing it as to be so changed; so-called “conative” representations like intention and desire exhibit the opposite “direction of fit” from their “cognitive” counterparts. Such intuitions and commitments can suggest that the expression “practical knowledge” could not but fail to denote a form of knowledge: if what it refers to is really practical, then it cannot really be knowledge, and if what it refers to is really knowledge, then it cannot really be practical. Anscombe offers a diagnosis: it is but a dogma of modern philosophy to conceive of knowledge as everywhere speculative, contemplative, beholden to the way things are anyway; this is to project features characteristic of a species of knowledge onto the genus. “Speculative” knowledge is “derived from the objects known”6, but practical knowledge is not: it is the cause of what it understands, and supposedly no less knowledge for that. Yet the remedy is far from clear: J. David Velleman—who is more sympathetic than most to this strand in Anscombe’s thought—expresses a common reaction when he writes that her account leaves agent’s knowledge “looking not just causally perverse but epistemically mysterious”7. A defence of the idea that “practical knowledge” denotes a distinctive form of knowledge therefore must tread a thin line, between advancing a conception of something genuinely cognitive, but only dubiously practical (or practical merely in topic), and advancing a conception of something genuinely practical that is only dubiously cognitive, knowledge only so-called. In this essay, I try to tread this line. Following Anscombe, I defend the idea that there is a cognition condition on intention and intentional ac4 5 6 7

Anscombe (1963), §48. McGinn (1996), 117. Anscombe (1963), §48. Velleman (1989), 103.

Practical Knowledge and the Structure of Action

135

tion: an agent who is doing A intentionally has practical knowledge that he is doing A, and an agent who intends to do A has practical knowledge that he is going to do A. However, well-known counterexamples, devised by Donald Davidson and Michael Bratman, are typically thought to pose a serious problem for this kind of view. Consequently, recent attempts to preserve a defensible kernel of the Anscombean thought have typically proposed weakening the kind of cognition involved in the condition, and/or weakening the condition’s content or scope. By contrast, I argue that it is a mistake to weaken the cognition condition in any such way, and that the felt need to do so is symptomatic of a deep failure to understand the possibility of practical knowledge. The true ground and meaning of the cognition condition, properly understood, lies in the (i) calculative and (ii) temporal structure of intentional action, some brief remarks about each of which are in order. (i) Many of the things we intentionally do, we do by intentionally doing other things. Many of these other things we do by doing yet further things, and so on. The agent’s practical knowledge is the source of this calculative structure; without this knowledge, what happens is not intentional. Thus, it will emerge, an agent intentionally doing something knows not just what he is doing, but how and why he is doing it. This practical knowledge is, in Anscombe’s phrase, knowledge “in intention”8, not knowledge that inheres in a belief that somehow accompanies intention. Accordingly, an expression of intention is itself an expression of knowledge, of what the agent is actually doing (if it expresses an intention in action, e. g. ‘I’m walking to school’) or of what the agent is actually going to do (if it expresses an intention for the future, e. g. ‘I’m going to wash the car tomorrow’). And just as intentional action is calculatively articulated, so is intention for the future: an agent who intends to do something knows not just what he is going to do, but how and why he is going to do it. (ii) An intentional action, like any event, unfolds in time. If what is happening falls under an event concept v, then it will be no accident if that which is v-ing ends up having v-ed. If what is happening falls under an intentional action concept do A, then it will be no accident if the agent who is doing A intentionally ends up having done A intentionally. In the specific case of intentional action, the agent’s practical knowledge is the source of this temporal structure; without this knowledge, what is happening is not intentional. Thus, it will emerge, an agent intentionally 8

Anscombe (1963), §32.

136

Will Small

doing something knows not just what he is doing, but that it will be no accident if he succeeds. Similarly with intentional action in prospect: an agent who intends to do something knows not just what he is going to do, but that it will be no accident if he gets it done. For one to be justified in thinking that it will be no accident if one ends up having intentionally done what one is doing intentionally (or intends to do), it is necessary, I argue, that one knows how to do what one is doing (or intends to do). In many cases, this demand is supplied by procedural knowledge: one knows that one can do C by doing B. But if one exercises this knowledge, in intentionally doing C by doing B, then one must know how to do B, and though one might know that one can do B by doing A, it seems obvious that procedural knowledge cannot be the whole story: there must be a way of knowing how to do something that doesn’t consist in knowing by what means one can do it. This point about the structure of practical knowledge mirrors the standard suggestion made in response to a corresponding regress that threatens our thought about the structure of action: though many of the things we do we do by doing other things (and so on), this picture appears to depend on the possibilty of basic actions, things we can do directly, without doing them by doing anything else. However, Michael Thompson and Sebastian Rödl have recently argued that reflection on the calculative and temporal structure of intentional action—the very structure that, I argue, explains and validates the cognition condition, properly understood—reveals that there can be no such thing as basic action. This creates two problems, one metaphysical, one epistemological. The metaphysical problem is to explain how intentional action is so much as possible, if it can appear compelling that basic intentional actions are necessary, as naïve reflection on the idea of doing one thing by doing another suggests, and yet impossible, as deeper reflection allegedly reveals. The epistemological problem is to explain how agent’s knowledge is possible: If there is no such thing as doing something without doing it by doing something else, it seems there can be no such thing as having non-procedural knowledge of how to do something. But in the absence of such knowledge, I will argue, there can be no such thing as knowing that it will be no accident that one will end up having done intentionally what one is doing intentionally (or intends to do), which knowledge is internal to knowing what one is doing intentionally (or intends to do). The resolution of both problems is the same (for they are but the same problem, considered, as it were, from different sides): we need to recognize the way in

Practical Knowledge and the Structure of Action

137

which one form of practical knowledge—knowledge in intention—depends, metaphysically and epistemically (but not inferentially), on another—skill. While it is not uncommon for philosophers of action to think that an account of intentional action must be grounded in skill, the way in which skill is commonly conceived in fact prevents it from playing this kind of role; thus, we shall have to rethink what it means for skill to be a form of practical knowledge, too. The path taken in this essay will accordingly be as follows: in §1, I introduce the idea of a cognition condition on intention and intentional action via Anscombe’s claim and the purported counterexamples to it. In §2, I argue that the true ground and meaning of the cognition condition begins to come into focus through reflection on the calculative structure of intentional action in progress, and in §3 I show how and why this structure belongs to intention for the future, which is intentional action in prospect. I then argue that the cognition condition can be articulated adequately only by taking into account the temporality of intentional action (in progress and in prospect), and the way in which the temporal structure of intentional action depends on its calculative structure (§4). These reflections enable me to show, in §5, why the purported counterexamples from §1 do not get a grip on the cognition condition, properly understood, and why we need and ought not weaken it in the ways that some have recently suggested. However, the very same reflections also appear to saddle us with a paradox, suggesting both that there must be basic action and yet that there cannot be (§6). In the final part of the essay (§7), I propose a solution that accords the concept of skill its proper place in the theory of action.

1 The Cognition Condition 1.1 Anscombe’s Claim G.E.M. Anscombe famously claims that, if an agent is doing A intentionally, he knows “without observation” that he is doing it.9 By contrast, if an agent is asked why he is doing A and he (sincerely) responds “I didn’t know I was”, then this suffices, she thinks, to show that his doing A is not an intentional action. Moreover, even if the agent does know that he is doing A, but he knows this only on the basis of 9

Anscombe (1963), e. g. §28.

138

Will Small

observation, then this too suffices to disqualify his doing A as an intentional action. The basic idea here, that when you’re up to something intentionally, you don’t have to observe or theorize yourself to know what it is that you’re up to, whereas someone else has to find or figure out what it is that you’re up to, is quite intuitive. A hum-drum example: I find you in the kitchen, apparently making breakfast. You take the eggs out of the fridge, and I suppose you’re going to cook them (I know you’re not the type to crack one in a glass, whisk it up, and drink it raw). But how are you going to cook them? I rule out boiling when you take out the frying pan: you’ll fry, scramble, or poach them. And then, when I watch you beat them as the oil heats up, I know you’ll scramble them. But wait! After you pour the eggs into the pan, you don’t start scrambling; and it’s only now that I spot a bowl of grated cheese on the side—you were making an omelette the whole time! Of course, instead of observing your actions, forming hypotheses, drawing inferences, and so on, I could have just asked you what you were doing: not only were you making an omelette the whole time, you knew you were making an omelette the whole time—you weren’t gradually coming to that conclusion on the basis of the evidence as I was. Anscombe’s point is just this: the concurrence of these two facts—that you were making an omelette the whole time, and that you knew without observation (indeed, without any kind of evidence) that this is what you were doing—is no accident.10 Now, while Anscombe’s claim is usually thought of as applying just to intentional action, in fact it has a wider scope. Readers of Intention will recall that the book opens by introducing the topic of intention “under three heads: expression of intention for the future, intentional action, and intention in acting”11, and almost immediately Anscombe 10 Anscombe’s illustrations (1963), §28, of the relevant contrast cases—things done in the absence of the relevant non-observational knowledge, which absence thereby reveals them to be things done unintentionally—ring true too: someone is asked “Why are you ringing that bell?” and replies “Good heavens! I didn’t know I was ringing it” (ignorance of the action); and someone, opening a window and thereby focusing a spot of light on the wall, is asked “What are you doing making that light come on the wall?” and replies “Ah yes, it’s opening the window that does it,” or “That always happens when one opens that window at midday when the sun is shining” (knowledge of the action only by observation). 11 Anscombe (1963), iii.

Practical Knowledge and the Structure of Action

139

insists that, temptation to proliferate senses notwithstanding, “it is implausible to say that the word is equivocal as it occurs in these different cases”12. And so her claim, that if someone is acting intentionally, he knows what he is intentionally doing, can presumably be extended to the other two “heads”: an expression of intention for the future (e. g. ‘I am going to do A’) will be an expression of knowledge about the future (i. e. ‘I am going to do A’); and if someone is doing A with the intention of doing B, then he will know, not only that he is doing A, but that he is doing A with the intention of doing B, and thus that he is thereby doing, or going to do, B. In both these cases, too, the knowledge will be “non-observational”, and though it is less clear here what the corresponding observational knowledge would be, our examples have brought out that the kind of knowledge of action in progress or in prospect that Anscombe means to exclude is not observational knowledge narrowly construed (that is, perceptual knowledge), but knowledge based on any kind of evidence whatsoever. The idea of a kind of knowledge not based on any evidence—indeed, incompatible with being based on evidence—might already sound suspicious. Perhaps one might cautiously admit the possibility of “groundless self-knowledge”; Anscombe’s claim, however, pertains not to knowledge of a mental state or event, but to the very transpiring of a public, material event: for, if you are making an omelette, then presumably an omelette is being made, and so knowing that you are making an omelette presumably includes knowing that an omelette is being made. It’s one thing to think that there might be groundless knowledge of the mind—but groundless knowledge of omelettes? However, to worry about the details and consequences of Anscombe’s claim at this point would be to get ahead of ourselves. Let’s call the condition Anscombe identifies—that if an agent is doing something intentionally, he knows without observation that he is doing it— the cognition condition. We said that her claim, that the cognition condition is a necessary condition on intentional action, is intuitive. We did not yet say it is true. 12 Anscombe (1963), §1. On the question of the univocality of ‘intention’, philosophers disagree: while Kieran Setiya claims that “The principal task of the philosophy of intention is to uncover and describe the unity of the three forms” (Setiya (2009a)), David Velleman “giv[es] up, once and for all, on the project of finding a unified analysis for the expressions in which [the words ‘intention’ and ‘intend’] appear” (Velleman (1989), 112 f.).

140

Will Small

1.2 Davidson’s Objection Donald Davidson thinks that it is not. He thinks that if an agent is doing A intentionally, he need not even believe—let alone know, let alone know without observation—that he is doing it. His well-known counterexample to Anscombe’s claim is this: “A man may be making ten carbon copies as he writes, and this may be intentional; yet he may not know that he is; all he knows is that he is trying”13. The core of the case—the idea of someone trying to do something that he doubts he will pull off and yet succeeding—can easily be adapted to target the corresponding Anscombean claims about the intention with which someone acts, and intention for the future. Davidson later deployed a slight variation of his carbon-copier case that targets the former: “in writing heavily on this page I may be intending to produce ten legible carbon copies. I do not know, or believe with any confidence, that I am succeeding”14. And Michael Bratman presents the following case against the latter: “I might intend now to stop at the bookstore on the way home while knowing of my tendency towards absentmindedness…. If I were to reflect on the matter I would be agnostic about my stopping there, for I know I may well forget”15. Philosophers disagree as to the force of carbon-copier style counterexamples. Bratman is very impressed by them; he concludes that “such examples are worrisome enough that [one] would do well to develop [one’s] account of intentions and plans in a way that does not require the strong assumption that to intend to A I must believe I will A”16. Kieran Setiya too thinks that these examples pose a difficulty for Anscombe’s claim, though he is less impressed by them than Bratman; he thinks they are “exceptions to [the] rule [that] serve only to confirm the basic insight,” and thus that “the challenge to Anscombe is limited”17. Davidson occupies an interesting position. While he holds that intentions are not, and do not include or entail, beliefs, he clearly does not think that agent’s knowledge is irrelevant to action: shortly after presenting the original carbon-copier case in ‘Agency’ he goes on to add that “[a]ction does require that what the agent does is intentional under some description, 13 14 15 16 17

Davidson (1971), 50. Davidson (1978), 92. Bratman (1987), 37. Bratman (1987), 38. Setiya (2007), 25.

Practical Knowledge and the Structure of Action

141

and this in turn requires…that what the agent does is known to him under some description”18. But despite these differences, there is some consensus: that, as Setiya puts it, “Anscombe’s doctrine is false, at least when it is unqualified”19. The idea, then, is that while Anscombe has suggestively pointed towards a putative requirement on intention and intentional action, she seems at best to have overstated it. Now, we could weaken Anscombe’s cognition condition by weakening either the kind, the content, or the scope of the relevant cognition (or some combination of the three). With respect to the kind of cognition, we might—out of “modesty,” as Setiya puts it—“[set] aside the claim to knowledge as well as the claim to knowledge without observation”20, and cast the cognition condition in terms of what the agent believes. This proposal has the advantage of postponing the inevitable difficulties that face accounting for the truth of the agent’s belief that he will in fact do what he intends to do, the justification with which the agent believes that he is doing, or will do, what he intends, and anything else that might be required for “agent’s belief” to amount to agent’s knowledge. In the light of Davidson’s and Bratman’s counterexamples, however, we might want to modify the content of the cognition. One possibility, suggested by Davidson’s initial presentation of the carbon-copier case, would be that if one is intentionally doing A, one knows (or believes) that one is trying to do A. 21 Setiya makes a different suggestion about the cognition’s content: if one is doing A intentionally, “there must be something he is doing intentionally, not merely trying to do, 18 Davidson (1971), 50 emphasis added. 19 Setiya (2007), 25. While Velleman thinks that carbon-copier cases fail in their aim, he too thinks Anscombe’s view must be qualified—indeed, in a way that inverts Davidson’s position: Velleman thinks that intentions are beliefs about the future, but that intentional action needn’t involve spontaneous agent’s knowledge. See e. g. Velleman (1989), 112 – 117; Velleman (2007), xv, n.7. 20 Setiya (2007), 25. 21 I hesitate to attribute the view to Davidson himself; in ‘Freedom to Act’, he writes: “[Armstrong] asks the question what we must add to ‘A tried to do x’ in order to have necessary and sufficient conditions for ‘A did x intentionally.’ Most of us would say—I certainly would—that trying itself isn’t necessary in many cases…” (Davidson (1973), 78). It is clear that Davidson thinks an agent must know some, but not all, of the descriptions under which his action is intention; it is less clear how to figure out which descriptions, if any, Davidson thinks are ones the agent must know, or if he envisages any principle that accounts for this.

142

Will Small

in the belief that he is doing it”22. The “something” need be A itself only if one’s doing A is a “basic action”, which making ten carbon copies does not appear to be—the copier may be making the copies intentionally while believing only e. g. that he is writing firmly on the top sheet. An alternative possibility for accommodating the counterexamples would be to limit the scope of the cognition condition. John Gibbons thinks that the cognition condition need not be a necessary condition on intentional action. This is not simply because “we shouldn’t expect to find such a thing, given the history of [such conditions’] failure,” but rather because the important questions about “privileged access”— whether to our intentional actions or to anything else—are “what kind of access, how it works, and why we have it”23. Crucially, “we shouldn’t expect an account of how you know when things go smoothly to deliver a non-trivial answer to the question of when things will”24—things don’t go smoothly for the carbon copier, but such cases are outliers that don’t impugn the account, providing it is properly understood. Similar permutations (kind, content, scope) arise when considering whether there is a cognition condition on intention for the future, and what form it takes if there is one.25 According to the account developed in this essay, all such weakening and finessing is a mistake. Once we have uncovered the real ground and meaning of the cognition condition, which lies in the calculative and temporal structure of intentional action, we will see that the carbon-copier example poses no threat, because it trades on confusion as to how to understand the possibility of failing to act as one had intended.

22 23 24 25

Setiya (2007), 26. Gibbons (2010), 84. Gibbons (2010), 84. It is worth noting, however, that while even Davidson accepts a weak version of the cognition condition on intentional action, intention for the future can appear to be yet further removed from cognition: Bratman’s bookstore case supposedly exemplifies intention not merely in the absence of belief in success, but in the absence of the belief that one will even try. But this is not necessary; it is easy to construct an example that simply transposes the features of Davidson’s carbon-copier example to intention for the future (indeed, Bratman himself does so). See Bratman (1987), 35 – 42.

Practical Knowledge and the Structure of Action

143

1.3 The Contemporary Debate and Beyond Before proceeding to my defence of the cognition condition, it is worth mentioning the way in which contemporary accounts of action and intention deal with the issue. A helpful, if somewhat coarse, division of the field yields two types of approach: cognitivism and non-cognitivism about intention. Cognitivists, like Velleman and Setiya, are primarily impressed by the fact that, like belief, an expression of intention, such as ‘I am walking to school’ or ‘I am going to walk to school’, makes a claim not merely about the speaker’s state of mind, but about what is actually happening, or going to happen, in the world. Finding the similarity overwhelming, cognitivists take the bold course of identifying intention with belief—albeit belief of a special kind. What is special about an intention is that, unlike a “regular” belief about the future, it represents, and in the good case represents correctly, the future as causally dependent on the belief.26 On the other hand, non-cognitivists, like Davidson and Bratman, reject the assimilation of intention to belief, which they consider to be a theoretical attitude. Instead, they insist that intention is a special kind of “pro-attitude”, a distinctively practical attitude that differs from mere desire by representing one’s course of action as “settled”; whereas mere desires only potentially influence conduct, intentions actually control it. As we saw above, Bratman takes carbon-copier-style counterexamples to decide the issue in favour of non-cognitivism. But in any case, if intention is a pro-attitude, then as such it has a different direction of fit from any kind of cognition: it must represent the world not as it is, but as it would in some sense be good to be, or it must at least consist in its bearer’s being disposed to act so as to bring the world into alignment with the representation. Thus, according to non-cognitivism, what is practical cannot be knowledge, and what is knowledge cannot be practical (cf. the epigraph by McGinn).27 26 Weaker forms of cognitivism are possible: the content of the belief or the scope of the condition may be weakened (see the discussion in §1.2 above); alternatively, one might hold—with e. g. Grice (1971)—that an intention is a hybrid state that contains belief as an element, or that intending to do A entails believing that one will do A, and so on. 27 See Bratman (1987), 15 – 20; cf. the closing pages of Davidson (1978), where Davidson distinguishes between desires, conceived of as expressed by “prima facie judgements”, and intentions, conceived of as expressed by “all-out judgements”. While Davidson situates both desires and intentions within “the genus of pro attitudes expressed by value judgements”, (Davidson (1978), 102), it is

144

Will Small

From the cognitivist point of view, however, this reflects a lack of perception and imagination. Intention and practical knowledge need not aspire to possessing contrary directions of fit at the same time. Rather, according to Velleman, intention has “the same direction of fit as belief but the same direction of guidance as desire,” where direction of guidance “consists in whether the attitude causes or is caused by what it represents”; thus, in virtue of combining the cognitive quality of belief with the practicality—the causal efficacy—of desire, intention is “a case of practical cognition”.28 In the context of the contemporary debate between cognitivist and non-cognitivist conceptions of intention, Anscombe’s position, with its emphasis on practical knowledge (as that which is “the cause of what it understands”), is easily viewed as a particularly strong version of cognitivism. But we might well wonder whether this is what Anscombe has in mind, given that by “practical knowledge” she is widely acknowledged to have meant more than merely the following: a cognitive mental state that is the efficient cause of intentional action.29

worth noting that Bratman explicitly disavows the “tendency to see intention as either a special kind of evaluation or a special kind of belief” (Bratman (1987), 110). I am inclined to think that the deepest motivation for a non-cognitivist view lies in the traditional thought that practical reasoning is distinguished from theoretical reasoning in taking as its formal object the good rather than the true. Unfortunately, a discussion of such a non-cognitivism, which in any case is not Bratman’s, lies beyond the scope of this essay. 28 Velleman (2000), 25. On Velleman’s view it is crucial not only that an intention is the cause of what it represents, but also that it represents itself as that cause; cf. Velleman (1989), 102; Velleman (2000), 25. In a recent essay, Setiya retains the idea that intention consists in a kind of causally-efficacious belief, but distances himself from the idea that it represents itself as the cause of the intended action (Setiya (2011), 194 n.65). 29 In §33 of Intention, Anscombe claims that we cannot make sense of practical knowledge without making sense of practical reasoning; she goes on to disparage the idea that practical reasoning is just theoretical reasoning about a special topic (i. e. action) or involving special concepts (e. g. ethical concepts), concluding that there is “a difference of form between reasoning leading to action and reasoning for the truth of a conclusion. …[T]he conclusion [of practical reasoning] is an action whose point is shewn by the premises” (Anscombe (1963), §33). The thesis that Anscombe is rightly understood as claiming that practical knowledge is the formal cause of what it understands, rather than merely its efficient cause, is elaborated and defended at length in Moran (2004); Velleman— who clearly hears “cause” only in its dominant, “efficient” sense—“doubt[s] whether Anscombe takes the word ‘cause’ seriously” on the ground that

Practical Knowledge and the Structure of Action

145

On one traditional conception of things, theoretical reason is the power of theoretical knowledge, practical reason the power of practical knowledge.30 And we might also go on to say, adopting the old terminology, that, on the one hand, a belief is an act of theoretical reason, the power of theoretical knowledge, such that if a belief is a proper, unimpeded exercise of that power, it is a case of theoretical knowledge; whereas on the other hand, an intention is an act of practical reason, the power of practical knowledge, such that if an intention is a proper, unimpeded exercise of that power, it is a case of practical knowledge. Anscombe says that practical knowledge is knowledge “in intention”31; so, following this line of thought, we might say that theoretical knowledge is knowledge “in belief”. Thus, in seeking to locate a possibly defensible kernel of her claim in the idea that a necessary condition on intention and intentional action is that an agent has certain beliefs, we run the risk of fundamentally misconstruing her view with an unwittingly prejudiced application of the principle of charity. Of course, a hasty sketch of long-forgotten faculty psychology and the quite possibly dubious invocation of some scholastic terminology proves nothing in and of itself. I bring it up only to raise suspicion of the quick and ready assimilation of Anscombe’s view to contemporary cognitivism about intention and to suggest an alternative strategy, both for interpreting Anscombe and for offering an account of agent’s knowledge. According to the broadly Anscombean view I develop in §§2 – 4, there is a cognition condition on intention and intentional action that is to be explicated not in terms of belief, but rather in terms of the cognitive character of intention itself—in terms, that is, of practical knowledge. (This explains why I speak of a cognition condition rather than a belief condition: the label is intended to be neutral not merely between knowledge and something less than that—belief, on the standard conception—but between the different types of “acts of reason” in which “[t]he connection that she posits between the agent’s knowledge and his action seems conceptual rather than causal” (Velleman (1989), 102, n. 24). 30 This traditional conception is found at least as late as Kant, according to Stephen Engstrom (2009), who emphasizes the continuity of the view as it appears in Kant with Aristotle and medieval philosophy; these latter, of course, are those to whom Anscombe credits, in the remark that appears as the epigraph to this essay, the concept of practical knowledge—“blankly misunderstood” by modern philosophy—which she seeks to rehabilitate (Anscombe (1963), §32). 31 Anscombe (1963), §32.

146

Will Small

knowledge might “inhere”.) The account of practical knowledge to be advanced here is based not on considerations pertaining to the admissible combinations of directions of fit and guidance that a given type of mental state might possess, but rather on figuring out how intentional action could have the structure that it has: practical knowledge is practical because it is the source of the calculative and temporal structure without which what happens would not be a case of intentional action. I propose to approach the connection between practical knowledge and the structure of action obliquely, via an initially obscure distinction Anscombe draws between an intention’s being contradicted and its “falling to the ground”: …the contradiction of ‘I’m replenishing the house water-supply’ is not ‘You aren’t, since there is a hole in the pipe’, but ‘Oh, no, you aren’t’ said by someone who thereupon sets out e. g. to make a hole in the pipe with a pick-axe. And similarly, if a person says ‘I am going to bed at midnight’ the contradiction of this is not: ‘You won’t, for you never keep such resolutions’ but ‘You won’t, for I am going to stop you’.32 Now suppose what he says [i.e. ‘I’m replenishing the house water-supply’] is not true. It may be untrue because, unknown to the agent, something is not the case which would have to be the case in order for his statement to be true; as when, unknown to the man pumping, there was a hole in the pipe round the corner. …[If this is the case], we may say that in face of it his statement falls to the ground, …but it is not a direct contradiction.33

These passages suggest that ‘You aren’t, since there is a hole in the pipe’ and ‘You won’t, for you never keep such resolutions’ are on a par with one another, and thus that if in the face of the former the intention (in action) to replenish the house water-supply falls to the ground, then so in the face of the latter the intention (for the future) to go to bed at midnight will likewise fall to the ground. But this seems faintly absurd. Maybe I do never keep such resolutions, but there’s a first time for everything, and so long as you’re not going to stop me, why shouldn’t I intend to go to bed at midnight? And for that matter, if I’m employing you to replenish the house water-supply and find, at the end of the day, that you gave up your task, and even stopped intending to carry it out, in the face of a simple hole in the pipe—well, I’ve every right to be annoyed: ‘Why didn’t you just plug it?’, I might say. Perhaps, though, the pipe is well and truly broken; it can’t be repaired. If it’s impossible to re32 Anscombe, (1963), §31. 33 Anscombe (1963), §32, emphasis added.

Practical Knowledge and the Structure of Action

147

plenish the water supply, surely this makes sense of the idea that the intention to do so “falls to the ground”. But on reflection, this is just as much of a dodge as before, and you won’t get the afternoon off work so easily! The pipe can’t be repaired, so the intention to replenish the water-supply by pumping water through the pipe cannot be executed, and thus falls to the ground; but the inhabitants need running water and there are two buckets over there—get to it! There may be many different ways to replenish the water supply; if so, there will be many ways of filling in one’s intention to do so. After all, if we gave up on everything at the first sign of trouble, we’d never get anything done. We should also note something else. If you spent the morning pumping fruitlessly, as it turned out, and the afternoon lugging buckets of water up to the house, then what you were doing all day was: replenishing the water supply. Of course, there are other ways of describing what you did with your day, notably pumping fruitlessly and then lugging, but replenishing the water supply applies as well. (‘Why didn’t you cut the grass as you were supposed to?’—‘Because it took all day to replenish the water supply’.) After all, as Michael Thompson points out, …we happily affirm, of someone who is napping, that she is organizing the peasantry; of someone who is sitting reading the paper, that she is baking a loaf of bread; and of someone who is playing a hand of poker, that she is building a house. If confusion arises, we…concede that our agents aren’t baking or building or organizing at the moment or right now, but rather reading, playing poker or napping.34

Thompson’s examples highlight an important feature of progressive judgments—judgments, that is, of the form ‘S is (was) doing B’—namely, their “broadness”.35 It might take me ages to get to your house for our rendezvous because I take a wrong turn, which eventually leads me back to my house, whence this time I take the correct route to your house. It would be mere pedantry on your part to insist that it didn’t take me ages to get to your house, that I just left late. So it might well be that the pumper’s intention to replenish the water supply falls to the ground in the face of the hole in the pipe. But he can pick it back up again. And if he does, his statement, ‘I am replenishing the water supply’, will not be untrue. Notice the difference between the following two relationships: on the one hand, that between the intention to replenish the water supply 34 Thompson (2008), 141. 35 On the broadness of the progressive, see also Galton (1984) and Falvey (2000).

148

Will Small

and the fact that there’s a hole in the pipe, and on the other, that between the belief that the pipe has no holes and the fact that there’s a hole in the pipe. If our pumper finds out that there’s a hole in the pipe, he must give up, on pain of irrationality, the belief that it has no holes; but he can keep the intention to replenish the water supply, as long as he has, or can come up with, a “Plan B” for its execution. But now compare the case of opposed intentions: X says ‘I’m replenishing the water supply’ and Y says ‘No you’re not—I’m going to stop you’. X’s intention to replenish the water supply cannot be retained as easily in the face of Y’s threat as it could in the face of the hole in the pipe. The problem with the hole in the pipe is that it is preventing X from replenishing the water supply by means of operating the pump. The hole in the pipe is not going to prevent X from trudging up to the house with buckets of water from the well. By contrast, Y’s intention to stop X from replenishing the water supply does track X’s intention through Plans B, C, and so on. Y’s intention to prevent the water supply’s replenishment is an intention to act, to do various things—make a hole in the pipe, throw the buckets down the well, tie X up, and so on—in order to stop X from doing the various things he needs to do in order to replenish the water supply.36 X’s intention to replenish the water supply can be made consistent with the fact that there is a hole in the pipe, whereas X’s belief that the pipe has no holes cannot be made consistent with that fact, and X’s and Y’s intentions cannot both be successfully executed. These differences illuminate a famous but cryptic remark of Anscombe’s, which appears as an epigraph to this essay, in which she complains of modern philosophy’s “incorrigibly contemplative conception of knowledge,” according to which “the facts, reality, are prior” to the knowledge, which is “judged as such by being in accordance with” those facts.37 This is how things are with the belief that the pipe has no holes and the hole in the pipe. The belief is false because of how things stand with the pipe, which is prior to the belief. But X’s expression of intention “I am replenishing the water supply” is not to be judged as false simply because there is a hole in the pipe. In the face of the hole in the pipe, the intention falls to the ground. Whether it is picked up or not depends on what X does, on whether he seeks out alternative means or gives up. If he seeks out and takes alternative means, his expression of intention is true. 36 Cf. Moran/Stone (2009), 155 – 156. 37 Anscombe (1963), §32.

Practical Knowledge and the Structure of Action

149

If he has given up, then presumably he no longer has the intention, and thus wouldn’t express it. But suppose he were to continue to give (linguistic) expression to his intention, while doing nothing whatsoever to execute it; here we can say that his utterance is false—he is not replenishing the water supply. However, the assessment of what he says as false does not depend on a failure of fit between what he says and “the facts” (i. e. the hole in the pipe), but rather on a failure of fit between what he says and what he does: he says he is replenishing the water supply, but he is doing nothing to execute his intention (and he’s not just taking lunch). For his expression of intention to amount to knowledge, it must be true that he is replenishing the water supply. (No knowledge without truth.) But whether or not he is replenishing the water supply is not something that can be ascertained independently of his knowledge of what he is doing and the means by which he is doing it.38 The purpose of the foregoing considerations was to begin elucidating some ways in which intention and practical knowledge might differ from belief and theoretical knowledge; they have uncovered, albeit haphazardly, the two structures of intentional action of which we need a more detailed and systematic investigation: (i) the calculative structure of action—the structure involved in, intentionally, doing one thing by doing another thing, and (ii) the temporal structure of action—the relations that hold between what an agent is doing, what he is going to do, and what he ends up having done. The grounds for thinking that there is a cognition condition on intention and intentional action go deeper than noticing that agents can say what they’re doing without having to look, and that when they say what they’re doing or going to do, 38 By contrast, a cognitivist conception of intention like Velleman’s, which divides the practical and cognitive aspects of practical knowledge into two factors, characterized by a cognitive direction of fit and a practical direction of guidance, must surely treat “the facts, reality, as prior” for the purposes of assessing whether or not an expression of intention amounts to knowledge. As cognitive, on this conception of what it is to be so, intention strives to represent what is true independently of its being so cognitively represented; that the intention is causally responsible for the truth of what it represents—that it has “practical direction of guidance”—is neither here nor there from the perspective of its cognitive direction of fit. Even if the content of the intention represents the agent as, say, pumping-because-of-this-very-intention, then though the intention will be true only if the agent is pumping as a result of his so intending, from the standpoint of assessing whether what the agent says, when he says what he is doing, is true, the practicality of the intention figures not as the way of knowing, but only as part of the content of what is known.

150

Will Small

their utterances are apparently indicative; the grounds lie in the structure of intentional action itself.

2 Doing One Thing by Doing Another I will argue that the source of the cognition condition lies in the calculative structure of intentional action—the structure of means and ends represented as such by the agent that is internal to intentional actions. By contrast, Davidson and his followers understand the structure of intentional action in terms of the causal relations that an action, which itself is internally structureless, bears to other events, in virtue of which relations there is structure in action descriptions. If the source of the cognition condition lies in the inner structure of intentional action, as I will argue that it does, and if this structure is missing in the Davidsonian account, then it will be no surprise that the cognition condition cannot be adequately comprehended within the Davidsonian framework that still dominates contemporary philosophical approaches to action. Thus it is crucial for my subsequent account of the cognition condition that the Anscombean and Davidsonian conceptions of the structure of action are accurately distinguished. I begin with some common ground between Anscombe and Davidson, concerning the individuation of actions. Suppose a man raised a paddle, and thereby lodged a bid and cast a shadow on his neighbour. He did three things; let us suppose he did the first two intentionally, the third unintentionally. But how many actions did he perform? Anscombe, Davidson, and the many who have followed them, say one: in saying he did three things, we mean that there are three descriptions that truly apply to his action. (The other answer, which was advanced by Alvin Goldman (1970) but has few adherents these days, is of course three.) 39 Clearly, the example I provided is grossly simplified: there are 39 The Anscombe–Davidson line on this is, I think, obviously correct; but it is easy to overstate: suppose a man slices some bread, then butters it, then slices some cheese, assembles these ingredients, and thereby makes a cheese sandwich. His slicing of the bread isn’t the same action as his slicing of the cheese, and neither slicing is identical with his making of a cheese sandwich. The two slicings, the buttering, and the assembling amount to four actions; but the making of a sandwich is not a fifth (as Goldman would have it): rather, the making of the sandwich is identical with the sum of the four sub-actions or parts. Quite different relations are denoted by ‘…and then…’ and ‘…and thereby…’; the

Practical Knowledge and the Structure of Action

151

indefinitely many descriptions that truly apply to any action, most of which will be quite recherch and of no interest to anyone. But among the chaos of these indefinitely many descriptions, there is order to be found, and Anscombe and Davidson each try to find it. Or rather, I should say, there are orders to be found.

2.1 Davidson and the Accordion-Effect It follows from the Anscombe–Davidson account of action-individuation that while the concept action can delimit a class of events, the concept intentional cannot delimit a class of actions. The action in our example was intentional under the descriptions “raised a paddle” and “lodged a bid” and unintentional under the description “cast a shadow on his neighbour”. An event belongs to the class of actions if it is intentional under some description, and we have at least one description under which it is intentional, so the event described in the example is an action. But, if we were to ask whether “the action itself” belonged to the class of intentional actions, we would have to say both yes and no, which is absurd, showing that the question makes no sense and that there can be no such class. In his well-known paper ‘Agency’, Davidson concludes from the fact that the same action—the same event—can be intentional under one description and unintentional under another that “although the criterion of agency [i.e. being intentional under a description] is [intensional], the expression of agency is itself purely extensional.” This suggests to his Quinean side that “the concept of agency is simpler or more basic than that of intention”; thus he looks for “a mark of agency that does not use the concept of intention,” hoping that “[t]he notion of cause may provide the clue”40. This is the context in which Davidson provides his most in-depth analysis of the orders that can be discerned in the welter of descriptions of an action. The clue Davidson finds in the notion of cause is this: actions can be described in terms of their effects, and there will be an order to these descriptions determined by the causal chains running out from the acAnscombe–Davidson line is appropriate for ‘…and thereby…’, but we must take care when thinking about ‘…and then…’. See my discussion of the part–whole relation in §2.2 below. 40 Davidson (1971), 47.

152

Will Small

tion. Roughly speaking, he thinks that what it means to say that S did B by doing A is that S did A, and S’s doing of A caused an event in virtue of which S’s doing of A may be legitimately redescribed as S’s doing of B. 41 Now, S’s doing of A might have caused the event that licensed that redescription by causing any number of intermediate events. According to what Davidson (following Joel Feinberg) calls “the accordion effect”, we can leave out links of the causal chain; as he puts it, “[t]here are…a great many tunes we can play on the accordion”42. Thus if the queen killed the king by poisoning him by emptying a vial of poison into his ear by moving her hand, then it would be true—if opaque—to say that she killed him by moving her hand. The chain must have a beginning: “[n]ot every event we attribute to an agent can be explained as caused by another event of which he is the agent”43. Thus there must be basic actions, of which Davidson famously said: “our primitive actions, the ones we do not do by doing something else, mere movements of the body—these are all the actions there are. We never do more than our bodies: the rest is up to nature”44. On Davidson’s view, then, all actions are basic actions, bodily movements with no intrinsic structure qua actions. It follows that an action-description may very well be complex—e. g. ‘killing the king’ means ‘doing something which caused (something which caused something which caused something which caused…) the king’s death’. But the action it describes is simple; it’s the first event in the chain. Complex descriptions apply to actions, but only in virtue of causal relations that hold between an action and other events it causes—relations extrinsic to the action itself.45 The action itself, under its most primitive description, is a bodily movement, which goes unanalyzed by Davidson as a simple element of his theory. The causal interpretation of the ‘by’-relation cannot, however, be exhaustive, for ‘by’ may denote a constitutive connection, too.46 In such cases, S’s doing of A is his doing of B not in virtue of causing another event that licenses redescribing his doing of A as his doing of B, 41 For example, Davidson says that ‘The queen killed the king by pouring poison in his ear’ and ‘The queen poured poison in the king’s ear thus causing his death’ “are equivalent” (Davidson (1971), 59). 42 Davidson (1971), 58. 43 Davidson (1971), 49. 44 Davidson (1971), 59. 45 Cf. Davidson (1985), 301. 46 As Davidson concedes, his analysis “works for some cases only” (Davidson (1971), 55).

Practical Knowledge and the Structure of Action

153

but rather in virtue of conventional, circumstantial, or metaphysical facts concerning the A—B connection that license the redescription. For instance, a man may vote by raising his hand (given certain conventions), take a giant leap for mankind by taking a small step for man (given special circumstances), or get some Vitamin X in him by eating pig’s tripes (in virtue of the latter containing the former). The ‘by’-relation is determinable, and can be determined in these ways.47 Like the causal determination of the ‘by’-relation, each of these kinds of determination can be instantiated by an agent intentionally, unintentionally, or even unknowingly. So even if we enrich Davidson’s account by adding to it these other kinds of ‘by’-relation, it seems that the structure of action—the order of actions and descriptions of action—still concerns things done (and connections between them) that may be unintentional and unknown, and thus that investigating the structure of action has no bearing on the question of whether there is a cognition condition on intentional action; such is the price of an extensional analysis.48

2.2 The Calculative Structure of Intentional Action Though both Davidson and Anscombe investigate what structure there is to be discerned in the descriptions that apply to an action, Davidson does so by locating action in a structure, while Anscombe locates structure in an action. The structure that she finds there is calculative, one of 47 Other versions of the metaphysical constitutive connection: one may tie a knot by tying a reef knot, which is a kind of knot, or cook fish by frying it, which is a way of cooking fish. On the variety of determinations of the ‘by’-relation, see Baier (1970), 648 f., Baier (1971), and Müller (1979), 106. 48 I should note that Davidson himself considers his analysis of agency successful only with heavy qualifications. Yet he thinks that it leads “to a vast simplification of the problem of agency, for it shows that there is a relation between a person and an event, when it is his action, that [because all actions are primitive] is independent of how the terms of the relation are described” (Davidson (1971), 61). Of course, the analysis of the relation between an agent and his action will depend on intention, but the concept of intention has been successfully expelled from the analysis of the structure of the action itself. Consequently, a case for a cognition condition would have to be made on other grounds than considerations arising from the structure of action; and the most promising grounds would seem to be psychological. This would be a reason to turn away from the question of a cognition condition on intentional action, towards the question whether intention involves belief.

154

Will Small

means and ends represented as such by the agent. The calculative ‘by’relation may be instantiated by any of the connections we have encountered; however, it is not instantiated by those connections insofar as they are not represented as means to end by the agent—that is, insofar as they are not intentional. To our list of determinations of the ‘by’-relation we may now also add the part–whole relation: one may do C by doing B’ and then doing B’’ and then…. Unlike our other relations, there is more to doing C than doing B—more, that is, for the agent to do. Davidson’s system is essentially hostile to this relation, in a way in which it is not to the other additions: the part–whole relation brings out that actions may be composed of other actions. The relationship between an action and a proper part of it (itself an action) cannot be cast in terms of two descriptions of the very same action. Thus this relation captures what Davidson is blind to: the fact that actions themselves—not just their descriptions—are structured.49 Accordingly, in the remainder of §2, I will explicate key aspects of the calculative structure of intentional action, in order to reveal, gradually, the way in which it is constituted by the agent’s practical thought. We will see that an intentional action is articulated into means and ends, themselves intentional actions, where this articulation is revealed by asking the agent why (for what purpose) and how (by what means) he is doing what he’s doing (§2.2), and further specified by practical reasoning in order to sustain and/or repair it (§2.3). In §2.4, I’ll argue for a kind of “calculative holism”, according to which the elements of the calculative structure can be apprehended as intentional actions only by appreciating their role in the structure of which they are elements, the structure constituted and sustained by the agent’s practical thought. These remarks will permit a more detailed formulation of the cognition condition on intentional action, according to which if an agent is doing something intentionally, he has knowledge in intention of what he is doing, and of how and why he is doing it (§2.5). I will then go on, in §3, to 49 Note that the part–whole ‘by’-relation too has non-intentional instances; only its intentional instances, of course, instantiate the calculative relation. In his later paper ‘Intending’, Davidson does consider an example of an action (writing the word ‘action’) “[s]ome temporal segments of [which] are themselves actions: for example, first I write the letter ‘a’” (Davidson (1978), 88). But the incompatibility of such cases with the doctrine of ‘Agency’ passes without mention, and a positive account of how the parts are unified in the whole is not provided.

Practical Knowledge and the Structure of Action

155

extend this conception of the cognition condition to intention for the future. We can illustrate the calculative structure of intentional action with Anscombe’s famous example, a portion of which I exploited earlier, in which a man is moving his arm (doing A), operating a pump (doing B), replenishing the house water-supply (doing C), and poisoning the inhabitants (doing D). These things are connected as a series of means to his end, poisoning the inhabitants: he is poisoning the inhabitants by replenishing the water supply, which he’s doing by operating a pump by moving his arm. Thus we can move down the ‘by’-chain from D to A by asking ‘How?’, where the sense of the question asks for the agent’s means: ‘How are you doing D?’—‘By doing C’; ‘How are you doing C?’—‘By doing B’; ‘How are you doing B?’— ‘By doing A’. Conversely, we can move up the chain from A to D by asking the corresponding question ‘Why?’, which asks for the agent’s reason for acting: ‘Why are you doing A?’—‘In order to do B’; ‘Why are you doing B?’—‘In order to do C’; ‘Why are you doing C?’—‘In order to do D’. It may be surprising to learn that Anscombe’s question ‘Why?’, which she spends so much of Intention elucidating, corresponds to the question ‘How?’, which receives comparatively little overt attention.50 In fact, however, it is simply another way of putting one of her central points, which is not explicitly made until towards the end of the book, that the order of practical reasoning—the order of calculation from “the thing wanted” to “the immediate action”51, the order, that is, of how to achieve one’s end—and the order of action explanation—the order defined by the question ‘Why?’—are “the same order”52. In light of this, if we say that the expression ‘In order to do…’ gives the intention with which the agent acts, we can say that the man is moving his arm (doing A) intentionally with the intention of operating the pump (doing B). And indeed, reminiscently of the accordion-effect, we can say, if obscurely, that he is moving his arm with the intention of poisoning the inhabitants of the house, or that he is poisoning the inhabitants by moving his arm.53

50 51 52 53

Anscombe Anscombe Anscombe Anscombe

notes the correspondence briefly at Anscombe (1963), §26. (1963), §41. (1963), §42. (1963), §26.

156

Will Small

Now, it might be thought that the “operations” performed on the A–D order by the questions ‘How?’ and ‘Why?’ are not symmetrical, because whereas both the question ‘How are you doing B?’ and the answer ‘By doing A’ mention doings, when the question ‘Why?’ is asked of doing A, the response ‘In order to do B’ apparently does not mention a doing—in other words, it doesn’t say that B is being done. However, Anscombe points out that, if the agent is doing A intentionally with the intention of doing B, then he is doing B intentionally, provided two conditions are met54 : he is in fact doing B—call this the Realizing Condition (in doing A the agent is realizing the concept do B), and his doing B is not an accidental consequence of his doing A—call this the Non-Accidentality Condition. When both conditions are met, the symmetry of ‘Why?’ and ‘How?’ is secured. That is, the same calculative nexus of doing A and doing B is captured, from different angles, so to speak, by the following exchanges: ‘Why are you doing A?’—‘Because I’m doing B’, and, ‘How are you doing B?’—‘By doing A’. Employing the calculative senses of ‘by’ and ‘because’ just elucidated, we can say that ‘I am doing A because I am doing B’ and ‘I am doing B by doing A’ are but different expressions of the same thought, as are ‘At Plataea the Greeks defeated the Persians’ and ‘At Plataea the Persians were defeated by the Greeks’.

2.3 Doing A in Order to Do B, but not Doing B Anscombe claims that if an agent is doing A intentionally with the intention of doing B, he is doing B intentionally, as long as the Realizing and Non-Accidentality Conditions are met. The Realizing Condition appears obvious: the agent must actually be doing B. Now, one way in which an agent can be doing A in order to do B, but not be doing B, is if he is not doing B yet (buying eggs in order to make an omelette, for example). However, Anscombe claims that the distinction between, on the one hand, cases of which we will happily grant that, in doing A, an agent is (actually) doing B, and, on the other, cases of which we will say only that, in doing A, he (merely) wants or intends or is going to do B, is not a sharp one:

54 Anscombe (1963), §23.

Practical Knowledge and the Structure of Action

157

…is there much to choose between ‘She is making tea [by putting on the kettle]’ and ‘She is putting on the kettle in order to make tea’—i. e. ‘She is going to make tea’? Obviously not. And hence the common use of the present to describe a future action which is by no means just a later stage in activity which has a name as a single whole. E.g. ‘I am seeing my dentist’, ‘He is demonstrating in Trafalgar Square’ (either might be said when someone is at the moment e. g. travelling in a train).55

I will discuss the question of what it is to be doing B in doing A—that is, the questions whether and how the achievement of the objective (doing B) can be, and be known to be, somehow “present” in what is happening (doing A)—in some detail in §4 below. For now, I will consider cases in which an agent is doing A in order to do B, but not doing B, where this is not simply because he cannot be said to be doing B yet, but rather because something is going wrong with his attempt to be doing B now. Earlier, we considered the case in which the pumper is intentionally pumping in order to replenish the water supply, but he is not replenishing the water supply because of a hole in the pipe. We said that his intention to replenish the water supply by pumping thus “falls to the ground”; pumping, in these circumstances, is in fact not a way in which to replenish the water supply—it’s not a part or species of replenishing the water supply, it doesn’t constitute replenishing the water supply in virtue of any special circumstances or conventions, and it’s not a cause of the water supply’s being replenished. We also said that the pumper could “pick up” his intention to replenish the water supply: either by doing something in order to restore the possibility of pumping-inorder-to-replenish-the-water-supply (e. g. fixing the hole in the pipe), or by taking alternative means to replenishing the water supply. However, might not the Realizing Condition go unsatisfied in a different way? Suppose there is no hole in the pipe; there is no problem with the pumper’s intention to replenish the water supply by means of pumping. And indeed, he is pumping (operating the pump) intentionally. But suppose that his stroke is too shallow; it doesn’t draw the water, and in fact he is only pushing air through the pipe. In this case too, he is not replenishing the water supply—not right now, anyway. But here there is nothing wrong with his intention (to replenish the water supply by pumping); it is his execution (his pumping) that is at fault: “the mistake is in the performance,” Anscombe famously says, fol55 Anscombe (1963), §23.

158

Will Small

lowing Theophrastus.56 However, it seems as if we can push on this case to bring it closer to the previous one (the hole in the pipe). For, we can ask the pumper how he is pumping, and thereby get him to further articulate his intention: to replenish the water supply by pumping, by making these movements with his arm. And now, just as pumping is not a way of replenishing the water supply if there’s a hole in the pipe, so making those movements with his arm is not a way of pumping; again, he needs to respecify his intention. In both cases, what he’s doing isn’t working, though there remains a difference: in the one case, the pump–pipe system isn’t working, in the other, the pumper—his pumping—isn’t working.57 Examples such as these illustrate what is invariably the case: that there is a kind of constant correction that goes on throughout the course of an intentional action, as the agent responds to the miniature successes and failures, obstacles and alternative possibilities both foreseen and unforeseen, that he encounters in what he’s doing and what he’s acting on and with. What I am bringing out is that this constant correction amounts to rationally further determining or respecifying one’s intention: the answer to the question ‘How?’ is constantly being finessed. This should be familiar to all of us who produce things like arguments, the construction of which involves many false starts and pursuits of different avenues, the argument only gradually taking on a determinate shape. Now, practical deliberation just is rationally specifying an intention (for the future or, as emphasized here, in action) in the light of one’s ends, powers, and circumstances. Thus we can see that practical reasoning continues for the duration of the action; it is not over and done with once one, say, decides that one will replenish the water supply by operating this pump.58 56 Anscombe (1963), §45. In saying that there is something wrong with the pumper’s intention in the case where he is pumping fruitlessly (because of the hole in the pipe), I do not mean to imply that the pumper is to blame (though he might be, if e. g. the maintenance of the pipe was his responsibility, or if the pipes hadn’t been used for some time, etc.). His intention just needs fixing (“picking up”). 57 Thanks to Anton Ford for this way of putting things, and for helpful discussion here. 58 In saying that practical reasoning is the rational specification of an intention, I am not denying that it concludes in intentional action; rather, I am elucidating a conception of practical reasoning and intentional action according to which whenever a piece of practical reasoning concludes in an intentional action, it does so by specifying an intention, and further determining or respecifying

Practical Knowledge and the Structure of Action

159

I am suggesting that when someone, for instance, as he is stirring the milk into his coffee, finds that he has not added quite enough and so adds in a little bit more, this “on-the-fly” correction amounts to a rational further specification of his intention in action and thereby to a piece of practical reasoning. One way in which someone might try to explicate this suggestion is with the idea that, running through the agent’s stream of consciousness, alongside his bodily activity, is some such train of thought as “This coffee isn’t reaching the desired colour; it would reach the desired colour if I added in a little bit more milk; so I’ll add in a little bit more milk”. It is not phenomenologically implausible to suppose that, on some occasions, agents do accompany these kinds of corrections with such dialogues, in their heads or even out loud. But it would be wildly implausible to suppose that they always do this; our agent might have a bit of song running through his head as he adds a little more milk in order to make it milky enough, or he might be busily engaged in a piece of deliberation pertaining to something quite removed from the business of making coffee. (In denying that practical reasoning must be either separate or separable from the intentional action whose structure it explicates I am not, of course, denying that it can be separate, that one can come up with a plan of action well before one begins to act. Intention for the future, and the practical knowledge that inheres in it, are the subject of §3.) This might incline one to suppose instead that the practical reasoning in which this further specification consists need not be conscious. But it is not as if, were we to ask the agent why he was putting some milk into coffee into which he had already added milk, he would not be able to simply say something like “Because it wasn’t yet milky enough for my taste” and thereby express, if perhaps only partially, his intention. It is plausible to think that an “unconscious intention”, if there could be such a thing, could not be expressed by simply avowing it in this way, and would need to be discovered by its subject, as precisely our man does not need to do.59 That his intention to add a little that intention in action (as necessary) throughout the action’s duration. The connection between the calculative structure of an intentional action and the action’s duration will become important in my investigation into the temporal structure of intentional action in §4 below. 59 As will be clear from what is to come, I do not think there can be unconscious intentions. I do not, of course, deny that there are unconscious mental states; but intention is, as e. g. desire is not, essentially a rational and self-conscious act of the will. On the connections between expression, avowal, and conscious-

160

Will Small

more milk in order to make it the right colour is conscious is shown by the fact that he can simply say, without having to find out, that this is what he is doing. This in turn shows that what we need to do is rethink what it would be for his intention in action, and the practical reasoning that respecifies it as his intentional action progresses, to be conscious. The sensitivity of an intentional action to its own progress is a mark of its intelligence, yet there is no need to picture this sensitivity—this rational further determination or respecification of the intention in action—as something standing outside the intentional action itself, as Gilbert Ryle points out: “When I do something intelligently, i. e. thinking what I am doing, I am doing one thing and not two. My performance has a special procedure or manner, not special antecedents”.60 The “special procedure or manner” that belongs to intentional action as such is that of being prosecuted in accordance with the form of Anscombe’s A–D order: a means is a means relative to an end just in case it is sufficient to produce that end (perhaps in concert with other means), so if one is doing A in order to do B, but doing A (like this) isn’t working, one must modify what one is doing in order for what one is doing to be a means to doing B.

2.4 Calculative Holism With the Non-Accidentality Condition, Anscombe reminds us that it is possible for one to be doing B by doing A, and doing A in order to do B, and yet not be doing B intentionally—if one’s realization of the conness, see Finkelstein (2003). Finkelstein argues, plausibly, that the ability to express a mental state by avowing it is a necessary mark of the state’s being conscious. 60 Ryle (1949), 32. Ryle objects to the idea that the rational specification of the intention in action must be conceived of as a special antecedent to the action that it is in; his remark is naturally taken as rejecting a picture on which the intention (or intentions, if that is how the specification is to be understood) in action efficiently cause the bodily movements that constitute (either in virtue of being so caused or together with that cause) the intentional action. (Think, for instance, of John Searle’s theory of action (Searle (1983), ch. 3), according to which an intentional action is the causally-complex combination of an intention in action and the bodily movement that is its effect.) But Ryle’s objection—“I am doing one thing not two”—would just as well apply to a conception of intentional action according to which intention (in action) “guides” action, but does not constitute it.

Practical Knowledge and the Structure of Action

161

cept do B is an accidental consequence of one’s doing A. (Suppose, for example, that to operate the pump, which unbeknownst to our man does not work, he has to stand on a platform, and that in doing so he has activated the automatic pump, of which he is not aware, and is thereby replenishing the water supply for as long as he stands on the platform.) This reminder reveals that to get intentional action in view we must consider it in its calculative context: that is, we must consider the chain D-by-C-by-B-by-A (equivalently, A-because-B-because-C-becauseD) as a unit, one whose complexity is not to be understood by conceiving of it as composed from independently intelligible atoms A, B, C, D, and the ‘by’/‘because’-relation. On the one hand, A, B, and C show the texture in practical reality, so to speak, of D—the particular way in which its possibility is being realized, the particular shape that doing D is taking on this occasion. The shape that poisoning the inhabitants of the house is taking on this occasion is via the water supply; but if the cook weren’t eagle-eyed and incorruptible, it might have been via the food. Only with an understanding of the practical texture that a deed takes can we make certain judgments about it, for example whether it was done quickly or slowly, as in an example of Davidson’s: “Susan says, ‘I crossed the Channel in fifteen hours.’ ‘Good grief, that was slow.’ … Now Susan adds, ‘But I swam.’ ‘Good grief, that was fast.’ We do not withdraw the claim that it was a slow crossing; this is consistent with it being a fast swimming”61. On the other hand, D gives the point or use or good of A, B, and C—it sets the standard for their success, for what counts as doing or having done them well, badly, or even at all (that is, for what counts as having finished doing them). The standard of success for replenishing the house water-supply depends on whether or not one is replenishing it for the sake of poisoning the inhabitants. Suppose our pumper replenishes the supply with unpoisoned water. If he is replenishing it in order to poison the inhabitants, this constitutes a mistake; if not, it doesn’t. And if he is replenishing the water supply because he’s poisoning the inhabitants, he needs to pump enough poison to kill them (supposing that to be his intention in poisoning them) and also e. g. pump enough water to avoid detection (at least before the poison does its work). Otherwise the re-

61 Davidson (1967), 106.

162

Will Small

plenishing, and all the pumping and arm-moving that went into it, will have been in vain.62 These considerations show not only that intentional action is structurally complex, but also that so must be its representation. This has an obvious implication for the cognition condition. If an agent is doing B intentionally, he is intentionally doing-B-by-doing-A and doing-B-because-he-is-doing-C. His doing B intentionally cannot be understood apart from its calculative context—how and why he is doing B. And this goes just as much for the agent’s representation of it enjoined by the cognition condition: if he is doing B intentionally, then he knows that he is doing B, and he knows that he is doing B by doing A and doing B because he is doing C. In short, knowing what you are doing involves knowing how and why you are doing it.63 Knowing why and how are internal to knowing what one is doing, not something external that might be known “in addition”. Only when situated in its calculative context can an intentional action be assessed as a success or failure in various ways, and—crucially—the agent needs to be in a position to make these evaluative judgments about his ongoing performance so that he can shape and re-shape it in the ways necessary to achieve his end: the pumper needs to know when to stop pumping, but he cannot know whether he’s pumped too little or gone on too long unless he knows that he’s pumping because he’s replenishing the water supply. It can only be in a useless and etiolated sense that anyone, and in particular the agent, can “know what the action is” independently of knowing why it was, or is being, done. 62 Note that this does not imply that it is not possible to do something intentionally for no reason. According to the conception of the cognition condition I am developing, if an agent is doing A for no reason, he knows why he is doing A: for no reason. Thus there are no further ends that can set normative or temporal constraints on his doing A. If he can be said—and I think in the normal cases of acting for no reason (doodling, tapping one’s fingers, whiling away the time, etc.) it would be a stretch—to be doing A well or badly, or to be half-way through doing A, then such judgments would have to depend on considerations intrinsic to doing A. But this, if possible at all, would of necessity be the outlying case. 63 It is not uncommon for philosophers inspired by Anscombe to recognize that knowing what one is doing involves knowing why one is doing it, but that knowing what one is doing equally involves knowing how one is doing it typically goes unthematized. See, e. g. Velleman (1989), chs. 1 – 2; Setiya (2007), 40; Setiya (2010), 89 and Moran/Stone (2009), 150, 160.

Practical Knowledge and the Structure of Action

163

2.5 The Cognition Condition Reformulated The calculative structure of intentional action is a case of self-conscious teleology, differing from the teleological structure found in the behaviour of plants, animals, and machines in that it is constituted by the agent’s representation of it. This is why, while no one is going to suggest that the teleological structure of the behaviour of plants, animals, and machines implies a cognition condition, we have reason to think things will be different in the case of intentional action. We understand the functional relations that hold between the different processes that constitute the workings of a machine, the distinction between function and malfunction, and the distinction between a part of the machine (including the improper part) “fulfilling” its function accidentally or non-accidentally, all in terms of the machine’s design. An explanation of the teleology of living things will not appeal to an external agency of design, but rather to the nature of the kind of creature under consideration— its life-form. In the case of intentional action, the teleology internal to the structure of action is explained by the agent’s practical thought. That is, the reason why doing D sets the standard on this occasion for the agent’s doing A, B, and C is that the agent himself represents doing D as his end, and doing A, B, and C as his means. My heart is beating because it is circulating blood around my body, and thus circulating blood by beating, and it does this whether I think the connection between beating and circulating or not. The form of explanation is biological, and as such it has nothing to do with thought. If I happen to think that my heart is beating in order to circulate blood around my body while it is in fact doing so, this is simply a coincidence.64 But it is not like this in intentional action: the representation of the teleology is constitutive of it.65 According to the account developed via the foregoing reflections on the A–D order, the source of the cognition condition on intention ac64 Obviously it is no coincidence in the following sense: if my heart weren’t beating, I wouldn’t be thinking. But if my heart is beating, then that it is beating in order to circulate my blood has nothing to do with any representation I might form of this connection. An important point is this: part of the explanation of why my heart is circulating blood around my body by pumping is that the heart circulates blood around the body by pumping. But it is no part of the explanation of why this pumper is poisoning the inhabitants of this house that “the” pumper “poisons” the inhabitants of “the” house. 65 Cf. Rödl (2007), 47 f.

164

Will Small

tion lies in the structure of intentional action itself. But this structure has been revealed to be the structure of practical reasoning: of the rational specification and dynamic further specification of means to ends. An agent engaged in a stretch of calculatively-articulated intentional activity knows what he is up to because he understands how and why the elements of his action, which are themselves intentional actions, combine such as to amount to him intentionally producing the whole thing. His understanding is practical: it is arrived at through practical reasoning, and abides in the calculative articulation of his intention, throughout the inevitable respecifications that he makes to it in the course of its execution. This is what it means, I think, to speak of agent’s knowledge as knowledge in intention, not a knowledge that somehow precedes or accompanies intention or intentional action while inhering in a belief merely contingently connected with intention or action. Thus, my investigation into Anscombe’s account of the calculative structure of intentional action has revealed that this structure is the source of the cognition condition, of which a more detailed and robust articulation is now available: if you are acting intentionally, you have knowledge in intention of: (i) what you are doing (e. g. replenishing the house water-supply), (ii) why you are doing it (e. g. because you’re poisoning the inhabitants), and (iii) how you are doing it (e. g. by operating this pump). However, this formulation suggests that the account just given of the structure of intentional action is, strictly speaking, incomplete. If the answer to the question ‘Why?’, asked of an intentional action, cites just another intentional action, and this can keep on going by extending the A–D order as far as we want, then it seems that we don’t understand any of the answers. Likewise, if the answer to the question ‘How?’ asked of an intentional action cites just another intentional action, and this can keep going on ad infinitum by extending the A–D order in the other direction, then it seems we don’t understand how anyone can act intentionally at all. The first worry concerns whether there are non-calculative reasons for acting and what they’re like if there are any. The second worry concerns whether there are basic actions and what they’re like if there are any. A full treatment of the topic of practical knowledge, which, in any case, lies beyond the scope of this essay, would require addressing both of these worries. In the context of my more limited ambitions here, however, it will suffice to indicate

Practical Knowledge and the Structure of Action

165

how it is so much as possible to put the second worry to rest. To see why this is called for, it will be useful to have an overview of the remainder of the argument. Whether or not there is a topic of which to give a full treatment depends on whether or not there is a cognition condition on intention and intentional action, and if so, what that condition is. In other words, it depends on the nature and success of a response to the counterexamples proposed by Davidson and Bratman that we canvassed in §1.2. The purpose of the present section was to argue that the cognition condition on intentional action is calculatively articulated into knowledge of what the agent is doing, and how and why he is doing it, on the grounds that intentional action itself is calculatively articulated, and the agent’s knowledge in intention is the source of that articulation. In the next section (§3), I extend this account to the cognition condition on intention for the future, before going on to argue (in §4) that intentional action is temporally structured, such that there is an internal connection between doing something intentionally, or intending to do something in the future, and being such as to end up having succeeded in doing it. In §5, I show that the account of intention and intentional action as calculatively and temporally structured, according to which the agent’s knowledge in intention is the source of this structure, makes possible a response to the counterexamples that purport to show that practical knowledge is not a necessary condition on intention and intentional action. The reason why it will be necessary to broach the topic of basic action is this. Reflection on intentional action’s calculative structure—on the idea of doing one thing by doing another—suggests that there must be calculatively basic action if there is to be intentional action at all. But the reflections on the temporal structure of intentional action (§4), which underpin the response to Davidson (§5), appear to entail the surprising result that there can be no such things as basic actions. And so the worry is that the argument will have secured the claim that if an agent is doing something intentionally, he has knowledge in intention of what he is doing, at the cost of proving that no one ever acts intentionally, because it is impossible to do so—a disappointing result. Accordingly, in §§6 – 7 I will address the worry about basic action.66 66 With respect to the first worry, about non-calculative reasons for action, it will suffice for the purposes of this essay to point out that any realistic account of acting for non-calculative reasons, whether they be lofty or mundane, will pre-

166

Will Small

3 From Intentional Action to Intention 3.1 Intention, Aspiration, and Knowledge How Anscombe claims that if an agent is doing something intentionally, he has knowledge in intention that he is doing it. In §1.1, I suggested that, given what she says about the unity of the three headings under which the concept of intention appears—intentional action, intention with which, and expression of intention for the future—we should think of the cognition condition as applying to intention with which and the expression of intention for the future as well. My basic strategy for “extending” the account given of the cognition condition in the last section from intentional action to intention for the future is straightforward, if contentious (a direct defence of it cannot, for reasons of space, be given here): intention for the future is intentional action in prospect, and thus falls under the same account. We simply shift from considering the calculative structure of intentional action in progress to that of intentional action in prospect, where we find the applicability of the same three questions (‘What?’, ‘Why?’, and ‘How?’), and for the same reasons.67 That is, if someone announces an intention to do something next Tuesday, we get a better idea of what he intends to do by asking him how he intends to do it and why. In finding out the answers to these suppose the account of the calculative inner structure of intentional action that I have sketched—after all, sometimes even acting from the motive of duty might involve crossing the street, and thus looking both ways, taking one step and then another, in order to cross the street, in order to perform one’s sublime act. Candace Vogler gives a more detailed argument, along these lines, for the claim that acting for calculative reasons is the fundamental form of intentional action in ch. 6 of her Vogler (2002). 67 By focusing (in the last section) on intentional action and (in this section) on intention for the future and its expression, it might seem as if I have ignored Anscombe’s third heading, intention with which. However, in examining the structure of intentional action, we have seen that it contains intention with which: if someone is doing A intentionally because he is doing B intentionally, he is doing A with the intention of doing B, and he knows that he is doing B. If, on the other hand, he is doing A with the intention of doing B, but he is not yet doing B, then he is doing A because he intends to do B, and an expression of his knowledge in intention will at the same time be an expression of intention for the future. This, then, is why I have not treated the idea that there is a cognition condition on intention with which separately.

Practical Knowledge and the Structure of Action

167

questions, not only do we have a better idea of what he intends to do, we also become convinced that he has the announced intention. Here the question ‘How?’ is crucial, as Annette Baier points out: whenever we announce an intention in a vague way, we must be prepared to back it up with a more precise specification or a demonstration of how we will do what we intend doing, if our hearers are to be satisfied that we can do it, and so be satisfied that what we have is an intention and not just an aspiration. … Answers given to the question “How?” or “What’s your method?” may themselves invite the same “how” question…. If the question can continue to come up, then it will remain a mystery how the man will do what he says he will do.68

Earlier on (§2.4), we imagined a scenario in which the pumper was doing what he intended (replenishing the water supply), but not in the way he intended (operating the pump with his arm); in fact, he was replenishing the water supply by standing on the platform next to the old hand pump, which activated an automatic pump. Because he didn’t know how he was doing what he was doing, he wasn’t doing it intentionally. The connection between his means and end was accidental; it was not contained in his intention, and thus not something known to him in intention. He was doing what he intended, but not intentionally; he was realizing his goal not through practical reasoning and knowledge, but simply through luck. Now, Baier’s remarks suggest the following parallel, to be explored and refined in the remainder of this section. Just as someone who is bringing off his end D without knowing how he is doing it is not doing D intentionally, so someone who “intends” to do D without knowing how he is going to do it does not really intend to do D: he merely aspires to do D. Aspiring to do something differs from merely wishing for some state of affairs or event to obtain or transpire. As Anscombe notes, “[a] chief mark of an idle wish is that a man does nothing—whether he could or no—towards the fulfilment of the wish”69. Even though a mere or idle wish represents its object as in some sense good, it does not amount to practical thought because it has no inner tendency towards action, which is what practical thought is ultimately for.70 By contrast, just as 68 Baier (1970), 653 f. 69 Anscombe (1963), §36. 70 Cf. Anscombe: “The primitive sign of wanting is trying to get” (Anscombe (1963), §36). The kind of wanting of which Anscombe speaks—which she distinguishes from (a) idle wish, (b) the “prick of desire at the thought or sight of an object, even though a man then does nothing towards getting the object”

168

Will Small

an intention has the inner tendency to further articulate and re-articulate itself in action, so an aspiration has the inner tendency to articulate itself such that it becomes an intention. This articulation too is the work of practical reasoning: an agent’s aspiration can become an intention, by his coming up with a plan—by reasoning from the end which is at a distance, to an immediate means that he can take. This inner tendency towards such articulation—an inner tendency towards perfecting itself in action and practical knowledge—is what marks practical thought out as rationally efficacious; that is, as practical. It is in the nature of the thought ‘I want (aspire, intend) to do D’ that it ends up articulated ‘I intend to do D by doing C by doing B by doing A’ and realizes that articulation in action and practical knowledge: ‘I am doing D by doing C by doing B by doing A’. Such inner efficacy is absent in mere wish.71 To illustrate, consider the following attitudes towards becoming rich. Someone who idly wishes he were to become rich does nothing towards getting rich (save perhaps purchasing the occasional lottery ticket, though this is by no means necessary); his wish is primarily manifested in fantasy—in imagining that he wins the lottery, or that he has a rich aunt who makes him the heir to her fortune, in imagining where he would live, what he would buy, and so on. Someone who aspires to become rich would be getting rich if he knew how, but he does not; in contrast to the idle wisher, his aspiration is manifested in deliberation and action, not merely in imagination—not in getting rich, however, but in figuring out how to get rich (he spends his time at the library, reading up on real estate, the stock market, spread betting, or bank robbery).72 Some(Anscombe (1963), §36), (c) hope, and (d) “the ‘I want’ of a child who screams for something” (Anscombe (1963), §40)—is, as I understand it, the most primitive kind of practical thought. 71 Stephen Engstrom (2009), 68 ff., makes some similar remarks, in a Kantian framework, about the way in which deliberation serves to transform “wish”—a “problematic” practical judgment specifying what the agent would do that, though merely “practical… in potentia” (Engstrom (2009), 68), is nevertheless by that very feature to be distinguished from a mere or idle wish—into “choice”—an “assertoric” practical judgment specifying what the agent will do by specifying how (by what means) the object of wish is to be realized. According to Engstrom, “the inner act of deliberation is the characteristic form of the efficacy proper to wish itself, an efficacy that can be described as practicality’s own self-actualization, and even (given practicality’s self-awareness) as its selfconscious self-actualization” (Engstrom (2009), 87). 72 Such a man might be walking to the library with the intention of reading a book about spread betting with the intention of figuring out how to get rich

Practical Knowledge and the Structure of Action

169

one who intends to become rich, however, knows how to get rich: he has worked out a plan, which he is putting, or going to put, into action. And if he is getting rich by executing his plan, he is getting rich intentionally. The distinction between intention and aspiration is easy to overstate, however. Surely it is possible for you to intend to go shopping next weekend, despite not having decided where exactly you’ll go or whether you’ll drive or take the bus; it would be decidedly off-key to lump you together with the man who wants to get rich but has no idea how to do so, or to insist that you provide us with “a more precise specification or a demonstration of how you will do what you intend doing” on pain of refusing to credit you with an intention. We do not want the implication that an intention is somehow fully determinate (fully calculatively-articulated), and that anything less than that must relegate it to the status of a mere aspiration. The idea that an intention for the future could be fully determinate is a philosophical fantasy; it would preclude what is always possible and usually necessary: the rational further specification of one’s intention in action.73 But then doesn’t this suggest that while someone who intends to do something knows what he intends to do and why he intends to do it, he need not know how he intends to do it, and thus that the intention/aspiration distinction is spurious? I think not. An intention need not be fully articulated into subordinate intended means, but, as practical, it is such as to so articulate itself through practical reasoning. It may be that the question of the means of transport to the shops need not be settled until the morning of departure. But one who intends to go shopping next weekend knows that the means of transportation needs to be decided upon at some point. There is a significant difference between leaving it unsettled which of several options one could take, one shall take, and not having any with the aspiration of getting rich. He would not thereby be getting rich intentionally. 73 The fantasy of a fully determinate intention for the future is not unmotivated. As far as I can tell, it arises from supposing that we can take the unity of a given intentional action for granted (or, what comes more or less to the same, conceive of it as unstructured), and then ask for its cause, while supposing that, as Hume puts it, “the effect is totally different from the cause, and consequently can never be discovered in it” (Hume (1977), 18). Part of the goal of this essay is to break down, by way of an analysis of the structure of intentional action (in progress and in prospect), the tendency to suppose that intention and intentional action must be “distinct events”.

170

Will Small

idea of what the options are, not having any idea how to do what one intends to do. This reveals that we need to be more precise in drawing the distinction between intention and aspiration in terms of knowing how to achieve one’s end. One can intend to do A without knowing how one shall do A so long as one knows how one can do A. But if one doesn’t know how one can do A, then one can merely aspire to do A (where the inner tendency of this aspiration is towards the acquisition of knowledge how to do A, and thus towards its own transformation from aspiration into intention). Thus, I may know how I can get to the shops without yet knowing how I shall get to the shops next weekend. This is sufficient for me to intend to go to the shops next weekend, in the knowledge that my intention, as such, must at some point prior to departure become articulated in some such way as ‘I intend to go to the shops by taking the #6 bus’ if I am to actually get there. But if I do not know how I can get to the shops, then all I can do is aspire to go shopping. This distinction between two levels of knowing how—general knowledge how to do A, and particular knowledge how one is going to do A on some particular occasion—is of importance, and applies to intention in action, too. If one is intentionally doing B by doing A, one knows how one is, here and now, doing B: by doing A. This particular knowledge how is practical knowledge, knowledge in intention. It is not derived from observing that the way one happens to be doing B is by doing A. It is possible only because one already has the corresponding general knowledge how, that one can do B by doing A, or, equivalently, that doing A is a way of doing B; one puts this general knowledge into practice on a particular occasion in doing B by doing A. 74 The distinction between intention and aspiration is not binary, allor-nothing. It is not that either I definitely have general knowledge how to do something, and can thereby intend to do it, or I have no idea how to do it, and can thereby harbour mere aspiration. I might well reason: ‘I want to do B; doing A might work; so I’ll do A’. Is my thought about doing B an intention or an aspiration? And if I did A and thereby did B, did I do B intentionally? Well, there need be no easy answer; but (as Dr Johnson reputedly said) the fact of twilight does not mean that we cannot tell between night and day. It is not always easy to tell whether an 74 Cf. Rödl (2011); Setiya (2009b), 135. That general knowledge how may be indexed to certain agents, conditions, and circumstances does not speak against, but rather specifies, its generality.

Practical Knowledge and the Structure of Action

171

action was performed intentionally or brought about merely through luck, or where exactly it fell on the spectrum along which terms like ‘adventitious’ and ‘fortuitous’ have homes, but in many cases it is easy enough to tell.75 This suggests that the inner tendency to calculative articulation that belongs to practical thought as such has two aspects: on the one hand, the specification of means through calculation (the development from the desire to do D to the intention to do D by doing C by doing B by doing A), and on the other hand, the strengthening of the calculative links (from ‘maybe doing B is a way to do C, who knows?’ to ‘doing B might be a way to do C ’ to ‘doing B is a way to do C ’).76 According to the picture I have developed, aspiration is a kind of practical thought that is not itself intention, but which has an inner tendency to develop itself into intention. The primary way in which this happens is through practical reasoning itself: I aspire to do B, but don’t know how to do B, so I figure out that I can do B by doing A, and thereby come to intend to do B by doing A. But it may be that reasoning is not enough. Kieran Setiya77 asks whether he can form the intention to dance the tango at his wedding, though as yet he doesn’t know how to dance the tango. The answer, on this view, is that he cannot; so far dancing the tango may figure only as the object of his aspi75 This is a simplification. Though I am proposing a conception of aspiration according to which it presupposes the concept of intention, from which it is understood by subtraction, aspiration nevertheless has a positive grammar of its own, which distinguishes it from (e. g.) idle wish. Just as it would be a confusion to conflate being an intender with being an aspirer, so it would be a confusion to conflate being an aspirer with being an idle wisher. Like intention, aspiration has its own wider context to which we must look before ascribing it. Someone who seriously aspires to do something, or to be able to do something, will be engaged in various kinds of activities and projects to that end. So it would hardly be mere luck if, were she to try to do that which she aspired to do, she succeeded in doing it on that occasion. The point is rather that a case of successfully doing what one aspired to do is fortuitous in a way in which successful intentional action proper is not. (It would be mere luck for someone who had never shot an arrow, or a gun, or played darts, etc., to hit ten out of ten centre shots the first time she picked up a bow and tried her hand at archery. However, for an aspiring archer, for whom three centre shots out of ten attempts is currently the norm, hitting ten out of ten on some particular occasion would hardly be mere luck—yet it seems clearly to be fortuitous, in a way in which an expert’s hitting ten out ten would not be.) Thanks to Jim Conant for helpful discussion here; see also §4.5 (especially fn. 119), and §5 below. 76 This paragraph is based on a conversation with Sebastian Rödl. 77 Setiya (2008), 406 f.

172

Will Small

ration.78 But not knowing how to dance the tango is unlike, say, not knowing how to get home if the subway isn’t running. If the subway isn’t running, you can come up with an alternative plan by apprising yourself of the bus schedules, looking up a phone number for a taxi, consulting a map to find a route you can walk, and so on; that is, you can engage in practical reasoning that takes you from your end to means that you can in fact take. By contrast, if you aspire to dance the tango at your wedding, your aspiration can become an intention only if you learn how to dance the tango, and there is more to learning how to dance the tango than engaging in practical reasoning; all you can intend, for the time being, is to learn. The difference between the two kinds of case is this: on the one hand, it is a matter of figuring out how to use one’s practical abilities in concert with one another given the practical circumstances (you want to do B, you know how to do A, you just need to figure out that you can do B by doing A), on the other, it is a matter of acquiring a new practical ability. I originally quoted Baier with a view to developing this parallel: just as someone who is bringing off his end D without knowing how he is doing it is not doing D intentionally, so someone who purportedly intends to do D without knowing how to do it does not really intend to do 78 Sarah K. Paul claims that this “get[s] things backward.” She asks: “Does not the rationality of my deciding to learn how to dance the tango depend in part on my having an intention to dance it at some future point, rather than the other way around?” (Paul (2009), 556). There is no need to suppose that one must intend to dance the tango in order to explain why one has reason to learn; wanting or aspiring to dance the tango gives one reason enough. Paul appears committed to unsatisfyingly representing an agent’s practical powers as conditions external to his practical thought, circumstances to be dealt with in the course of executing his intention and not conditions internal to having an intention. But there is a qualitative difference between, on the one hand, the relation between an intention to dance the tango and the discovery of a circumstance that necessitates re-articulating—so as to “pick up”—one’s intention (the discovery, say, that one’s shoes are unsuitable for dancing), and, on the other, the relation between an ostensible intention to dance the tango and the discovery that one doesn’t know how to dance it. By contrast, Setiya thinks that “it is possible, if inadvisable, to form the intention of doing something I do not know how to do” (Setiya (2009b), 136). It would be inadvisable, he thinks, because the belief in which the intention would consist on his cognitivist view would be unjustified. Thus he presupposes an account of intention that is normatively constrained, but constitutively unconstrained, by knowing how to do what one intends to do. Such an account cannot mark the distinction between intention and aspiration.

Practical Knowledge and the Structure of Action

173

D; rather, he merely aspires to do D. 79 Our discussion has primarily focused on drawing out the way in which intentions, like intentional actions, are themselves calculatively structured (though perhaps so only potentially).80 The general knowledge how to do things implicated by the calculative structure of practical reasoning is knowledge of means, where this is understood on the model of deliberation: of figuring out what would be sufficient to achieve one’s end, and selecting one such (set of) sufficient means. Such deliberation is often thought about in terms of planning81—though as we saw when thinking about the rational re-specification of an intention in action in §2.3, the deliberative process need not be a “mental” process distinct, or even distinguishable, from the progress of the intentional action itself. But we have now uncovered another way in which an agent might not know how to do what he aspires to do, namely by failing to possess the skill or skills his plan requires of him. This suggests that there are two kinds of knowledge how to do something—procedural knowledge how and skill—to which I will return in §6. Assuming that what one intends to do is neither a basic action nor something one is doing for a non-instrumental reason, to intend to do something is to intend to do it for the sake of doing something else, and (at least by when the time to execute it comes) to intend to do it by means of doing something else. Intention, whether it is in action or for the future, is, essentially, calculatively articulated. This point is not refuted by the possibility of intentions to do things for which the means have not yet been decided upon, because insofar as such a thought is an intention, the efficacy that belongs to it as practical thought is realized (in part) by figuring out and selecting such means—by deriving particular from general knowledge how—when or before the time to take them comes. As we have already seen, calcu79 The initial proposal, made at the beginning of this section, was that someone who purportedly intends to do D without knowing how he is going to do it (i. e., without having particular knowledge how) does not really intend, but merely aspires, to do D. The refined proposal articulated here replaces the reference to particular knowledge how with one to general knowledge how. 80 When I say that an intention to do something may be calculatively structured only potentially I mean more than that it is (merely) possible that it ends up becoming so articulated; I mean that it is in its nature to so articulate itself, and that it will be no accident if it does and some accident if it doesn’t. 81 The popularity of this helpful but ultimately one-sided way of thinking about practical thought is primarily owed to Michael Bratman (1987).

174

Will Small

lative articulation is self-conscious teleological articulation: the means–end structure of intention (whether in action or for the future) obtains in virtue of the agent’s knowledge of it. Thus, if you intend to do something, you know what you intend to do, why you intend to do it, and how to do it, and you either know how you intend to do it or that you will know how you intend to do it at some point before doing it. The cognition condition applies to intention for the future as well as intentional action. This falls out of the fact that intention for the future is intentional action in prospect: what it is that one now intends to do is something whose subsequent realization will take the calculatively-articulated form we have already discussed, where this form will obtain precisely because the agent represents it as obtaining; one’s present representation of one’s action in prospect must also take this calculativelyarticulated form if it is to be the representation that is the source or cause or principle of the action, when its time comes.82

3.2 Thinking that You’ll Try and Thinking that You’ll Succeed This is enough, I think, to see that Bratman’s bookstore case from §1.2 can be dispensed with. In his example, Bratman (purportedly) intends to go to the bookstore, but he’s agnostic with respect to the question whether he will even try to go there when the time comes, because he knows he’s so forgetful.83 Bratman concludes from this example that an intention to do A doesn’t “involve” even the belief that one will try to do A, let alone the belief or knowledge that one is actually going to do A. But I think that we can now see that Bratman’s “intention” to go to the bookstore reveals itself to be simply a mere aspiration or wish: if he’s considered his bad memory (and this, after all, is the ground of his agnosticism), then it’s a pretty hopeless plan that doesn’t involve writing a reminder on his hand or leaving a post-it on the dashboard or something. This is in part an ad hominem objection, because for Bratman intentions are just plans writ small. But it is not just ad hominem: if someone avows an intention to do something, we can investigate

82 This section owes much to conversations with Jim Conant, Sebastian Rödl and Candace Vogler. 83 Bratman (1987), 37.

Practical Knowledge and the Structure of Action

175

whether he really has an intention and not a mere aspiration or wish by asking him how he’ll do what he purportedly intends.84 The kind of case that should concern us, then, is one where someone putatively intends to do something, despite being agnostic as to whether or not he will succeed, not whether or not he will try. Such a case would be directly in line with Davidson’s original carbon-copier case. And indeed, someone who is impressed by carbon-copier counterexamples might feel aggrieved by the discussion so far, on the grounds that it seems tangential to the real issue: belief in—or even worse, foreknowledge of—success. It may be all very well to suppose that if you are doing something intentionally, you know why and how you’re doing it, but this isn’t the same as knowing that you will end up having done it. Likewise, it may be all very well to suppose that if you intend to do something, you know why and how you are going to do it, but the real question is whether or not you will do it. Knowledge of success is the issue; it is scepticism about eventual success that props up the agnosticism on which carbon-copier style worries trade. We must turn, then, to the question of the relationship between doing something (or intending to do something), being such that one will end up having done it, and knowing that one will end up having done it. This question does not concern in the first instance the calculative structure of intention and intentional action, but rather their temporal structure. Again, I begin with action, by examining the connection between two thoughts that, I will claim, belong to the knowledge in intention of an agent who is doing B intentionally: ‘I am doing B’ and ‘I am going to do B (i. e. end up having done B)’.

4 The Temporal Structure of Intentional Action 4.1 Stopping and Finishing Here, a comparison with Davidson can take place rapidly, because— somewhat surprisingly—his account of the logical form of action sentences is silent about the truth-conditions of the kinds of sentence that comprise our topic, i. e. ‘S is doing B’ (or more precisely ‘I am 84 Of course, most people can and do simply intend to go to the bookstore; the point is that the context Bratman describes is one in which the status of his “intention” is in question, legitimately subject to the question ‘How?’.

176

Will Small

doing B’). The object of his inquiry is completed actions, and thus sentences of the form ‘S did B’, which, as is well known, he analyses by quantifying over events along these lines: (9x)(did B(S, x)).85 Anscombe provides the clue for thinking that this amounts to an oversight: “A man can be doing something which he nevertheless does not do, if it is some process or enterprise which it takes time to complete and of which therefore, if it is cut short at any time, we may say that he was doing it, but did not do it”86. What would we quantify over in such a case? Picking up on this clue, one strand of recent work in action theory has made much of the temporal structure of action. Rather than theorizing only completed actions, philosophers have increasingly been drawn to examine the temporal structure of actions in progress.87 The crucial element that this brings into the discussion is marked in language by what linguists call aspect. In applying a state-predicate F to an object a, the form of predication admits contrasting tenses (a was/is/will be F). By contrast, in applying an event- or process-predicate do B to an agent (or patient) S, the form of predication admits contrasts of both tense (past, present, future) and aspect (imperfective, perfective); the aspectual contrast that interests us is between one type of imperfective, the progressive (S was/is/will be doing B), and the perfective (S did/will do B).88 Davidson’s investigation into the logical form of action sentences characterized by perfective aspect conceals the fact that actions (and events more generally) are temporally structured. As Bernard Comrie puts it, perfective meaning “presents the totality of the situation referred to…without reference to its internal temporal constituency: the whole 85 Davidson (1967) 86 Anscombe (1963), §23. 87 The most influential work on this topic is Michael Thompson’s ‘Naive Action Theory’ (2008), pt. 2. See also Falvey (2000); Rödl (2002); Rödl (2007), ch. 2; Moran/Stone (2009); Boyle/Lavin (2010). 88 There is no present perfective, at least for events and processes that take time to occur: ‘John walks to school’ does not say of the present what ‘John walked to school’ and ‘John will walk to school’ (i. e. ‘It will be the case that John walked to school’) say of the past and the future; rather, ‘John walks to school’ says that John habitually walks to school. By contrast, ‘John is walking to school’ says of the present what ‘John was walking to school’ and ‘John will be walking to school’ say of the past and the future. In general, ‘S does B’ says that S does B habitually; there are other readings too (describing non-durative events; the historical present; certain performative uses), none of which concern us here.

Practical Knowledge and the Structure of Action

177

of the situation is presented as a single unanalysable whole, with beginning, middle, and end rolled into one”89 ; but the whole of an event that takes time to happen cannot be located at a single present moment. Perfective meaning does not imply that the event or process is temporally structureless; it simply abstracts from its internal temporal constituency.90 By contrast, imperfective meaning, which is expressed in progressive judgments, “is crucially concerned with the internal structure of the situation, since it can both look backwards towards the start of the situation, and look forward to the end of the situation”91. By reflecting on the temporal structure of action, revealed by the aspectual contrast passed over by the standard Davidsonian approach, I will argue that the cognition condition on intentional action (in progress and in prospect) must encompass knowledge of eventual success, properly understood. I have already noted (in §1.3 above) one important feature of progressive judgments, their broadness: a progressive judgment ‘S is doing B’ may be true even if there is nothing that S is doing right now that amounts to doing B or taking some means to doing B. In this section, I focus on the other important feature of progressive judgments, their openness. The openness of the progressive is precisely what Anscombe points to when she notes that “a man can be doing something which he nevertheless does not do.” That is, while the truth (at t1) of ‘S is doing B’ entails (at t2) the truth of ‘S was doing B’, neither entails—at any time—the truth of ‘S did B’.92 Thus it could be true that S is (or was) doing B, even though S will never have completed this act of B-ing. The intelligibility of the thought that S was doing B, but didn’t do B (and isn’t still en train) depends on the event-predicate do B admitting a contrast between two ways in which something that is doing B can cease doing so: by stopping doing B and by finishing doing B. The predicate cross-the-street admits this contrast: one can have been crossing the street even if one didn’t and never will make it to the other side, if one’s street-crossing was cut short, perhaps by an oncoming bus. The predicate walk (on its primary reading) does 89 90 91 92

Comrie (1976), 3. Cf. Comrie (1976), 21. Comrie (1976), 4. Thus it is a mistake to assume that that an event- or process-predicate denotes a type. For an incomplete walking across the street falls under the concept walkacross-the-street even though there is no completed event—the token that would instantiate the type. See Thompson (2008), 134 – 38, especially fn. 21.

178

Will Small

not: there is no contrast between stopping walking and finishing walking. Following Comrie93 and Antony Galton94, I’ll call predicates that admit the stopping/finishing contrast telic, and those that do not atelic. To put the point slightly differently, telic predicates are those whose instances can be interrupted. An atelic predicate can be transformed into a telic predicate by specifying an end, either in terms of an objective to be achieved (e. g. from walk to walk-to-school) or in terms of a length of time (e. g. from walk to walk-for-an-hour). By specifying an end, one specifies the proper terminus of the event-form to which the predicate refers itself, whose instantiation on an occasion may then be subject to improper termination. All of this applies regardless of whether the event in progress is an intentional action or not: it marks what belongs to the representation of movement, a topic preliminary to ours, the representation of intentional action, which is a specific form of it. Armed with these conceptual resources, we are now in a position to make progress with our present guiding question: what is the relationship between doing A intentionally and ending up having done A intentionally? Clearly we are interested in the question as it pertains to telic concepts: our question pertains to the successful achievement of an end. We want to know whether knowing that one is doing A involves knowing that one will end up having done A—whether knowing what you’re doing involves knowing (or believing) you’ll succeed.

4.2 The Quick Argument and What It Shows There is a quick argument, based on the aspectual considerations already outlined, for the claim that it doesn’t. If John knows that he is walking home, then it is true that he is walking home. But we have seen that the proposition ‘John is walking home’ is inferentially unrelated, so to speak, to the subsequent truth of ‘John walked home’ (and its negation), and thus to the present truth of ‘John will have walked home’ (and its negation). So what John knows, when he knows he is walking home, does not involve knowing that he will succeed in walking home: his knowledge in intention does not amount to cognition of success. And given the suggestion I made in §3 that we conceive of intention for 93 Comrie (1976), 44. 94 Galton (1984), 67.

Practical Knowledge and the Structure of Action

179

the future as intentional action in prospect, it would appear that if John intends to walk home, then his knowledge in intention that walking home is what he is going to do is not in fact cognition of success; that is, his “knowledge” that he is going to walk home turns out not to be knowledge that he will, in fact, walk home. The quick argument, combined with the assimilation of intention for the future to intentional action in prospect, apparently shows that the cognition condition cannot extend to cognition of success. But the quick argument is no good. It slides from the fact that neither ‘S is doing B’ nor ‘S was doing B’ entails ‘S did B’, ‘S did not do B’, ‘S will do B’, or ‘S will not do B’ to the claim that ‘S is doing B’ and ‘S was doing B’ are indifferent to the eventual doing of B. But this is not so. Success and failure are not on a par: if they were, there would be no grounds for ever thinking that a sentence of the form ‘S is doing B’ is true.95 We can see this by looking at a case in which something was doing something, even though it didn’t end up having done it: Unlucky John, who was walking home from work, even though he didn’t end up having walked home, because he was run down by a car. To walk home from work, John needed to walk down Avenue A, take a right at Boulevard B, and then turn left into C Close. Suppose John has walked down A to B, and is walking along B towards C: he is walking, but has not yet walked home; all the while he was walking down A, and during his progress so far along B, he has been walking, but has not walked, home. Then, while crossing B in preparation to turn left into C, he is hit by the car; he doesn’t even start to walk down C, and hence doesn’t end up having walked home. Surely it is true, though, that he was run over while walking home. But if eventual success and failure were on a par with one another, then surely all we would be entitled to say is that what he was doing was walking down A and then along B. And in fact, it would be more appropriate to say, not that John was walking down A to B and then along B to C, but only that he walked down A and then along B to the fateful spot. That is, we are forced into scepticism about the very idea of an event’s being in progress. According to such a scepticism, events are things that have happened; there is no such thing as them happening. But such a position is incoherent, for the very idea of something’s having happened presupposes the idea that there was a 95 Cf. Rödl (2007), 29 ff.

180

Will Small

time at which it was happening, and the idea of something’s happening is what is presently denied.96 The argument from the premise that someone can be doing something that they nevertheless will not do to the conclusion that intentional action does not involve knowledge of success displays too little sensitivity to the aspectual considerations upon which it relies. Consequently, it proves too much: if the possibilities of eventual success and failure were on a par with respect to the truth of ‘S is doing B’, then the concept event would be annihilated.97 If the fact that something is doing something neither entails successful completion nor is indifferent to the possibilities of eventual success and failure, then the general conclusion we should reach is this: if S is doing B, then it will be no accident if S succeeds in doing B, and it will be some accident if S fails.98 This is no surprise: our understanding of the telic concept do B has two possibilities built in for ways in which something that is instantiating that concept can cease to instantiate it—either by finishing doing B or by stopping short of finishing. Stopping and finishing are not on a par; finishing doing B is proper to a process of doing B, stopping short improper.

4.3 The Epistemological Argument and What It Shows We said that for S to be doing B, it must be no accident that S ends up doing B, and that to deny this, as the quick argument implicitly recommends, would be to annihilate the concept event. But now a second sceptical possibility rears its head (epistemological rather than metaphysical): how can we know that S is doing B? And how could S know that he is doing B? Our sceptic grants that if S did B, there was a time when S was doing B; and at that time, it would, as things turned out, have 96 I am not supposing, absurdly, that the idea of something’s happening is intelligible independently of the idea of something’s having happened; the point is that the two ideas are interdependent, and equally implicated in the concept event. 97 Again, nothing turns on the fact that the event in my example is an intentional action (or rather, it would have been, had John made it home). I might just as well have illustrated the point with a tree falling over or a boulder rolling down a hill. 98 “Success” and “failure” here are to be understood in terms of the end specified by the telic concept; thus, one might succeed in dying or falling over.

Practical Knowledge and the Structure of Action

181

been true to say ‘S is doing B’. But what could have justified that assertion then? It looks as if all we could ever be justified in saying while an event is in progress is that it seems to be in progress. And it looks like the same could be said for the agent of an intentional action: if asked what you’re doing, perhaps all you are justified in saying is something like, “As far as I know, I am doing B”.99 While we can be justified, by the eventual occurrence of the event, in saying, later, that it was (then) in progress, and thus, though it turns out that it can be true that an event is in progress, such truths escape our knowledge until after the event’s completion. The objection is not that, if we are ever to claim that S is doing B, we must know that S will do B, for we have established that the progressive judgment does not entail the corresponding perfective judgment. Even if we could know, on an occasion, that S is doing B on the basis that we know that S will do B (though it is not clear how we could rule out the myriad possibilities that would prevent S from doing B), this cannot be the only criterion for knowing that S is doing B, or else we could never know that S was doing B but didn’t end up having done B. Thus the sceptic does not claim that we retreat from ‘S is doing B’ to ‘S seems to be doing B’ on the grounds that we have not ruled out the possibility that S might stop doing B before finishing doing B. Rather, the worry is that, faced with a situation in which S is putatively doing B, we have no criterion by which to say whether S is doing B rather than some observably compatible B’ or B’’ or… until S has done B, and if S doesn’t end up having done B, no criterion by which to say what, if anything, S was doing at all.100 We can bring out the worry by comparing the case of Unlucky John with this case presented by Kevin Falvey: A friend approaches me as I’m at the track one afternoon preparing to run. “What are you doing?” he asks, as I take off down the track. “I’m running a four-minute mile,” I call back, only to stop short coming off the first turn, exhausted, and plod slowly back to the starting line. “What happened?” my friend asks. “Well,” I say, “I didn’t run a four-minute mile—in fact I guess I can’t run a four-minute mile. But I was running a four-minute mile; I just didn’t complete it. You know, just as someone

99 Cf. Galton (1984), 86 f. 100 The point may be illustrated with a famous example: what makes it the case that someone who has counted ‘2, 4, 6, 8…’ is following the rule ‘plus 2’ as opposed to following the rule ‘quus 2’? And how might this be known?

182

Will Small

can be crossing the street but not cross the street, I was running a four-minute mile, even though I didn’t do it.’’101

The point, I take it, is not only that Falvey didn’t end up having run a four-minute mile, but that he was never running a four-minute mile, and he was unjustified in claiming that he was. But if Falvey couldn’t transform a few strides down the track into a case of running a four-minute mile by making his claim, what was it that made John’s stretch of walking a case of walking home and entitled him to claim that that was what he was doing? If we are ever to be justified in applying the present progressive to a situation—if, that is, we are ever to be justified in claiming that something is happening—then we must see our way to legitimately distinguishing between Falvey and Unlucky John, even though neither of them ended up succeeding in their tasks. We need an account that explains, on the one hand, why John was walking home and how he knew that he was, and, on the other hand, why Falvey wasn’t running a four-minute mile, and thus why his claim that he was running one didn’t amount to knowledge. But how can we explain this? The first thing to notice it that our question—how to distinguish between cases where ‘S is doing B’ is true and cases where it is false—arises regardless of whether the event in progress we consider is an intentional action or not. As Anscombe notes, of the openness of the progressive: “The point…is in no way peculiar to intentional action; for we can say that something was falling over but did not fall (since something stopped it),” and she concludes from this that “we do not appeal to the presence of intention to justify the description ‘He is Y-ing’”102. Her remark, though, also serves to develop a result we arrived at earlier, that if S is doing B, then it will be no accident if S succeeds in doing B, and it will be some accident if S fails: in Anscombe’s example, something was falling over but did not fall over because something else stopped it. Something external, accidental, to the process must enter into the explanation of how it could be that S was doing B, is no longer doing B, but didn’t end up having done B; nothing external to the process needs to be mentioned to explain how it came to be that S, which was doing B, ended up having done B. This allows us to distinguish between Unlucky John and Falvey, as we need to if we are to rebut the sceptical challenge provided by the 101 Falvey (2000), 23. 102 Anscombe (1963), §23.

Practical Knowledge and the Structure of Action

183

epistemological argument. Neither of them achieved his end ( John didn’t walk home, and Falvey didn’t run a four-minute mile). The reason John didn’t make it home is because, though he was walking home, he was involved in an accident that interrupted his progress. (You can only be interrupted doing something if you’re doing it.) But Falvey wasn’t interrupted while running a four-minute mile; the reason he didn’t run a four-minute mile is because he was never running one. What Falvey was trying to do (if indeed he was even trying to run a four-minute mile, and not simply blustering) was something difficult; something that lay beyond his power. By contrast, John’s failure to make it home is not aptly characterized as him failing to bring off a difficult task; it is best characterized as him suffering an awful interruption in the course of carrying out an easy task.103 Unfortunately, even though we can distinguish the cases in this way, we haven’t answered the sceptic’s challenge. It is true that John was interrupted and Falvey was not, but John’s being interrupted presupposes that he was doing something; we need to entitle ourselves to the idea that he was doing something, that what he was doing was walking home, and that this fact could have been known—indeed, that it could have been known by John himself. The possibility of interruption presupposes a process’s being underway; if a process is underway, then, if it is not interrupted, it will reach completion. As Sebastian Rödl puts it, “[t]he prospective end is present in the movement itself, even while it has yet to be attained”104. And this is what we need to understand, if we are to understand thoughts of the form ‘S is doing B’, and thus the form of thought in which we are primarily interested: the first-person progressive expression of intention in action, ‘I am doing B’. In other words, to justify the application of the present progressive—to legitimately judge that S is doing B—we need to know that S will do B if it is not interrupted. This is weaker than needing to know that S will in fact end up having done B, but stronger than needing to know that it is possible that S will end up having done B. 105 How can we 103 On the deep differences between the superficially similar statements ‘I would have run a four-minute mile if I could’ and ‘I would have made it home if I hadn’t been run down’, see Austin (1956). 104 Rödl (2005), 182, my translation. 105 To see this, suppose that S is putatively moving from a to s, when we encounter S at p. It is possible that S will stop short of its putative end s, and end up having moved from a to p, or from a to q ; and it is possible that S will overshoot s and end up having moved (e. g.) from a to y. More is required to know

184

Will Small

know this? How can what is happening here and now contain a tendency towards some particular end, such that it will be no accident if it accomplishes that end? What justifies the application of the concept do B to an event in progress?

4.4 Justifying the Application of the Present Progressive It depends on the value of B. If a stone is rolling down a hill, then its being no accident that it will end up having rolled down the hill is explained by mechanics: once the stone (which has a certain shape and mass) has begun rolling (at a certain speed and acceleration) down the hill (which has a certain decline), it will be no accident that it reaches the bottom—indeed, it will be no accident that it reaches the bottom at a certain time, and having followed a certain trajectory. Only if something interferes will the stone not reach the bottom of the hill. If a dog is moulting, then its being no accident that it will end up having moulted is explained by appeal to the dog’s life-form: moulting is a non-accidental feature of the life of dogs as such. (In other words, dogs moult.) We can know that our dog was moulting (and that he was about half-way through) before he met his untimely end because we know that dogs moult and what moulting is. On the other hand, it is possible to judge wrongly that a dog is moulting; somehow or other the cause of the hair loss we mistook for moulting or its onset comes to light. Similarly, it is possible to judge wrongly that a stone is rolling down a hill; if, early in its decline, the hill plateaus, then a stone’s rolling down to the plateau might be mistaken for its rolling down to the bottom of the hill. “That stone is rolling down the hill,” I say. “No,” you say, “it was only pushed hard enough for it to roll down to the plateau.” Physical, chemical, and biological laws specify types of physical, chemical, and biological processes, respectively, and explain why such processes come to completion if nothing interferes.106 It is often objected that “if nothing interferes” is but shorthand for an indefinitely long that S is doing B than that S is engaged in a process of which it is possible that the process ceases with S having done B. 106 I am abstracting from the important differences between such laws and thus such types of process—in particular, from what is distinctive about genuinely goal-directed processes—which a full treatment of the representation of movement would have to attend to.

Practical Knowledge and the Structure of Action

185

disjunction “unless p or q or r or…” the net effect of which is to render vacuous the claim about what will happen: S will do B, unless it doesn’t. However, I am in sympathy here with Rödl, who writes that “[t]he phrase ‘if nothing interferes’ denotes not an unknown content but rather a known form…. That generic statements [i.e. statements that express laws] specify what happens if nothing interferes characterizes their form, not their content”107. And Michael Thompson helpfully suggests that, [t]hough actual uses of the sign “going to V” often escape the symbol to which an apt Begriffsschrift would restrict it, so that it can be said in some cases to express simple futurity, … it is, in its primary and most interesting use, an instrument for the expression of imperfective aspect—it expresses, as we might put it, the “prospective imperfective”, and no tense at all.108

I propose that the judgment ‘S is going to do B’, understood as expressing present tense and “prospective imperfective” aspect, is equivalent to the judgment ‘S will do B if nothing interferes’ as I just explicated it— that is, where the qualification signifies the form of the judgment, not part of its content. From now on I will use ‘is/was going to…’ to express present and past tense prospective imperfective judgments, understood in this way.109 It follows that neither ‘S is going to do B’ nor the equivalent ‘S will do B if nothing interferes’ is falsified by the subsequent failure of S to do B, as long as the reason S did not do B was 107 Rödl (2005), 197 f., my translation. Cf. Rödl (2005), 182, my translation: “The progressive statement looks beyond the here and now. It does not look forward, backward, or sideways; it directs its gaze upwards to what happens in general”. It is clear that what happens in general—what a law says happens—does not always happen, because a process exemplifying what happens in general—a process exemplifying a law—may be interrupted, in which case it will stop without finishing. This shows that the generality of statements of what happens in general cannot be understood through the device of universal quantification. But it cannot be understood through some special kind of generic quantification, either: if the intelligibility of statements exhibiting progressive and perfective aspect depends on the generic statements they exemplify, then the generic statement cannot in turn be understood as quantifying over such statements. As Rödl puts it, “[t]he generality of generic statements characterizes not the quantity of the subject but rather the way in which subject and predicate are conjoined” (Rödl (2005), 192, my translation). 108 Thompson (2008), 142. 109 Just to be clear: I am not claiming that typical utterances in English of the form ‘S is going to do B’ necessarily express prospective imperfective judgments, and thus admit ‘S was going to do B but never did’; ordinary language is used for much more than doing philosophy.

186

Will Small

that something interfered and prevented S from doing B. We can judge, after the fact, that S was going to do B, that S would have done B, but that S didn’t do B because something interfered. Of course, if someone judges that S is going to do B, but in fact S does not do B, then what he judged was going to happen did not happen; but whether or not his judgment was true or false depends not on this, but on whether S would have done B were it not for interference. By contrast, even if S would have done B had the interference not occurred, the fact that S did not do B falsifies an earlier statement of “simple futurity”—that is, where subject and event-predicate are joined with future tense and perfective aspect—‘S will do B’.110 Thus, though the thought that S will do B (simple future) is not contained in the thought that S is doing B, the latter nevertheless contains the thought that S is going to do B (prospective imperfective). What makes it the case that what S is doing amounts to doing B is that, in doing what it is doing, S is going to do B. All knowledge of what is happening is, for this reason, knowledge of what is going to happen; knowledge, that is, of what will happen if nothing interferes, not knowledge of what simply will happen. By contrast, to know what will happen is to have knowledge of what is going to happen plus the knowledge—hard to come by, this—that nothing will interfere.111 110 Cf. Comrie (1976), 64 f.: “It is important to appreciate the difference between … expressions of prospective meaning and expressions of straight future time reference, e. g. between Bill is going to throw himself off the cliff and Bill will throw himself off the cliff. If we imagine a situation where someone says one of these two sentences, and then Bill is in fact prevented from throwing himself off the cliff, then if the speaker said Bill will throw himself off the cliff, he was wrong, his prediction was not borne out. If, however, he said Bill is going to throw himself off the cliff, then he was not necessarily wrong, since all he was alluding to was Bill’s intention to throw himself off the cliff, i. e. to the already present seeds of some future situation, which future situation might well be prevented from coming about by intervening factors. Indeed, Bill is going to throw himself off the cliff might well be shouted as a warning to some third party to prevent the future situation from coming about”. Our question, of course, is what it would be for the seeds of some future situation to be present. It is clear that it cannot always be intention, for The house is going to burn down and The house will burn down (both said while the house is burning) exhibit the same contrast; and even if Comrie is right that Bill’s intention is that wherein those seeds of the future presently lie, it remains to be seen what intention must be if it is to function in that way. 111 Cf. Anscombe (1963), §52; Hampshire (1965), 65 and Galton (1984), 110.

Practical Knowledge and the Structure of Action

187

4.5 The Temporal Structure of Intentional Action Is Explained by Its Calculative Structure On this picture, then, we distinguish between a case in which S is actually doing B from one in which S is only apparently doing B but is in fact doing B’ by apprehending what is happening as exemplifying a law that is part of a system of laws (e. g. chemical laws) that together determine a generic kind of process (e. g. chemical process); the law that what S is doing exemplifies and through which it is understood specifies what doing B is and explains why S is going to do B. 112 However, when I start intentionally walking to school, even though I have a certain mass, and am moving at a certain speed, it is not these physical facts and the kind of mechanical explanation they are fitted to enter into that explains why it is no accident if I end up having walked to school intentionally. It is not as if once I start walking to school, I am carried there by something like gravity. Nor is there a chemical or biological explanation. In fact, there aren’t any laws of the relevant kind about people walking to school, or even about students walking to school, as there are about dogs moulting.113 When I am walking to school intentionally, I am not actualizing a tendency that is described by a natural law and which law thereby explains, and is exemplified by, what is happening. Rather, what I am actualizing, or rather executing, is my intention to walk to school. Thus, what explains why it is no accident if I end up having walked to school intentionally is that (i) I have knowledge in intention that I am walking to school, that is, I intend to walk to school, which as we have seen involves (ii) knowing why I am making my way to school—to return a book to the library—and (iii) knowing how I’m getting there—I know I’m getting there on foot, I know how to get there on foot (I know the route), and I know how to get about on foot (I know how to walk). 112 How this works in detail for various kinds of processes is an interesting question that deserves a fuller discussion; but it would be out of place here, where our concern is what kind of knowledge of the future is present in an agent’s knowledge in intention. 113 Of course, there might be statistical laws about, say, students walking to school. But a statistical law cannot explain the temporal unity of the events that exemplify it, because it generalizes over events that are antecedently comprehended (cf. fn. 107 above). A law of the relevant kind would explain the temporal unity of the events that exemplify it. Perhaps the laws that, according to Kant, we “give to ourselves” do this; unfortunately I cannot pursue the question here.

188

Will Small

This knowledge in intention is what accounts (i) for the fact that my walking to school has a definite end-point (the returns desk in the library), and so I know what will count as my having finished walking to school, my being half-way done, and so on, and (ii) for the fact that the parts and phases of my walking-to-school process amount to one process that luck and accident might interfere with, but which they do not constitute.114 We can reach the same point by reflecting on the concept of a means, as it figures in the calculatively-articulated knowledge in intention that permeates the intentional action. The intention in action is articulated, and articulates itself, into means and ends, and it is internal to the idea of the means–end relation that the means are ( jointly) sufficient for the production for the end. Thus, the knowledgeable comprehension of the means as sufficient to produce the end, which is internal to the calculative articulation of intention in action, is thereby a knowledgeable comprehension of the fact that it will be no accident if one ends up intentionally producing the end. For, if it would be some accident that the end would be produced, then this would show that the means are insufficient, and that any thought of them as being sufficient could not have been knowledge. What I have said in this paragraph and the last concerning intention in action (about what explains why it will be no accident if I end up having walked to school intentionally when I am walking to school intentionally) can be straightforwardly extended to the case of intention for the future, to explain why it will be no accident if I end up having walked to school intentionally when I intend to walk to school (when, that is, I am going to walk to school intentionally): in this case too, the calculative articulation of the intention for the future (which may still be in potentia) shows why it will be no accident if the intended end is produced. Moreover, this account also explains why there is another way in which an intention (in action or for the future) might end up not being executed, other than if something external interferes: the agent 114 Recall from fn. 64 above the contrast between, on the one hand, my heart’s pumping because it is circulating blood round my body, the teleology and temporal unity of which is explained by the generic statement ‘the heart pumps blood in order to circulate blood round the body’, and, on the other, the pumper’s pumping because he is poisoning the inhabitants of this house, the teleology and temporal unity of which is explained not by the (false, or perhaps unintelligible) generic statement ‘the pumper pumps in order to poison the inhabitants of the house’, but rather by the pumper’s intention.

Practical Knowledge and the Structure of Action

189

may change his mind. If something external interferes, then there is a real process underway, or that is going to be underway when the time comes and towards which there is thus already a real tendency, that is prevented from reaching completion by something accidental to it. If the agent changes his mind, then the process (or tendency towards it) simply is no more; the real tendency towards the completion of the action has not been interfered with, but abolished. As with the “qualification” if nothing external interferes, this possibility is properly represented in the form of the agent’s knowledge rather than the content. However, unlike that “qualification”, which belongs to the form of knowledge of what is going to happen (knowledge of movement in progress or in prospect), the “qualification” unless I change my mind belongs specifically to the form of knowledge in intention (of intentional action in progress or in prospect), reflecting intention’s distinctively selfconscious causality. We saw, in §2.5 above, that the distinction between doing C by doing B accidentally and doing it non-accidentally was explained by the source of the relevant teleology: in a machine’s design, in an organism’s life-form, in an agent’s calculatively-articulated intention. Here, we see that the non-accidentality of the accomplishment of the end specified by do B, or equivalently, the temporal unity of an event falling under the concept do B, has a similar kind of underlying explanation. Mechanical laws underwrite the temporal unity of mechanical events; chemical laws underwrite the temporal unity of chemical events; biological laws underwrite the temporal unity of biological events.115 The design of a machine underwrites the temporal unity of the actions it carries out. And intentional actions, through their calculative articulation, underwrite their own temporal unity: the calculative structure of an intentional action constitutes its temporal structure. So while Anscombe was right to insist that the presence of intention is not what explains the legitimacy of present progressive judgment in general, it in fact turns out that when the concept do B is an intentional action concept, the legitimacy of its present progressive predication does depend on the presence of intention in the agent. If an agent is doing B intentionally, then he is going to do B, where this consists in his intend-

115 Cf. Rödl (2002), §3; Rödl (2005), 188 f. and Galton (1984), 142.

190

Will Small

ing to do B, of which intention his doing B intentionally is the execution (in progress).116 We are now in a position to distinguish properly between Falvey and Unlucky John. What makes it the case that John was walking home, not walking to his neighbour’s, is that he was walking home intentionally; he intended to walk home and he was executing that intention, and thus he knew that what he was doing was walking home. By contrast, the explanation of why Falvey wasn’t running a four-minute mile, though he may briefly have seemed to be, is that he wasn’t in a position to have the relevant intention, let alone execute it. In our earlier discussion of intention and aspiration (§3.1), I said that an intention to do B requires knowing how one can do B (that is, general knowledge how to do B) though it need not require knowing how one shall do B 116 It follows from these remarks that if S is doing B intentionally, then S intends to do B, which would presumably count as a version of what Michael Bratman (Bratman (1987), 111 ff.), calls “the Simple View”, the truth of which is much debated in the literature. I cannot engage that debate here; it must suffice for now to say that the account of the calculative and temporal structure of intentional action (in progress and in prospect) that I have been developing here constitutes what I take to be the deep ground for a defensible version of the Simple View. However, it would be a mistake to read me as claiming here that we are to understand intentional action as the execution of a calculatively-articulated intention, where the latter notion is antecedently understood. Such an account would be received as unproblematic by many contemporary theorists, but our reflection on the temporal structure of action suggests that it should not be. The difficulty arises when considering how to represent the object of desire or intention, as Matthew Boyle and Douglas Lavin point out: “to want to do A is not merely to want to be in some terminal state…but rather to want what is essentially a goal-directed course of action: ‘enough A-ing to have A-ed’. …When I want to do A, in short, the content of my want is of a form such that the world can only come to conform to that content insofar as it not only comes to be a certain way, but does so as the outcome of a goal-directed process guided by the agent. …To represent my doing A [i.e. as the content of a desire] is to represent, as it were, a kind of state of affairs whose obtaining is my having intentionally caused it to be” (Boyle/Lavin (2010), 173). What Boyle and Lavin say of wanting can be said of intending, too: we have no idea what it is to intend to do something without knowing what it is to do something. Thus we cannot appeal to intention to explain the possibility of intentional action; cf. Anscombe (1963), §4. So, although I have argued that we cannot understand ‘S is doing A’ independently of ‘S intends to do A’ (where do A is an intentional action concept), I take it that converse holds too: we cannot understand ‘S intends to do A’ independently of ‘S is doing A’. Hence, on the view I am advancing, the two forms are interdependent.

Practical Knowledge and the Structure of Action

191

(that is, particular knowledge how one is going to do B on this occasion), though the intention understands itself as needing to become further calculatively-articulated so as to settle this question when (or before) the time to do B comes. Now Falvey cannot run a four-minute mile, and so he cannot know how he can run a four-minute mile; in other words, he doesn’t know how to run a four-minute mile, and so cannot intend to run one.117 Thus he is like Setiya with respect to dancing the tango; right now, Falvey can at best aspire to run a four-minute mile, but all he can intend, at the moment and in this respect, is to train (indeed, if he seriously aspires to run a four-minute mile—if it is not a mere wish—then he had better intend to train). One might wonder whether we need to say all this; after all, if Falvey cannot run a four-minute mile, doesn’t this suffice to explain why he wasn’t running a four-minute mile? (One cannot do what one cannot do, and thus one cannot be doing what one cannot do.) 118 Why bother taking this detour through whether or not Falvey was in a posi117 Some, e. g. Snowdon (2003), will want to resist this suggestion, and claim that what distinguishes Falvey from someone who can run a four-minute mile is not the latter’s having more knowledge, but the latter’s having more power. I discuss the relation between practical powers and knowledge how in §§5 – 7. 118 Care must be taken with this point. The principles (i) that it is possible to have been doing B intentionally, but yet not end up having done B (and not through change of mind), and (ii) that one cannot be doing what one cannot do, are consistent only if we deny the superficially plausible principle (iii) that failure to do B if one tries suffices to show that one cannot do B. At the very least, (iii) must be modified as follows: one’s failure to do B in particular conditions C even though one tried suffices to show that one could not do B in those conditions (i. e. conditions in which something interfered). But against even this, I sympathize with J. L. Austin, when he writes: “Consider the case where I miss a very short putt and kick myself because I could have holed it. It is not that I should have holed it if I tried: I did try, and missed. It is not that I should have holed it if conditions had been different: that might of course be so, but I am talking about conditions as they precisely were, and asserting that I could have holed it. There is the rub. Nor does ‘I can hole it this time’ mean that I shall hole it this time if I try or anything else: for I may try and miss, and yet not be convinced that I could not have done it; indeed, further experiments may confirm my belief that I could have done it that time although I did not. …According to [the traditional beliefs enshrined in the word can], a human ability or power or capacity is inherently liable not to produce success, on occasion, and that for no reason (or are bad luck and bad form sometimes reasons?)” (Austin (1956), 166 n.1). See also the discussion of fallibility in §5 below.

192

Will Small

tion to intend, and thus whether he knows how, to run a four-minute mile? The answer is this. It seems that there is no such thing as having run a four-minute mile but only through luck. However, the possibility of merely lucky success is a feature that permeates the landscape of intentional action: most types of intentional action are such that an agent might “perform” a token of the type, but, because of the way in which his “success” depended on luck, it is commonly accepted, he did not do so intentionally (under the relevant descriptions). It is easy to imagine scenarios in which agents have replenished a water supply or walked to school or made a three-point shot or baked a cake or become rich or pulled off a coup or a heist only through luck; their plans and skills—their knowledge how to do the things they did—were either inadequate to their goals as the situations played out or, though they were adequate, it was not in fact through them that those goals were achieved. By contrast (though perhaps this reflects a failure of imagination on my part), it is hard to see how a piece of behaviour could attract the description “running of a four-minute mile” while resisting the description “intentional running of a four-minute mile”, because (e. g.) being carried along quicker over the final 100 yards due to a large gust of wind would, I suppose, put pressure on the applicability of “running”. Falvey’s example is unusual in this respect. In “lucky success” cases, the agents “did” what they “intended” or “aspired” to do, but that their doing those things depended substantially on luck leads us to judge that their actions were not intentional (under the descriptions that make reference to the goals). Now we know why this is so: the calculative articulation of practical thought is the rule that runs through an action; it is that which thereby constitutes the action’s temporal unity and which accounts for its tendency towards completion. If there is nothing that accounts for the tendency towards completion, then nothing answering to the relevant description (the one that makes reference to the point of completion) is happening; if there is such a tendency, but something other than a sound, calculatively-articulated practical thought (i. e. an intention) accounts for it, then what is being done is not being done intentionally, and if it ends up having been “done”, or having “come to pass”, it will not be an intentional action; if there is action at all in such a case it is because the agent did something intentionally, but not what he aspired to do, and the coming to pass of that to which he aspired must be traced to an origin external to his practical thought (and powers). Thus, if someone who aspires to

Practical Knowledge and the Structure of Action

193

do B brings it about that he has done B, we can ask two questions: was the present progressive judgment ‘He is doing B’ ever true? And, if it was, what explained it? If its truth is explained by the agent’s sound practical thought, we have a case of intentional action; otherwise, we have something less than that—though how much less will depend on how far away from being an intention the agent’s aspiration was.119

4.6 The Cognition Condition Reformulated Again I have argued that someone who is doing B intentionally, or who intends to do B in the future, is going to do B. Even though it is possible that he will not do B, it will be no accident if he ends up having done B. This is likely to be met with scepticism, for it follows by contraposition that if the agent is not going to do B, then neither is he doing B intentionally nor does he intend to do B. The distinction drawn between the prospective imperfective ‘is going to do B’ and the simple future ‘will do B’ shows that this is not absurd, but it will be suspected that it renders the point trivial. At the very least, it will surely seem as if the account I have offered must fail to do justice to what many think of as the central insight of Anscombe’s claim about agent’s knowledge: the idea that an expression of intention says something not just about the speaker’s state of mind, but about what is happening or going to happen in the world—just as might an expression of belief. That someone who is doing B intentionally is going to do B (going to end up having done B) is due to his intention to do B, I said; yet it is compatible with 119 As I emphasized in §3.1 (see especially fn.75), the distinction between intention and aspiration is not all-or-nothing: the closer the serious aspirer comes to meeting the conditions of intention, the more clearly we can, so to speak, discern the signal of the rule in the noise of what is happening, and the closer what is happening comes to being an intentional action (under the relevant description). By contrast, an agent who purports to aspire to do something might reveal himself to have something more like a mere wish or fantasy if he is not in business of trying to acquire the kinds of plans and skills the lack of which prevent him from being able to form the relevant intention. (It is tempting to read Falvey as depicting himself as such a fantasist, rather than a serious or even incipient aspirer; however, one could surround the example with further details that would push it in one direction or another.) The role we credit luck in the success of a serious aspirer ought to differ from the role we assign it in the case in which the fantasist “succeeds” in “doing” what he was “trying” to do. Thanks to Jim Conant for helpful discussion here.

194

Will Small

this that he ends up not having done B. Thus, the objection runs, the successful prosecution of the action is present merely in the agent’s mind as an object of intention; the only kind of knowledge to be had here is knowledge of that state of mind, not knowledge of what will happen in the world. This is so regardless of any linguistic sleights of hand I might have made regarding stipulative uses of the expressions ‘is going to do B’ and ‘will do B’. If it turns out that knowledge of what is going to happen is not, as such, knowledge of what will happen, but rather knowledge merely of what the agent intends to do, then the spirit of Anscombe’s claim has been abandoned, whether or not lip service has been paid to the letter. The objection will not do, however. Consider a parallel case: a boulder has begun to roll down a hill. We know that it is going to roll down the hill to the bottom, that it will reach the bottom if nothing interferes (we do not know that nothing will interfere). There is a real process underway here; real tendencies, which are systematically captured by the laws of mechanics, are being actualized. This is not something “merely subjective”, something not about the world but merely concerning the mind. Likewise, when someone has begun to walk to school intentionally, this too is a real process underway, a process which to be understood as the very process that it is must be understood as one that will terminate in that person’s having walked to school unless something interferes. That we are not certain that he will end up having walked to school does not make it the case that our claim that he is going to walk to school expresses something only about his or our states of mind.120 One might be prepared to grant this, however, and yet resist the parallel, and seemingly stronger, claim concerning intention for the future: that knowing that someone who intends to go shopping at the weekend is, by that very fact, going to go shopping at the weekend amounts to knowledge of anything more than his state of mind. But such resistance amounts to a refusal to acknowledge the fact that intention is efficacious. It is true that if you know that someone wishes she could fly to the moon or believes that Hume died in 1776 then what you know is something about that person’s state of mind and not (sim120 Cf. Anscombe: “If I say I am going for a walk, someone else may know that this is not going to happen. It would be absurd to say that what he knew was not going to happen was not the very same thing that I was saying was going to happen” (Anscombe (1963), §52).

Practical Knowledge and the Structure of Action

195

ply as such) something about the world and what is going to happen in it. This is because neither mere wishes nor beliefs are practical thoughts: they are not as such efficacious. But an intention, as distinct from other species of desire such as aspiration and more-distantly related acts of mind such as mere wish, is sufficient to produce the intended act, as being moved with a certain degree of force is sufficient (albeit in a specifically different way) for a boulder of a certain mass, size, and shape to roll down a hill of a certain decline. Assuming that what the agent intends is to do something “worldly”—like go shopping or walk home or write a novel—and not merely “mental”—like sing a song in her head or imagine what it would have been like to be Napoleon—then knowledge of her intention is knowledge of a real tendency towards something’s happening in public reality. Our knowledge of what someone is going to do intentionally is no less objective and worldly than our knowledge, of an event of any kind, that it is going to happen. But does the agent who is doing B intentionally need to know that it’ll be no accident if she ends up having done B? Isn’t it enough that it’ll be no accident—why should she need to know it? One might worry that the account I have offered amounts to an implausibly intellectualized conception of agency, one that supposes that an agent somehow grasps the complicated and quite possibly tendentious metaphysical views outlined in the present section; in other words, one might worry that I am supposing that one must become a philosopher of action before one can act intentionally.121 But this would be a mistake. I have tried to situate my account of the temporal unity of intentional action within a broader framework, one which reveals that intentional action is a distinctive species of event in virtue of the form that the temporal unity of intentional actions takes. But the specific form that this unity takes is simply the calculative structure that belongs to intentional action insofar as it is, as it always is (at least in potentia), articulated into means and ends. The idea that someone who is doing B intentionally knows that she is going to do B, that it will be no accident if she ends up having done B, is just this: she knows that she is doing B by doing A, where this is the ‘by’ of calculative unity, and thus she knows that doing A is as such sufficient for doing B, and thus, in virtue of this sufficiency, that it will be no accident if, in doing A, she ends up having done B. Conversely, to suppose that it could be no accident that she would end up having done B despite not herself knowing that this was so would be to suppose that its being no 121 Thanks to Anton Ford for pressing me to address this objection.

196

Will Small

accident derived from something other than her own practical thought, and thus to suppose that she is not doing B intentionally. We can now conclude our investigation into the temporal structure of action. If an agent is doing B intentionally, then she is going to do B: it will be no accident if she ends up having done B, and she will end up not having done B only if she is prevented or changes her mind. That her success will be no accident is explained by her intention; if someone merely aspires to do B, and brings it about that she has done B, this is not an intentional action. For her to be doing B intentionally, she must know that she is doing B intentionally, and thus she knows that it will be no accident if she succeeds. That is, she knows she will succeed if she is not prevented. The same goes for intentional action in prospect as for intentional action in progress: if an agent intends to do B on Tuesday, then she will do B on Tuesday unless she is prevented (or unless she changes her mind, in which case the antecedent is no longer satisfied and thus her doing B on Tuesday is no longer on the horizon). It may be objected that she does not know with absolute certainty that she will succeed; but neither does she know with absolute certainty that the sun will rise tomorrow—for that event may be prevented, too. Anscombe’s intuitive idea with which we began in §1.1—that if I am doing B intentionally then I know (without observation) that I am doing B—has been revealed to be both more complex, and more deeply grounded in the structure of intentional action, than we might at first have thought. An intentional action is calculatively and temporally structured, and thus so must be its representation. This is so whether the action represented is in progress or in prospect. The calculative structure, which belongs essentially to the agent’s practical thought, underwrites the temporal structure, which thereby too belongs to the agent’s practical thought. If I am doing B intentionally, then I am doing it by doing A intentionally and because I am doing C intentionally, and for each of A, B, and C it will be no accident if I succeed in bringing them off.122 Moreover, this calculative and temporal structure 122 Two qualifications. First, I am assuming that my doing B is not a basic action, of which there will be more discussion below. Secondly, it is possible to do B intentionally because one takes oneself to be doing C intentionally, and yet one is not doing C intentionally either because one is not doing C or because one is doing C but only through luck (i. e. either the Realizing Condition or the Non-Accidentality Condition might not obtain). (A parallel situation is one in which one is doing, or intends to do B, because one aspires to do C.) If one is not doing C intentionally, where this is not just because one is going

Practical Knowledge and the Structure of Action

197

is what I have knowledge in intention of, where this is practical knowledge: the structure obtains precisely because it is so known. This is what it is for an expression of intention in action ‘I am doing B intentionally’ to be true. And likewise for an expression of intention for the future ‘I am going to do B’.

5 The Necessity of the Cognition Condition 5.1 Davidson’s Objection Reconsidered We have derived a more adequate understanding of the cognition condition from its source in the calculative and temporal structure of intentional action (in progress and in prospect); now we are in a position to respond to those examples, like Davidson’s carbon-copier, that purport to show that one may be doing something intentionally, or intending to do something, while remaining agnostic about whether one is doing it, or going to do it. (We saw in the previous section how and why one’s knowledge in intention that one is doing A contains knowledge of eventual success—knowledge that one is going to end up having done A.) To defend the cognition condition I have outlined, then, I must show that either the carbon copier does know that he is making the copies while he is intentionally making them, or, if he doesn’t know that he’s making them, he isn’t making them intentionally. Now, as it stands, Davidson’s carbon-copier case is underdescribed (this is unsurprising—it originally appeared as a parenthetical remark). Happily, though, when we flesh it out, either it ends up being a case of intentional action that satisfies the cognition condition, or it ends up not being a case of intentional action. The relevant kind of fleshing out that is required consists, unsurprisingly, in providing the calculative to do C but is not doing it yet, then one does not have knowledge in intention that one is doing C, or that one is going to do C (in the relevant sense in which that knowledge is, or rather would be, contained in the knowledge that one is doing C). Does this impinge upon whether or not one is doing B intentionally, or whether or not one knows that one is? It depends—on whether or not the error in one’s practical thought or performance in virtue of which one, though doing B intentionally, is not doing C intentionally is grievous enough that it impugns the status of C as the measure by which one assesses and regulates one’s doing B.

198

Will Small

structure within which we are to situate, and thus grasp, the copier’s action and its temporal unity. First, consider the following mundane expansion of the story. The copier needs to produce ten copies of a document to give to his boss. He reasons that the fastest way to do this would be to interleave carbon paper between the ten pages, and make the copies all in one go. So he does this, and he presses very hard on the pages as he writes. Suppose that when he inspects the documents, he finds that the impression carried through to the seventh copy, but not beyond. What happens next? Remember, the copier needs to give ten copies to his boss; thus, having made only seven copies, his inspection reveals not that he has failed but that he hasn’t finished yet. (It is not as if, not having made all ten copies in one go, he won’t be able to give ten copies to his boss.) So he places the seven copies he has made to the side, and writes again on the eighth, again pressing firmly, so that the impression carries through to the ninth and tenth pages. He has now made ten copies, intentionally; and all the while, from beginning to write on the top sheet until checking the tenth impression, he was making ten copies. Moreover, he knew that this is what he was doing; his practical knowledge that he was making ten copies, not seven, is what sustained his action through to the next phase: the calculative unity of his action constitutes its temporal unity. It’s true that he was initially trying to make all ten copies at once, but the calculative and temporal context of his action—importantly, the facts that he needed to make ten copies, that it didn’t really matter whether he did them all in one go or not, and that he continued to do things with the intention of making more copies until there were ten of them—reveals that this was just one possible means he could have taken. As it happened, the means he chose only amounted to getting seven-tenths of the job done; so what he thought was the whole of his action turned out to be a proper part. But this is no objection to maintaining that while he was doing the parts, he was doing the whole, and thus no objection to maintaining that while he was doing the parts, he knew he was doing the whole.123 123 One might worry that this vindication of the claim that the carbon-copier, in making ten copies intentionally, thereby knows that he is making ten copies, fails to show that his knowledge is (all) in intention, and that it is not derived, at least in part, from observation. Perhaps agent’s knowledge consists of two factors, non-observational knowledge of what one intends to be doing, and observational knowledge of what one is actually doing. (Donnellan (1963) advan-

Practical Knowledge and the Structure of Action

199

What happens if we flesh out the case in a way where it specifically matters that the copier not only produces ten copies, but that he produces all ten copies in a single act of carbon-copying—a case, say, in which his job or his life hangs on his being able to perform this out-of-the-ordinary feat? In such a case, if the first inspection reveals that only seven copies have been made, then this does amount to the revelation that the copier has failed, and not merely that he hasn’t finished yet, because what determines whether or not he has finished is not his having made ten copies, but his having finished writing on the top sheet (once). But it’s not clear that a copier under such a strange and unusual demand is in a position to intend to make all ten copies at once. He can aspire to, perhaps (though it strikes me that even the concept of aspiration is being pulled out of shape in such an example); he can give it a shot and hope for the best. But if you hope to make ten copies and you bring it about that ten copies are made, you haven’t necessarily acted intentionally. If you bring it about that ten copies are made through intention, or—what comes to the same thing—through knowledge in intention, then this is intentional action; if you bring it about through luck, it is not. So if the copier doesn’t have knowledge in intention that he’s making ten copies, because he’s not in a position to genuinely intend to make ten copies, then he is not making ten copies intentionally, even if he ends up having brought it about that there are ten copies of the document.124

ces such a view; see Falvey (2000) for criticism.) But this would be a mistake: when the copier observed that he had only made seven copies, he was not observing what he was doing, but rather what he had done. As Falvey puts it, “Observation enables the agent to notice and correct mistakes; but what counts as a mistake here is determined by what the agent is doing” (Falvey (2000), 29); observing a mistake presupposes knowing what you’re doing, which knowledge therefore cannot be observational. Cf. in this context Anscombe’s remarks on the role of the eyes as “an aid” in writing (Anscombe (1963), §29), and her discussion of the “two knowledges,” one practical, the other speculative, involved in any intentional action (Anscombe (1963), §48). 124 Cf. Thompson (2011), 209 f; but compare his remarks, in ‘Naive Action Theory’, on whether intending to do something involves thinking that one will end up doing it (Thompson (2008), 102 f, 145 n.29); or whether asserting ‘S is doing B’ in any way commits the speaker to the eventual truth of ‘S did B’ (Thompson (2008), 144).

200

Will Small

5.2 Failure and Fallibility One way to take Davidson’s carbon-copier example would be as material for a sceptical argument, an action-theoretical parallel to the argument from illusion. According to this argument—call it the argument from failure—an agent who is doing A intentionally does not know simply in virtue of his practical thought and agency that he is doing A; it “practically seems” to him as if he is doing A—that is, he is trying to do A and he knows this—but it would be unjustified of him to infer from the way things seem to the way they are, because when an agent tries and fails to do something, things are for him, from the point of view of his practical thought and agency, just as they would be if he succeeded. Knowledge that he is actually doing what he intends to be doing or is trying to do would require supplementing his practical thought with something extra-practical: observational knowledge, say, of what he is doing. If the carbon-copier ends up having succeeded in making his copies, then it will have been the case while he was making the copies, Davidson tells us, that he was making them and making them intentionally. But while he is making them, supposing that he is in fact making them, things are for him as they would be if he were merely trying to make them; thus, all he knows is that he is trying. But the argument is too quick. To see this, return to our second expansion of the case, where the carbon-copier tries but fails to make his ten copies all in one go. Consider what was true when he was doing whatever he was doing in order to make the copies—pressing very hard with his pen, say. Plausibly, it is true that he was then trying to make the copies, but it is not true that he was then making the copies. Contrast Unlucky John, considered before his accident: he was then actually walking home. (Perhaps it is also true that he was trying to walk home, but this should not detain us.) Our reflections in §3.1 and §4.4 above have put us in a position to distinguish between two different notions of “failing to do something”: (i) being either prevented from doing something one had a genuine intention to do or interrupted in the course of intentionally doing it, and (ii) not bringing about a state of affairs or effect that one merely aspired to bring about, which, with respect to the relevant description, one was never doing intentionally. The failed copier’s failure is roughly of a piece with Falvey’s type-(ii) failure to run a four-minute mile, not of a piece with Unlucky John’s type-(i) failure to make it home; and, as we have seen, bringing about a state of affairs or effect that is the content of a mere aspiration is not acting intentionally,

Practical Knowledge and the Structure of Action

201

under the relevant description. The argument from failure simply equivocates between these two notions of failing to do something. Trying to base an argument from failure against the possibility of practical knowledge on the fraughtness of difficult tasks like making ten carbon-copies at once or running a four-minute mile is like trying to base an argument from illusion about perceptual knowledge on the difficulty in distinguishing far-away square towers from far-away round towers: no one should be convinced by either.125 However, my remarks about the second way of fleshing out the carbon-copier case might make it sound as though actions that involve doing something in one shot, so to speak, can never be intentional. For what made it the case (in the first variation) that the copier was making ten copies the whole time was that it didn’t matter whether he made them all in one go and he knew this and knew how to cope with fact that, as it turned out, his first pass resulted in the production of seven copies. But surely it is possible to intend to sink a birdie putt, and indeed to sink one intentionally, even though if you miss it the best you can then do is make par; there are a whole host of kinds of actions that depend on being done in one shot, or might depend on it in particular circumstances. And it might seem that I have committed myself, implausibly, to saying that, if someone missed a birdie putt because of an error of performance—not because someone yelled while she was striking the ball, or because a gull swooped and made off with the ball while it was rolling towards the hole, but just because she mis-hit it—she thereby reveals herself to have had merely the aspiration of making a birdie, and the intention (of underwhelming ambition) to just finish the hole, however many strokes it takes. Saying this would be a mistake, but the conception of the cognition condition on intention and intentional action I have been advancing 125 If you “try” to do something and fail, it may or may not be clear to you or anyone else whether you failed to “do” it because you were prevented or because you weren’t in a position to do it intentionally. (The cases of temporary bodily paralysis that are popular in the literature can I think be construed, with enough back story, either way; I doubt that whatever intuitions we might have about such esoteric cases amount to data that an account of intention, action, or trying is under an obligation to respect.) The same thing goes for trying and succeeding, as I have already emphasized (§3.1 and §4.5 above): there is a difference between bringing about one’s goal through luck and bringing it about through knowledge (that is, between lucky action and intentional action); sometimes it is hard to make out in practice, and difficult cases can be argued about.

202

Will Small

does not require it. The birdie-putt case brings out a third type of failure of action, in addition to types (i) and (ii) above: (iii) failing to do what one intended to do because one’s action is the exercise of a fallible capacity. (It is worth pointing out here that though we have been led to the topic of fallible capacities to act by a worry arising from the specific response I gave to Davidson’s carbon-copier case, providing an account of this topic ought to be a general requirement on any account of intentional action and practical knowledge. Of course, the brief remarks that follow do not constitute a full account.) That the skills of putting, shooting a basketball, and so on are fallible is obvious to anyone who possesses or admires them; but we can and sometimes do make mistakes in every area of intentional action. Our ability, in many cases, to “pick up” an intention that has “fallen to the ground” means that we can, in such cases, overcome our mistakes and do, and know that we are doing, what we intend to do. But it would be wrong to infer that, when the possibility of picking up the intention and carrying on is excluded due to the one-shot nature of what one intends to do, one’s exercising a fallible capacity successfully counts as a piece of luck that undermines the status of one’s action as intentional and one’s practical thought as an intention. That a skilled golfer might have missed her putt, placing a birdie irrevocably beyond reach, does not impugn the status of the putt that she holes through the exercise of her skill: that her capacity is fallible does not mean that every exercise of it is flawed. This is not to say that a skilled golfer can never be said to have been lucky in making a putt—she might hit a putt slightly off-line, due to a flawed exercise of either putting judgment or execution, that nevertheless “lips in” when it might as well have “lipped out”—but only that she should not always be said to be lucky in this sense. (Keep in mind, as always, that the distinctions between intention and aspiration, between intentional action and making things happen merely by luck, are not allor-nothing.) We must not conflate the first and third types of failure (prevention/interruption and flawed exercise of a fallible capacity, respectively): the world may, from time to time, “do her a favour” by, so to speak, permitting what ought to have been insufficient to get the job done, but the only sense in which it does her a favour every time she succeeds is by not permitting e. g. a gull to fly off with the ball. Indeed, it is part of what a skill is that it provides (fallible, yet resilient) security against particular points of vulnerability that beset the kinds of action and activity for which the skill is useful; not only is it

Practical Knowledge and the Structure of Action

203

no accident that when an agent who possesses the skill of doing A exercises that skill, she ends up having done A, it is no accident if she ends up having done A despite the presence of certain typical hampering factors, which might interrupt or prevent an agent who possessed the skill to a lesser degree. 5.3 Why not to Weaken the Cognition Condition This might seem like a lot of fuss to make. Why shouldn’t we just accommodate apparent counterexamples along the lines of Davidson’s carbon-copier, by weakening either the content or the scope of the cognition condition? Although space does not permit me to go into the details of all of the various proposals made on this score, we are already in a position to see that the basic answer is this: if we suppose that the carbon copier could be making ten copies intentionally without knowing that he is doing so—either because he knows only that he is trying to make ten copies (Davidson), or that he is doing something “more basic”, like writing firmly on the top sheet with the aim of making ten copies (Setiya), or because it’s just an atypical case (Gibbons)— then we are simply taking for granted the calculative and temporal unity of the action while at the same time denying the presence of that which could account for it, the agent’s knowledge in intention. Anscombe, commenting on her deployment of Aquinas’s slogan, “practical knowledge is the cause of what it understands”, writes that it “means more than that practical knowledge is observed to be a necessary condition of the production of various results; or that the idea of doing such-and-such in such-and-such ways is such a condition. It means that without it what happens does not come under the description—execution of intentions—whose characteristics we have been investigating”126. The cognition condition is woven into the very fabric of the intentional action that is cognized in agent’s knowledge. The pull of the carbon-copier case is felt only if we imagine that the structure of action will take care of itself, independently of the agent’s knowledge of it; but because her knowledge is constitutive of the action, calculatively and temporally, we can’t simply take the action for granted and then ask whether or not she has knowledge of it, and if so how. Thus reflection on the calculative and temporal structure of intentional action reveals both why we must and how we can reject the idea that carbon-copier 126 Anscombe (1963), §48.

204

Will Small

style cases pose a threat to the cognition condition, and therewith reject the idea that they must be accommodated by weakening the condition’s content or scope. 5.4 Practical Knowledge? However, even if we should not weaken the content or scope of the agent’s cognition, perhaps we should weaken its kind. Recall that contemporary cognitivism’s appropriation of Anscombe’s thought proceeds in accordance with a programme of epistemological “modesty,” as Setiya puts it, of “setting aside the claim to knowledge”127 in favour of a claim about what the agent believes. In §1.3, I expressed some reservations, concerning the risk of “theoreticizing” practical thought, about characterizing the agent’s cognition that may or may not hit the heights of knowledge as a belief. In any case, the point of weakening the cognition condition from knowledge to a kind of less-than-knowledge cognition can’t be to respond to the worry that the cognition might be false, because if I am doing A intentionally, then my thought that I am doing A is true: the cognition condition is a condition on something that, if it obtains, guarantees the truth of the cognition in question.128 Rather, the point of weakening the claim in this kind of way must be to respond to a worry about justification. Consequently, one might reasonably wonder how my account of an agent’s cognition in intention is supposed to amount to knowledge in intention: what justifies the agent’s true practical cognition, and thereby makes it a case of practical knowledge? A number of proposals about the epistemology of agent’s knowledge have been made in the recent literature. Unfortunately, I cannot consider them all here. I want to focus on what strikes me as the most plausible suggestion, in light of the account developed thus far in this essay. The idea, due to Setiya, is this: “One is epistemically justified in forming the belief that one is doing A involved in doing A intentionally, only if, and because, one knows how to do A”.129 Now, the knowledge how that is supposed to justify one’s cognition in intention must be general knowledge how, not particular knowledge how, for the latter is already con127 Setiya (2007), 25. 128 We have also seen how and why if I intend to do A, it is true that I am going to do A. 129 Setiya (2009b), 128. See also Setiya (2008).

Practical Knowledge and the Structure of Action

205

tained in one’s cognition in intention.130 Setiya’s suggestion, understood in this way and shorn of its reference to belief, is a good fit for the story I have told: whereas I focused on the metaphysical role of general knowledge how in constituting the calculative and temporal unity of an intentional action (in progress and in prospect), Setiya points to its epistemological role.131 And given the account of the structure of intentional action provided here, according to which the metaphysics and epistemology of intentional action are really the same thing looked at from different angles, as it were, it should be no surprise if general knowledge how is the metaphysical basis of intentional action and the epistemological basis of knowledge in intention. My proposal, then, is this. An agent’s possession of general knowledge how to do B makes it possible for her to intend (rather than merely aspire) to do B; her exercise of general knowledge how to do B in doing B makes it the case that it will be no accident if she ends up succeeding in having done B; and thus her exercise of general knowledge how to do B is what makes her thought that she is doing B intentionally true. Moreover, her possession of general knowledge how to do B constitutes her entitlement to claim, when she is exercising that knowledge, that she is doing (rather than merely trying to do) B, and her corresponding entitlement to claim, when she intends to exercise it in the future, that she is going to do B (rather than that she is merely going to try to do B).132 The agent’s practical thought, that she is doing B intentionally, amounts to practical knowledge only if she is in fact doing B intentionally, to be sure (no knowledge without truth); but she is doing B intentionally 130 See §3.1 above. If one’s knowledge in intention is knowledge of what one is going to do in the future, it may as yet contain particular knowledge how only potentially. 131 I do not claim that Setiya envisages general knowledge how playing the metaphysical role in constituting the unity of intentional action (in progress and in prospect) that I have suggested. 132 Note that the suggestion is not that the agent infers her way to knowledge that she is doing B from knowledge of her intention to do B and knowledge of how to do B. The general knowledge how has already been taken into account in ascribing to her the intention that she knows that she has; it cannot provide additional information that could license drawing any conclusion about what she is doing that she would not have been entitled to draw directly from her knowledge of her intention. On the account provided here, according to which knowledge of one’s intention simply is knowledge of one’s intentional action in progress or in prospect, there is no gap for an additional premise to bridge, and thus no need to adopt the view that agent’s knowledge is inferential.

206

Will Small

only if what is happening is the exercise of her general knowledge how to do B, the possession of which knowledge gives her the authority to make claims of the form ‘I’m doing B’—claims that thereby express practical knowledge that she is doing B, which knowledge is the cause of what it thereby understands. By contrast, an unjustified practical thought—the thought, say, on the part of agent lacking general knowledge how to do B, that, in doing what she is doing, she is doing B intentionally—could not be such a cause, because such a thought does not (practically) understand what it is about; such a thought could not be true. It follows that one cannot think truly, but without justification, that one is doing something intentionally. This marks a significant difference between practical and theoretical thought, because one can have true, unjustified beliefs. The difference is due to the fact that theoretical knowledge is derived from the object known, which is prior and does not depend for its existence on its being known, by contrast with the object of practical knowledge. I have now completed my outline of what I take to be a promising account of a distinctively practical form of knowledge, an agent’s knowledge of her intentional actions in progress and in prospect. The account is based on the idea that it is a necessary condition on intentional action in progress and in prospect that the agent has practical knowledge of what she is doing, or going to do, where the true ground and meaning of this condition is found in the calculative and temporal structure of intentional action. In the remainder of this essay, I want to raise and respond to a worry about my account. The worry concerns basic actions and the knowledge how of which they are the exercise. It takes the shape of a dilemma with a familiar form: on the one hand, the account of the structure of intentional action and practical knowledge that I have offered seems to presuppose that there must be calculatively basic actions—actions performed without performing other actions as means to them—and non-procedural general knowledge how (skill)—knowledge how to do something that does not consist in knowing that one can do it by doing something else one knows how to do— on pain of a regress that would leave it seeming impossible how anyone could ever do anything intentionally; on the other hand, it is difficult to see how any conceptions of basic action and skill could so much as play the foundational roles required of them. After articulating this dilemma (§6), I will sketch a solution (§7).

Practical Knowledge and the Structure of Action

207

6 Basic Action and Skill: a Dilemma 6.1 Why There Must Be Basic Action As we have already seen, the A–D order is many things at once: it is the order of action explanation and the order of practical reasoning, and it displays the structure of intentional action and the structure of the agent’s knowledge in intention. An intentional action’s calculative structure can be traversed, as we have seen, via the reason-seeking question ‘Why?’ and the means-seeking question ‘How?’. Anscombe thinks that the chain of ‘Why?’s, which articulates and extends the A–D order considered as the order of action explanation, comes to an end. It ends when the question ‘Why are you doing D?’ receives in response a “desirability characterisation” (e. g. ‘Because it’s fun’ or ‘Because it’s pleasant’, ‘Because it’s good for my health’, and ‘Because it befits an X to do D’); issues concerning the “decency as an answer” of such desirability characterisations to the question ‘Why?’ transcend action theory—they “belong[…] properly to ethics”133. It seems that the chain of ‘How?’s must also come to end as well, on pain of regress. Given what we said about the A–D order, we could express the regress in different ways. With an eye towards the way in which the A–D order captures the structure of intentional action, we might think that, while many of the things we do we do by means of doing other things, there must be some things we can just do, without taking any means to them. Reflecting now on how the A–D order illuminates the structure of agent’s knowledge, we might think with Jennifer Hornsby that “some things—at the end of [the] ‘by’-chains, as it were—must be done without knowledge of procedures. These are things the agent does ‘directly’. They are basic things, …which we are inclined to say the agent is able to simply do”134. And recalling that the A–D order is also the order of practical reasoning, we might think that if, as Anscombe puts it, “[t]he mark of practical reasoning is that the thing wanted is at a distance from the immediate action, and the immediate action is calculated as the way of getting or doing or securing the thing wanted”135, then, unless practical reasoning is to go on forever in vain, without the agent being enabled to act, there 133 Anscombe (1963), §40. 134 Hornsby (2005), 114. 135 Anscombe (1963), §41.

208

Will Small

must be immediate actions, the doing of which need not be mediated by calculation resulting at some supposedly ‘more’ immediate action for which, of course, the same problem would arise.136 From the fact that the A–D order is the order of the structure of intentional action, the order of agent’s knowledge, and the order of practical reasoning, it is plausible to conclude that these regresses, of the dependence of intentional action on intentional action, of the agent’s knowledge how to do things, and of practical reasoning, are in fact the same regress. We need an account, then, of something that stands to the question ‘How?’ as the notion of a desirability characterisation stands to the question ‘Why?’. And whatever we make of Anscombe’s move in proposing that the question ‘What is intentional action for?’ belongs not to action theory but to ethics, a parallel move with respect to our question ‘How is intentional action possible?’—to defer it to some other branch of philosophy—seems implausible. The standard suggestion, familiar from the work of Hornsby among others,137 is that the regress is stopped by what we might call calculatively basic action. 138 The calculatively basic things done are things “the agent is able to simply do,” as Hornsby puts it—that is, things that the agent can do without doing them by means of doing something else. And they will be things the agent knows how to do, which knowledge (i. e. skill) is exercised in her simply doing them.139 On this conception of basic action, 136 This does not affect the point, made in §2.3 above, that practical reasoning continues while the agent is acting. 137 Hornsby (1980a; 1980b; 2005; 2007). Enç (2003) and O’Brien (2007) make similar proposals, on which I will occasionally draw as well. 138 What I call “calculatively basic”, Hornsby refers to as “teleologically basic” (Hornsby (1980b), 88). I distinguished between calculative and other forms of teleology in §2.5. The term basic action has philosophical currency beyond the topic, to which the present discussion is restricted, of calculatively basic action. 139 An agent’s knowledge how to do things cannot be exhausted by procedural general knowledge how. Indeed, the proposal made in the previous section for understanding the justification of practical cognition in terms of the agent’s possession of general knowledge how appears to depend on the agent’s procedural general knowledge how bottoming out in non-procedural general knowledge how—skill. This is easy to see: an agent could know that one can do B by doing A, and yet not herself be in a position to form the intention to do B by doing A or be entitled to claim that she is going to do B by doing A. She might have come to know—by reading it in a book, perhaps—that one can do B by doing A. But she could not put such knowledge into practice without know-

Practical Knowledge and the Structure of Action

209

different things may be basic for different agents. This is because different agents have different skills, acquired through learning and practice, as an example from John Searle illustrates: For a good skier, making a left turn can be a basic action. He just intends to do it and he does it. For a beginner to make a left turn, he must put his weight on the downhill ski while edging it into the slope, stem the uphill ski, then shift the weight from left to right ski, etc., all of which are reports of the content of his intention in action.140

Searle’s example brings out another important point, that although bodily movements may be among the basic things done (as shifting one’s weight is, for the novice skier), so may actions “further out” (as making a left turn is, for the good skier). Now, the good skier’s left turn may be composed of a sequence of movements of just the same types as the beginner, but the point is that they do not figure for the former, as they do for the latter, as objects of his practical thought and attention, means taken to his end of making a left turn. As Anscombe puts it, “[i]n general, as Aristotle says, one does not deliberate about an acquired skill; the description of what one is doing, which one completely understands, is at a distance from the details of one’s movements, which one does not consider at all”.141 However, even with these remarks about the scope of basic action in place, the regress-stopping proposal remains highly schematic. For we have no idea yet what it is to be “able to simply do” something, or to do ing, herself, how to do A. And if she could not put such knowledge into practice, possessing it would not entitle her to claim that she is doing, or is going to do B, and could not explain the calculative and temporal unity of the purported action. The idea of procedural knowledge how is the idea of a kind of practical knowledge that accounts for an agent’s ability to intentionally do things by doing other things intentionally, things she already knows how to do. 140 Searle (1983), 100. 141 Anscombe (1963), §30. Cases in which the agent does not conceive of a bodily movement as his means to his end—where the thing “further out” is done “directly”—are not hard to find. A favourite example of Hornsby’s is that of performing speech acts, where, in one’s mother tongue at least, one does not make detailed adjustments to one’s mouth, tongue, etc., as a means towards correct pronunciation; see e. g. Hornsby (1980a), 389; Hornsby (2005), 120 ff. and Hornsby (2007), 173. Similarly with the complex movements of the fingers made in typing or playing the piano—or even tying one’s shoelaces: if the agent made it his end to make those movements of the fingers, he would be best off typing or playing the piano or tying his shoelaces, for those are the things he does directly.

210

Will Small

something “readily”, “directly”, “forthwith”, or “just like that”—to use some other phrases common in the literature.142 Indeed, our problem is precisely that our grip on what it is to do something intentionally has so far centrally involved the notion, captured by the A–D order, of doing it by doing something else. Similarly, we do not yet understand how skill is supposed to make basic action possible, and how it is supposed to be able to play the metaphysical and epistemological roles assigned to general knowledge how in §§3 – 5 above. Worse still, there is some reason to think that a genuinely basic action could not be an intentional action at all, and that skill could not be a genuine form of knowledge. If this is so, then there will be no way to block the regress.

6.2 Why There Cannot Be Basic Action Michael Thompson argues that intentional actions are shot through with calculative complexity of the part-whole variety. The idea of calculative part-whole complexity has obvious application to those intentional actions that are …intuitively resoluble into a heterogeneous collection of sub-actions that are themselves clearly intentional—organs, as it were, of the whole. Such is the relation of egg-breaking and egg-mixing to omelet-making, of brick-laying and door-framing to house-building, and of writing the letters ‘a’ and ‘c’ to writing the word ‘action’.143

But it seems equally obvious that this intuitive resolution “will come to a limit”144 : there are, intuitively, no intentional sub-actions that stand as parts to writing the letter ‘a’. At such a point, then, we seem to find our final resolution of a nonbasic action—writing the word ‘action’— into basic actions, which themselves have no intentional sub-actions as parts.

142 For example: one is “able to simply A” (Hornsby (2005), 115); one is “simply able to [A]” (Hornsby (2007), 168); one can do A “just like that” (Hornsby (1980b), 88); one “has immediate access” to doing A (Enç (2003), 52); one knows how to do A “without needing to know how to do it” (O’Brien (2007), 164); one knows how “to execute [A] just like that” (O’Brien (2007), 164); etc. Such locutions are rife in contemporary action theory, independently of focused discussions of basic action. 143 Thompson (2008), 106 f. 144 Thompson (2008), 107.

Practical Knowledge and the Structure of Action

211

However, Thompson claims, “such resolution is not necessary”145. He argues that for any purportedly basic action B that an agent has performed, there was a more basic action A that she did because she was doing B: “Even actions that, like arm-raising, do not divide in this way [sc. into intuitive ‘organs’] need not, after all, be viewed as pointlike”146. The argument for this claim is given in terms of considerations pertaining to those actions that involve moving things, including moving oneself. The idea is that if, for example, an agent intentionally raised his arm from his side (a) to his ear (y), in order to scratch it, then he raised his arm from his side to, say, his shoulder (p), and intentionally: he raised it to his shoulder because he was raising it to his ear, and he knew while he was doing it that he was doing it and why, as we could have discovered if we had stopped him and asked him. If this is right, then it seems that every segment of the trajectory a– y was an intentional action. Thus, Thompson claims, “[a]cts of moving something somewhere intentionally always have an initial segment that is also an act of moving something somewhere intentionally”147. Even the movement a–b isn’t safe: it must have an “initial segment” a–a’, which itself is an intentional action that has as its own parts intentional actions, and so on. If this argument (call it the initial segment argument) is sound, then all intentional actions that involve moving something somewhere, even if it is only one’s limbs that are thus moved, are irreducibly calculatively complex, where the complexity is of the part–whole variety. One might suspect that the initial segment argument turns on the peculiarities of local motion. But, as Sebastian Rödl points out, the notions of part and whole are applicable in virtue of the in/complete contrast that belongs to the temporal structure of action: ‘S did A’ can be true only if there was a time when S was doing A, but had not yet done A, a time when the action was underway, but not yet complete. The idea of someone’s having performed an action has application only where the idea of his having been performing it has application.148

But what S had done, while S was doing but had not yet done A, is an action by means of which S was doing A, to which the same argument

145 146 147 148

Thompson (2008), 107. Thompson (2008), 107. Thompson (2008), 111. Rödl (2002), 329 f., variables modified.

212

Will Small

applies.149 According to Thompson and Rödl, then, every part, in the relevant sense, of an intentional action is itself an intentional action, and every intentional action has proper parts in the relevant sense; thus there cannot be basic actions.150 The calculative structure of intentional action and its temporal structure apparently issue conflicting demands: there must and yet cannot be basic actions.

6.3 What Does the Initial Segment Argument Show? The upshot of the initial segment argument is not immediately obvious. By virtue of the thought that a calculatively basic action may be “at a distance from the details of one’s movements,” in Anscombe’s phrase, a defender of (this conception of) basic action evades the objection that it is a kind of action-theoretical prime mover. We need not commit ourselves to thinking that a basic action must be atomic—that it has no parts that are themselves actions. Recall Searle’s example of the good skier. The basic thing done is making a left turn. There is a sense in which he does this by putting his weight on the downhill ski while edging it into the slope, etc. But whereas the beginning skier would make those movements in order to make a left turn, the expert skier can simply make a left turn—perhaps in order to make those movements. But are the component parts of the expert skier’s turn intentional actions or not? If each part is an intentional action, then it depends on practical thought. Now, it would be absurd to suppose that every non-basic action is preceded by an occurrent thinking of a procedural fact. So it is not the presence or absence of such a thought process that makes an action non-basic or basic. Furthermore, it is not the case that mere possession of a piece of procedural knowledge of the form ‘One can do B by doing A’ makes it the case that one’s doing B necessarily employs this knowledge, or that one is doing B by doing A (in the calculative sense of ‘by’)—the expert skier might have taught a beginner those facts ten minutes ago. Rather, it is whether the action is an exercise 149 Thus Rödl says that though Thompson’s argument shows “that it is not impossible that every part of an intentional action is, again, an intentional action,” he (Rödl) “claim[s] that it is necessary” (Rödl (2002), 332, n.12). 150 The initial segment argument presupposes that every phase of an action is a part of it (though it does not require, what is obviously false, that every part is a phase). I discuss this argument in greater detail in Small (unpublished).

Practical Knowledge and the Structure of Action

213

of procedural or non-procedural knowledge how. But if procedural knowledge is the medium of calculative thought, as our focus on the calculative structure of intentional action has suggested, and if calculative articulation is essential to intentional action, as Thompson and Rödl think, then it seems that behaviour that issues from non-procedural knowledge cannot be intentional action. What, then, of the possibility that a basic action has parts, which are not themselves intentional actions? The problem is this. The calculative unity of these parts in the action is not accounted for, and thus neither is the temporal unity of the action as a whole; the only possible source of the action’s structure lies beyond the agent’s practical thought, in which case the action is not intentional. We can bring this out by first seeing how an end that lies beyond the agent’s practical thought cannot rationally constrain the means taken towards it. Suppose I have set you to work doing C by doing B. You know all right that you are doing B, and you know that you are doing B by doing A. But suppose you don’t know that you are doing B for the sake of doing C—I have concealed the purpose of your work from you. As far as you’re concerned, you are doing B, not because you are doing C, but because you are doing D—carrying out orders, earning your pay. In this case, though you are doing C (by doing B by doing A), and though there is a purposive nexus between your doing B and your doing C, you are not doing C intentionally; I have kept the purposive nexus from you. The source of the “means-end unity” of doing B and doing C is external to your practical thought, alien to you. For instance, if my plan fails, and you are doing B, but not in such a way as to be getting C done, then this is my problem, not yours: that you are, as far as my calculation is concerned, meant to be doing C in doing B can set no standard for your doing B if this is not how you conceive of the point of doing B. Return now to a purportedly basic action, A, with parts, 1, 2, 3, that are not themselves intentional actions. The agent is not doing A by doing 1, 2, and 3 where the sense of ‘by’ is to introduce means. There is still a sense in which she is doing A by doing 1, 2, and 3; 1, 2, and 3 are, after all, parts of her doing A. But that they are parts, and that they contribute to her doing A in whatever ways they do, are, by hypothesis, beyond the agent’s practical thought, though she may know of them by observation, hearsay, or the memory of her training. From the agent’s practical point of view, doing A appears “pointlike”; the details of her movement are alien to her. She might know that

214

Will Small

she is doing A, but she cannot understand her doing A: the fact that, in making these movements, she is doing A (and that in thereby doing A, she is doing B, which goal imposes constraints on how she does A on this occasion) is not something that can exert rational constraint on her performance of those movements. And so her knowledge that she is doing A cannot be practical knowledge, because practical knowledge is the cause of what it understands, the source of the calculative, and thereby temporal, unity of the action. Thompson argues that, as Aristotle (for example) teaches, skill or craft or techne¯ often drives out deliberation. What is done in accordance with skill in doing B, or in the exercise of a practical capacity to do B, is not, as such, determined by deliberation or reflection—unless by a peculiarity of the skill itself (which might involve measurement and calculation, say, as laying carpeting does). But the absence of reflection does not make the action thus skillfully performed, making a pot of coffee, as it might be, or raising a hand, into a sort of unanalyzable whole; egg-breaking certainly does not lose its character as an intentional action after the agent’s thirty-fourth omelet. Why should we suppose that acquisition of the type of skill that interests us, skill in moving a limb or object along this or that type of path, must deprive movement along sub-paths of their status as intentional? 151

A skilled omelette-making is no less calculatively articulated for being skilled as one done in accordance with the steps of a recipe—one is still seasoning the eggs in order to make it taste nice. But the idea of knowing how to make an omelette that figures in the thought that making an omelette can be a basic action for somebody seems to treat omelette-making as an “unanalyzable whole”. And once it is so treated, it begins to be unrecognizable as an intentional action. I want to draw attention to a common way of approaching the topics of skill and basic action, one that I think our reflections on the structure of intentional action and practical knowledge enable us to see as quite misguided. Setiya suggests that, whereas my knowledge how to perform a non-basic action “consists in knowing basic means and knowing how to take those means,” my knowledge how to perform a basic action A “consists in the disposition to do A in execution of my intention”152. Now, according to our proposal from §5.4, the agent’s skill of A-ing must do metaphysical and epistemological duty: it must account for the unity of the intentional action that is the exercise of the skill, and 151 Thompson (2008), 108. 152 Setiya (2009b), 135, variables modified.

Practical Knowledge and the Structure of Action

215

it must justify the agent in claiming that she is doing, or is going to do A. How is a disposition to do A in execution of an intention to do A supposed to discharge these duties? Let’s focus first on skill’s epistemological role. How does being disposed to do A in execution of one’s intention to do A justify an agent’s practical thought, such that it amounts to practical knowledge of intentional action is progress or in prospect? We can say at least this. If I know how to take those basic means, this knowledge consists in the disposition to execute the corresponding intentions. Since I have this disposition, it is no accident that, when I intend and thus believe that I am [taking the basic means], I am doing so in fact. Exercising basic knowledge how ensures non-accidentally true belief. To say this is not to endorse an epistemology on which its being no accident that a belief is true suffices for it to count as knowledge. But it does preempt a residual source of skepticism, that the truth of beliefs formed without sufficient prior evidence could only be a matter of luck. When they are constituted by intentions and one knows how to perform the relevant actions, that is not the case.153

As Setiya appears to acknowledge, there must be more to say here. On the one hand, it is hard to see why, absent a reliabilist framework, the metaphysical account of why the agent’s thought that she is doing A is non-accidentally true thereby amounts to an epistemological account of her knowledge that she is doing A. 154 But what justificatory role could the disposition itself play? Perhaps the agent’s disposition to do A when she intends to do A is also a disposition to think that she is doing A when she is thereby doing it. But how can being disposed to think something constitute her justification for thinking it, even if what she thinks is (non-accidentally) true? More generally, why should being disposed to do A when one intends to do it count as have knowledge how to do A? Indeed, the worry that a disposition to act is the wrong kind of thing to count as a form of knowledge is a thought that has received expression in the recent attacks on Ryle’s account of knowledge how, according to which knowledge how consists not (or not solely) in knowledge of propositions, but rather in certain complex, “multi153 Setiya (2009b), 136. 154 Setiya, rightly, does not think that she is justified in thinking that she is doing A intentionally on the basis of an inference from the premises that she intends to be doing A and that she is disposed to do A when she so intends: agent’s knowledge is for him, as it is for me, non-inferential.

216

Will Small

track” dispositions to act.155 Setiya suggests that the view that knowledge how to perform a basic action “does epistemic work helps explain why it should count as knowledge”156, but it remains quite unclear how it is supposed to be able to perform any epistemic work unless it already counts as knowledge.157 On the other hand, the agent’s disposition to do A intentionally when she intends to do A secures the non-accidental truth of her practical cognition that she is doing (or is going to do) A only if it explains the temporal unity of her action. She is doing A (and thus her practical cognition ‘I’m doing A’ is true) only if what is happening is such that it will be no accident if she ends up having done A intentionally. But what will explain that? According to our account in §4.5 above, the temporal unity of an intentional action is constituted by its calculative unity. Yet a basic action is supposed not to be calculatively articulated: an action (B) is calculatively articulated when it is being done in order to do something else (C), and done by means of doing something else (A). Berent Enç conjectures that [o]nce the intention to tie one’s shoe lace is formed, no subsequent practical reasoning is required to figure out how to satisfy the intention. If the content of the intention matches an item in one’s repertoire of basic acts, and if the conditions are right, the formation of the intention becomes sufficient for its execution by the lower subsystems.158

And he summarizes his view in the claim that “knowing how to do something, B, without needing to use one’s knowledge of how to do something, A, in order to do B boils down to a capacity rational agents have of getting B done without needing to cognitively control how it is done”159. This is a good example of the conception of basic action and 155 For Ryle’s account, see Ryle (1949), chs. 2 and 5; for the attacks, see, e. g., Stanley/Williamson (2001); Snowdon (2003). 156 Setiya (2009b), 134 f. 157 Setiya’s claim (2008), 406 ff., that his account shows why knowledge how cannot be propositional doesn’t quite work: he claims that if it were, it would justify knowledge in intention inferentially, but knowledge in intention is non-inferential. But procedural knowledge how is, or at least can be, propositional; knowledge in intention of non-basic action is not by that fact inferential knowledge. What this shows is that one piece of propositional knowledge may noninferentially justify another. Therefore the argument that non-procedural knowledge how—skill—is non-propositional must be made on other grounds. 158 Enç (2003), 71. 159 Enç (2003), 71.

Practical Knowledge and the Structure of Action

217

practical knowledge that I want to reject.160 Implicit is the idea that cognitive control could only be involved via the exercise of procedural knowledge. The control over the parts of a basic action and their unity is assigned not to practical thought, but to lower, sub-personal systems. But this precisely undermines the status of the action as intentional: the explanation of why it will be no accident if an agent in the midst of performing a basic action A will end up having done A is to be located beyond her practical thought. This puts her doing A on a par with her digesting a piece of meat, the temporal unity of which event is similarly explained by the characteristic function of certain sub-personal systems. One cannot save this kind of view by suggesting that “the lower behavioural systems only ‘know how’ to make good on certain commands”161 or that one’s fingers “just get on with the job by themselves, as it were, now that I have learned to type”162—the scare quotes and ‘as it were’s give the game away. The language of delegation (and the like), which is so pervasive in contemporary literature on basic action, has its home in describing the interactions of multiple rational agents, not one rational agent and various sub-agential, non-rational systems.163 It seems we are left without a response to the initial segment argument and its incredible conclusion. Taking skill to consist in a disposi160 The agent’s cognitive relationship to the action, on this picture, leaves it such that any complexity in the action is external to her practical thought. It is as if the command to make an omelette is issued by the mind, whose work here is now done, and automatically “carried out” by the body, or sub-personal systems within it. It is not clear whether or not Setiya would flesh out his appeal to dispositions along the lines suggested by Enç (though see his remarks about the “‘sub-intentional’ components” of intentional actions at Setiya (2007), 55.). 161 Enç (2003), 71. 162 Steward (2009), 301. 163 Hornsby is sensitive to these difficulties. When she tries to explain what it is to be able to simply do A, to do A directly—without mediation by calculation, practical reasoning, procedural knowledge—she warns us against a variety of misunderstandings of the notion of simplicity or directness in play. The exercise of skill is not “something that we simply do” in the sense in which digesting is something that we simply do—something that lies beyond our control, cf. Hornsby (2005), 126. Nor are skills themselves simple, or simple as such; many of them are very complex. And the actions and activities in which the exercise of skills consists need not be simple, as reflection on the good skier and the initial segment argument have already shown; cf. Hornsby (2007), 173. But these clarifications amount neither to a positive account of what it is to be able to simply do A, nor to an account of skill as a genuine form of practical knowledge.

218

Will Small

tion (or set of dispositions) to execute instructions issued by practical thought, while at the same time supposing that the disposition does not itself fall within the scope of the agent’s practical rationality, encourages the thought that basic action ends up outside the sphere of practical thought altogether, by making the parts, and thus the whole, of a basic action non-intentional (and thus the non-basic action performed by performing the basic action non-intentional, and so on up the chain). It is untenable to think that basic actions have no parts, or that their parts are not intentional. But our only grip on what it would be for the parts to be intentional is for them to owe their articulation to calculation. Yet an action that is done by doing its parts, which are calculated to amount to doing the whole, is precisely not a basic action. What we need to understand, then, is this: how can there be actions that are calculatively articulated— remember, this means that the agent understands the articulation, and his understanding is its source—which calculative articulation is not owed to calculation?

7 The Structure of Practical Knowledge 7.1 A Parallel Problem It helps, I think, to consider what I take to be a parallel problem in thinking about perceptual experience. I’ll quickly rehearse a line of thought familiar from the work of John McDowell. If the only way in which a belief could be justified were by another belief, then we would not be able to think of the world bearing rationally on our thought; for, only thought could rationally bear on thought. Thus there is a great temptation to think that it must be possible for a belief to be justified by something other than a belief; and the natural candidate—for it is the rational bearing of the world itself that we are interested in—is experience. But it is surprisingly difficult to give an account of experience on which it is fit for this purpose, and as long as this is so the very ideas of beliefs with empirical content, and of believing one thing on the basis of another, remain in jeopardy. It is easy to lapse into a version of the Myth of the Given: the thought that experience provides external constraint on thought from outside the sphere of rational relations and yet somehow this constraint is supposed to be rational. The difficulty consists in conceiving of experience as somehow simultaneously within thought—so that its bearing can be rational—and

Practical Knowledge and the Structure of Action

219

yet beyond thought—so that it can provide external constraint. As is well known, McDowell’s elegant solution to this puzzle is to conceive of experience as, like thought, having conceptual content; however, while conceptual capacities are actively employed in thought, they are passively drawn into operation in experience. That conceptual capacities are operative in experience itself, and not merely in the response in thought to experience, secures the rationality of the transaction between mind and world; that their operation is passive secures the external constraint on thought. McDowell offers a diagnosis of why such a view seems difficult: we have difficulty seeing how a natural phenomenon like sensory perception could be “anything but externally related” to conceptual activity.164 What we need, he thinks, is to relax our conception of the natural, to allow that a transaction in nature—the receipt of sense-impressions— could at the same time be a transaction in the space of reasons. And he notes, perhaps surprisingly, that his focus on problems concerning perception “was not essential”165 : Kant says “Thoughts without content are empty, intuitions without concepts are blind”. Similarly, intentions without overt activity are idle, and movements of limbs without concepts are mere happenings, not expressions of agency. I have urged that we can accommodate the point of Kant’s remark if we accept this claim: experiences are actualizations of our sentient nature in which conceptual capacities are inextricably implicated. The parallel is this: intentional bodily actions are actualizations of our active nature in which conceptual capacities are inextricably implicated.166

What could it be for conceptual capacities to be inextricably implicated in action—in particular, in basic action, whose basicness, McDowell agrees, means that one does not do “the thing in question by doing something else in a sense that brings one’s means-end rationality into operation”167 ? His response, in a recent paper, is less illuminating than one might hope: So what difference does it make, according to me, for activity to be permeated with rationality…? … When a rational agent catches a frisbee, she is realizing a concept of a thing to do. In the case of a skilled agent [i.e. one who is performing a basic action], she does not do that by realizing 164 165 166 167

McDowell McDowell McDowell McDowell

(1996), (1996), (1996), (2007),

89. 89. 90. 327.

220

Will Small

other concepts of things to do. She does not realize concepts of contributory things to do, in play for her as concepts of what she is to do by virtue of her means-end rationality in a context in which her overarching project is to catch the frisbee. But she does realize a concept of, say, catching this. … When a dog catches a frisbee, he is not realizing any practical concept; in the relevant sense, he has none. The point of saying that the rational agent, unlike the dog, is realizing a concept in doing what she does is that her doing, under a specification that captures the content of the practical concept that she is realizing, comes within the scope of her practical rationality—even if only in that, if asked why she caught the frisbee, she would answer “No particular reason; I just felt like it”.168

The idea seems to be that basic action is “permeated with rationality”— it “inextricably implicates conceptual capacities”—insofar as it comes within the scope of the agent’s practical rationality by being subject to the reason-seeking question ‘Why?’. But though the basic action is apt to be part of a larger whole whose articulation and unity is the deed of the agent’s means-end rationality, whatever structure the basic action has is not aptly characterized in terms of the operation of that means-end rationality. It is hard to see this picture as more than a recapitulation of our problematic. We knew from the beginning that a basic action must be capable of being subject to the question ‘Why?’, for it must be capable of answering the question ‘How?’—indeed, capable of answering it in a special, regress-stopping way. Our problem has been that we cannot see how a basic action, insofar as any structure it has is not the deed of the agent’s means-end rationality, could genuinely be an answer to the question ‘How?’, genuinely subject to the question ‘Why?’. An agent without a practical understanding of the articulation of his movement cannot be expressing practical knowledge when he says ‘I am doing A’. The shape that McDowell’s picture takes is this. A basic action, if it is to stand a chance of doing the work that it needs to do, must be subject to the question ‘Why?’; it must fall within the scope of the agent’s means-end rationality. Yet that means-end rationality cannot enter into the constitution of the basic action itself, or else it would no longer be basic. It can merely relate a basic action to a larger action or plan. However, this starts to seem a bit as if a basic action is mythical in the way in which the Given was: it is supposed to stand in rational relations while being intrinsically nonrational. And thus to alleviate this suspicion, McDowell insists that even basic action is permeated with rationality, 168 McDowell (2007), 327.

Practical Knowledge and the Structure of Action

221

that it inextricably implicates concepts. But this makes it clear that the inextricable implication of concepts cannot consist in the fact that a basic action comes within the scope of the agent’s practical rationality by being subject to question ‘Why?’, for its being so subject presupposes the inextricable implication of concepts. (This is another way of putting what is wrong with the ‘delegation-to-subpersonal-systems’ approach to basic action.) The trouble is that we only have one picture of what the implication of concepts in the practical domain looks like: their deployment in calculative thought. The problem about perception was resolved by distinguishing two different ways in which the same (theoretical) conceptual capacities could be operative. In Mind and World, the idea was that we could distinguish between the active putting together of concepts into propositional form in judgment and the passively given, propositionally articulated, content of experience. In more recent work, McDowell’s thought is rather that the passively given content of experience is not propositional, but intuitional; this intuitional unity of the sensory manifold in experience is still conceptual, even though it is not discursive, because …every aspect of the content of an intuition is present in a form in which it is already suitable to be the content associated with a discursive capacity, if it is not—at least not yet—actually so associated. That is part of the force of saying, with Kant, that what gives unity to intuitions is the same function that gives unity to judgments.169

The parallel between action and perception suggests, then, that we need to be able to distinguish between two ways in which practical conceptual capacities could be operative in action. And the idea of a second way in which practical conceptual capacities could be operative in action might be able to do duty as the substantive account, for which we have been looking, of the purportedly-regress-stopping proposal that basic actions are the exercise of skill.

169 McDowell (2008), 264.

222

Will Small

7.2 Skill as a Rational Capacity What we need is for basic actions to have a rational structure of the same form as actions structured by the A–D order, but where this structure is not the result of active mediation in practical thought, that is, of calculation from general ends to particular means. The suggestion is this: the capacity for calculation, whose exercise in practical reasoning accounts for the articulation and unity of the A–D order, also accounts for a corresponding articulation and unity in basic action.170 The same function—the capacity for practical thought—gives unity to both. Now, the idea that conceptual capacities are passively operative in basic action seems to be an obvious non-starter. The contrast we need to exploit is not that between activity and passivity; rather, it is that between mediated and immediate activity. The idea we need is one of the immediate realization of rational structure: the rational structure of a basic action is due not to its being worked out, but to its being immediately instantiated. In calculation, the capacity for practical thought is actualized in the synthesis of different representations of action: in deriving my intentional action (doing B by doing A, here and now), from an intention to do B and procedural knowledge that one can do B by doing A, I actively bring together a representation of doing B as end and a representation of doing A as means, and my doing A in order to do B is nothing other than this synthetic representation. This is what it is for calculation to take place, for means and end to come together in practical thought. The calculative articulation of a non-basic intentional action by practical reasoning is the result of actively synthesizing practical representations. An exercise of skill, too, must be calculatively articulated, on pain of not being an intentional action, not subject to Anscombe’s question ‘Why?’, and thus not fit to be the means by which the agent does something non-basic. But though this articulation must ultimately have the same source (the same function, the same capacity for practical thought), it must be actualized in a different way. It cannot arise through active synthesis, through being derived by practical reasoning; it must lie already articulated in the skill itself, waiting to be immediately instantiated in the skilled action. A skill, then, must be a conceptually-, and thus rationally-, articulated capacity, whose exercises are conceptually—that is, 170 Here I echo McDowell: “The capacity whose exercise in judging accounts for the unity of the content of judgments—propositional unity—also accounts for a corresponding unity in the content of intuitions” (McDowell (2008), 260).

Practical Knowledge and the Structure of Action

223

calculatively—articulated intentional actions, the parts of which are themselves intentional actions, unified by practical thought in its nonprocedural form. In seeing that skill cannot be a mere capacity or disposition, but must be a rational and conceptual capacity or disposition, we are able to assuage the worry about the foundational role accorded to general knowledge how, and thus ultimately to skill, in the proposal for the epistemology of knowledge in intention (§5.4). A full treatment of the epistemology of skill would vindicate the claim that it constitutes a genuine form of knowledge, and explain why this knowledge consists in a rational capacity to act, rather than a piece of propositional knowledge that may or may not be associated with a non-rational capacity to act. It must suffice here, however, to have shown that the metaphysical role that basic action plays in the structure of intentional action is not fulfilled if basic action is but the exercise of a kind of general knowledge how that merely secures the point that it will be no accident if an agent’s cognition in intention is true; for basic action to play that role, the agent’s cognition of it must amount to knowledge, and thus the capacity whose exercise accounts for the cognition’s truth must be a rational, conceptual capacity. Thus we preempt the worry that the epistemology of intentional action must bottom out in reliabilism.171

Conclusion I have argued that there is a cognition condition on intention and intentional action. If an agent is doing B intentionally, he has knowledge in intention that he is doing B, that he is doing B because he is doing C and, if it is a non-basic action, that he is doing B by doing A; his knowledge in intention that he is doing B is knowledge that he is going to do B, knowledge of success, in this sense: he knows that it will be no accident if he ends up having done B. A parallel condition holds, perhaps surprisingly, for an intention for the future to do B and the agent’s knowledge in it that he is going to do B. In both cases, the agent’s knowledge is of what is happening, or what is going to happen, in the world; it is not knowledge merely of his state of mind. This knowledge is practical: it is the cause of what it understands; without it what is 171 I expand on this proposal in a more detailed discussion of skill and basic action in Small (unpublished).

224

Will Small

happening, or what is going to happen, is not an intentional action, and any practical thought of the agent’s about what is happening, or what is going to happen, does not hit the heights of intention. The cognition condition is grounded not in reflection about the directions of fit and guidance of different kinds of mental states, nor in the indicative content of linguistic expressions of intention, but in the calculative and temporal structure of intentional action itself. Examples that purport to show that the content or scope of the condition must be weakened, or that the condition should be completely rejected, have been revealed to be ineffective, because they trade on confusions about the different kinds of failure to which intentional action may be susceptible, confusions remedied via reflection on the structure of intentional action. Cognition in intention amounts to knowledge in intention because intention and intentional action presuppose general knowledge how to do things, which comes in two kinds, procedural knowledge how and skill. The calculative structure of intentional action—of doing B by doing A— and the structure of procedural knowledge—knowledge that one can do B by doing A—presuppose the possibility of basic action, and that basic action is the exercise of skill. Though reflection on the temporal structure of intentional action initially suggested that this demand could not be satisfied, we saw our way to resolving the puzzle by recognizing that skill cannot be a non-rational capacity to act, but must rather be a rational capacity that is characterized by the same structure as practical thought more generally, albeit in a specifically different way.172

References Anscombe (1963): Gertrude Elizabeth Margaret Anscombe, Intention, (Reprinted 2000), Oxford. Austin (1956): John L. Austin, “Ifs and Cans”, in: Proceedings of the British Academy 42, 109 – 132, (cited as reprinted in J. L. Austin (1961): Philosophical Papers, Oxford, 153 – 180). 172 This essay draws on material from my doctoral dissertation. Thanks to Jason Bridges, Jim Conant, Anton Ford, Felix Koch, Eliot Michaelson, Sebastian Rödl, Justin Shaddock, and Candace Vogler for helpful comments and discussion. Earlier versions of parts of this essay were presented at the New School for Social Research Wittgenstein Workshop and at the Practical Philosophy and Wittgenstein Workshops at the University of Chicago; thanks to the participants.

Practical Knowledge and the Structure of Action

225

Baier (1970): Annette Baier, “Act and Intent”, in: Journal of Philosophy 67, 648 – 658. Baier (1971): Annette Baier, “The Search for Basic Actions”, in: American Philosophical Quarterly 8, 161 – 170. Boyle/Lavin (2010): Matthew Boyle/Douglas Lavin, “Goodness and Desire”, in: S. Tenenbaum (ed.), Desire, Practical Reason, and the Good, Oxford, 161 – 201. Bratman (1987): Michael Bratman, Intention, Plans, and Practical Reason, Cambridge/MA. Comrie (1976): Bernard Comrie, Aspect: An Introduction to the Study of Verbal Aspect and Related Problems, Cambridge/MA. Davidson (1967): Donald Davidson, “The Logical Form of Action Sentences”, in: N. Rescher (ed.), The Logic of Decision and Action, Pittsburgh (cited as reprinted in: Davidson (1980), 105 – 148.). Davidson (1971): Donald Davidson, “Agency”, in: R. Binkley/R. Bronaugh/ A. Marras (eds.), Agent, Action, and Reason, Toronto (cited as reprinted in: Davidson (1980), 43 – 61). Davidson (1973): Donald Davidson, “Freedom to Act”, in: T. Honderich (ed.), Essays on Freedom of Action, London (cited as reprinted in: Davidson (1980), 63 – 81). Davidson (1978): Donald Davidson, “Intending”, in: Y. Yovel (eds.), Philosophy of History and Action, Jerusalem (cited as reprinted in: Davidson (1980), 83 – 102). Davidson (1980): Donald Davidson, Essays on Actions and Events. Oxford. Davidson (1985): Donald Davidson, “Adverbs of Action”, in: B. Vermazen/M. B. Hintikka (eds.), Essays on Davidson: Actions and Events (cited as reprinted in the 2001 edition of Davidson (1980), 293 – 304). Donnellan (1963): Keith Donnellan, “Knowing What I am Doing”, in: Journal of Philosophy 60, 401 – 409. Enç (2003): Berent Enç, How We Act: Causes, Reasons, and Intentions, Oxford. Engstrom (2009): Stephen Engstrom, The Form of Practical Knowledge: A Study of the Categorical Imperative, Cambridge/MA. Falvey (2000): Kevin Falvey, “Knowledge in Intention”, in: Philosophical Studies 99, 21 – 44. Finkelstein (2003): David H. Finkelstein, Expression and the Inner, Cambridge/ MA. Galton (1984): Antony Galton, The Logic of Aspect: An Axiomatic Approach, Oxford. Gibbons (2010): John Gibbons, “Seeing What You’re Doing”, in: T. S. Gendler/J. Hawthorne (eds.), Oxford Studies in Epistemology, vol. 3, Oxford, 63 – 85. Goldman (1970): Alvin I. Goldman, A Theory of Human Action, Englewood Cliffs/NJ. Grice (1971): Paul H. Grice, “Intention and Uncertainty”, in: Proceedings of the British Academy 57, 263 – 79. Hampshire (1965): Stuart Hampshire, Freedom of the Individual, New York.

226

Will Small

Hornsby (1980a): Jennifer Hornsby, “Action and Ability”, in: R. Haller/W. Grassl (eds.): Language, Logic, and Philosophy, Vienna, 387 – 391. Hornsby (1980b): Jennifer Hornsby, Actions, London. Hornsby (2005): Jennifer Hornsby, “Semantic Knowledge and Practical Knowledge”, in: Proceedings of the Aristotelian Society, supplementary vol. 79, 107 – 130. Hornsby (2007): Jennifer Hornsby, “Knowledge and Abilities in Action”, in: C. Kanzian/E. Runggaldier (eds.): Cultures: Conflict-Analysis-Dialogue, Frankfurt/Main, 165 – 180. Hume (1977): David Hume, An Enquiry Concerning Human Understanding, Indianapolis. Jones (1960): O. R. Jones, “Things Known without Observation”, in: Proceedings of the Aristotelian Society 61, 129 – 150. McDowell (1996): John McDowell, Mind and World, Cambridge/MA. McDowell (2007): John McDowell, “Response to Dreyfus”, in: Inquiry 50, 366 – 370, (cited as reprinted in McDowell (2009a), 324 – 328). McDowell (2008): John McDowell, “Avoiding the Myth of the Given”, in: J. Lindgaard (ed.), John McDowell: Experience, Norm, and Nature, Oxford, (cited as reprinted in McDowell (2009b), 256 – 272). McDowell (2009a): John McDowell, The Engaged Intellect: Philosophical Essays, Cambridge/MA. McDowell (2009b): John McDowell, Having the World in View: Essays on Kant, Hegel, and Sellars, Cambridge/MA. McGinn (1996): Colin McGinn, The Character of Mind: An Introduction to the Philosophy of Mind. Oxford. Moran (2004): Richard Moran, “Anscombe on ‘Practical Knowledge’”, in: J. Hyman/H. Steward (eds.), Agency and Action, Cambridge, 43 – 68. Moran/Stone (2009): Richard Moran/Martin J. Stone, “Anscombe on the Expression of Intention: An Exegesis”, in: C. Sandis (ed.), New Essays on the Explanation of Action, Houndmills, 132 – 168. Müller (1979): Anselm Winfried Müller, “How Theoretical is Practical Reason?”, in: C. Diamond/J. Teichman (eds.), Intention and Intentionality: Essays in Honour of G. E. M. Anscombe, Ithaca/NY, 91 – 108. O’Brien (2007): Lucy O’Brien, Self-Knowing Agents, Oxford. Paul (2009): Sarah K. Paul, “Intention, Belief, and Wishful Thinking: Setiya on ‘Practical Knowledge’”, in: Ethics 119 (3), 546 – 557. Rödl (2002): Sebastian Rödl, “Practice and the Unity of Action”, in: G. Meggle (ed.), Social Facts and Collective Intentionality, Frankfurt/Main. Rödl (2005): Sebastian Rödl, Kategorien des Zeitlichen: eine Untersuchung der Formen des endlichen Verstandes, Frankfurt/Main. Rödl (2007): Sebastian Rödl, Self-Consciousness. Cambridge/MA. Rödl (2011): Sebastian Rödl, “Two Forms of Practical Knowledge and Their Unity”, in: A. Ford/J. Hornsby/F. Stoutland (eds.), Essays on Anscombe’s Intention, Cambridge/MA, 211 – 241. Ryle (1949): Gilbert Ryle, The Concept of Mind, London. Searle (1983): John R. Searle, Intentionality, Cambridge. Setiya (2007): Kieran Setiya, Reasons Without Rationalism, Princeton.

Practical Knowledge and the Structure of Action

227

Setiya (2008): Kieran Setiya, “Practical Knowledge”, Ethics 118 (3), 388 – 409. Setiya (2009a): Kieran Setiya, “Intention”, in: E. N. Zalta (ed.), The Stanford Encyclopedia of Philosophy, http://plato.stanford.edu/archives/fall2009/entries/intention/. Setiya (2009b): Kieran Setiya, “Practical Knowledge Revisited”, Ethics 120 (1), 128 – 137. Setiya (2010): Kieran Setiya, “Sympathy for the Devil”, in: S. Tenenbaum (ed.), Desire, Practical Reason, and the Good, Oxford, 82 – 110. Setiya (2011): Kieran Setiya, “Knowledge of Intention”, in: A. Ford/J. Hornsby/F. Stoutland (eds.), Essays on Anscombe’s Intention, Cambridge/MA, 178 – 197. Small (unpublished): Will Small, “Basic Action and Practical Knowledge”, manuscript. Snowdon (2003): Paul Snowdon, “Knowing How and Knowing That: A Distinction Reconsidered”, Proceedings of the Aristotelian Society 104, 1 – 29. Stanley/Williamson (2001): Jason Stanley/Timothy Williamson, “Knowing How”, in: Journal of Philosophy 98 (8): 411 – 444. Steward (2009): Helen Steward, “Sub-intentional Actions and the Over-mentalization of Agency”, in: C. Sandis (ed.): New Essays on the Explanation of Action, Houndmills, 295 – 312. Thompson (2008): Michael Thompson, Life and Action: Elementary Structures of Practice and Practical Thought, Cambridge/MA. Thompson (2011): Michael Thompson, “Anscombe’s Intention and Practical Knowledge”, in: A. Ford/J. Hornsby/F. Stoutland (eds.): Essays on Anscombe’s Intention, Cambridge/MA, 198 – 210. Velleman (1989): J. David Velleman, Practical Reflection, Princeton. Velleman (2000): J. David Velleman, The Possibility of Practical Reason. Oxford. Velleman (2007): J. David Velleman, Practical Reflection, second edition, Stanford/CA. Vogler (2002): Candace Vogler, Reasonably Vicious, Cambridge/MA.

Three Forms of the First Person Plural Matthias Haase 1. Introduction Assertion is a self-conscious act. My utterance is not an assertion unless I know that I’m thereby claiming, for instance, that a is F. In knowing this, my knowledge has to reach in a certain way beyond this act. For I have to know what the elements of my utterance mean if I am to know that I’m making an assertion in deploying them; and it belongs to their having meaning that they can be used on other occasions as well, either by me or by other speakers. In knowing that I’m making an assertion I have to know the language to which the elements of my assertion belong. An account of the self-conscious character of assertion must therefore include an account of the epistemology of language. Contemporary philosophy of language tends to approach the epistemology of language from the perspective of an interpreter developing a theory of meaning for the language of another. But if knowledge of language is to become intelligible as a constitutive part of self-consciousness in assertion, then the epistemology of language has to be approached from the perspective of the speaker and be conceived as a form of self-consciousness. As the language I’m speaking is also spoken by others, this self-consciousness is paradigmatically expressed in a statement in the first person plural. This paper investigates the logical form of this deployment of ‘we’. It proceeds through a discussion of the account of semantic self-consciousness put forward in the work of Stanley Cavell.1 The linguist in the jungle doing field research is one of the favorite characters of the philosopher of language. This time, let us suppose that 1

The paper was originally written for a conference on Stanley Cavell in Frankfurt a.M. An earlier version of this paper appeared in the Deutsche Zeitschrift fr Philosophie: Haase (2007). I’m very grateful to Matthew Boyle for tremendous help with my English and Jim Conant, Andrea Kern, John McDowell, Daniel Smyth, Michael Thompson, Sebastian Rödl and Jennifer Whiting for invaluable comments.

230

Matthias Haase

the linguist wants to report on his research to the friendly natives themselves. Provided his translation manual has advanced to an extent that allows such high-level translations back into the indigenous language, he will say something like ‘In English the expression E is used thusand-so’ or ‘English speakers use E thus-and-so’. An English speaker could utter the same sentence in describing how E is used. As a native speaker, however, he can also say ‘We use E thus-and-so’. Let us call the latter statement S, and the statement of the linguist T. Although S and T describe the same state of affairs and have the same truth-value, they are not arbitrarily interchangeable. S cannot be replaced by T in intensional contexts. The linguist cannot assert S, since he is not an English speaker. (The fact that he has translated his research report into English with the help of a translation-manual does not yet make him into a competent speaker of English.) The expressions ‘we’ and ‘English speakers’ have – to use Frege’s terminology – the same referent but different senses.2 In “Must We Mean What We Say?”, Stanley Cavell points out that the difference between S and T reflects a difference between the positions from which the native speaker and the linguist speak. While the field linguist confronts the English language as an object of empirical investigation, the native speaker speaks, as Cavell puts it, “from within the language.”3 Cavell holds that S-type statements are statements of a very specific kind, and he asserts that the recognition of their grammatical specificities is of central importance for the philosophical understanding of thought and language. We will investigate whether that is so, and what analysis of the relevant propositions Cavell himself recommends. In the process, it will emerge that his thesis that the nature of language exhibits itself in the grammatical features of S-type statements is a deep insight, but that his own interpretation of the form of these statements is faulty. Cavell thinks that we need to distinguish between ‘knowing’ and ‘being able to say’ and that S-type statements express, not knowledge, but rather something that underlies our knowledge: “our mutual attunement,” “our agreement with one another.”4 In what follows I will argue that this is a mistake. After investigating Cavell’s understanding of S-type statements (§§2 – 3), I will suggest that the picture of language that follows from this understanding amounts, in the end, to a 2 3 4

See Frege (1994). Cavell (1976a), 16. Cavell (1979), 168.

Three Forms of the First Person Plural

231

form of meaning-skepticism (§§4 – 6). Finally I will claim that this skepticism rests on a misconception of the logical form of the statements with which we say “what we say” (§7). In his investigation of S-type statements Cavell focuses, and I think rightly so, on their first person character; I will argue however that he fails to appreciate the fact that the apparent uniformity of the linguistic category of the first person plural covers over crucial distinctions between the different logical roles that ‘we’ can play in a sentence: it can figure as what I will call a ‘Distributive We’ with which I refer to a set of which I’m an element, as a ‘Communal We’ with which I refer to a community of which I am a member and, finally, as a ‘Generic We’ with which I articulate the form of which I am a bearer. As we shall see, Cavell’s failure to distinguish these logical roles leads him to conflate at crucial points in his considerations the Communal We and the Generic We and, in consequence, two different senses in which an activity can be called ‘social’.

2. The Representation of Language According to Cavell, the difference between judgments of type S and judgments of type T emerges clearly as soon as we pose the question: What justifies the native speaker and the linguist in their respective claims? While the linguist is justified in asserting T only if he has gathered sufficient “evidence of a certain kind”, the native speaker’s assertion is not based on evidence of any kind whatsoever. Cavell’s argument for this thesis takes the form of a reductio. Were the native speaker’s assertion evidence-based, we would have to assume that he had undertaken an unsystematic version of linguistic fieldwork, and that his assertion rests on correspondingly shaky grounds. Since this is absurd, it follows that “there is no evidence (which it makes sense, in general, to say) he has: the question of evidence is irrelevant.”5 This reductio is supposed to show that the relation between a speaker and his language cannot be characterized by appeal to a certain concept of knowledge: namely, the concept of the sort of knowledge that rests on evidence. Knowledge that rests on evidence is observational knowledge. Thus, in asserting that “the question of evidence is irrelevant”, Cavell is claiming that it would be a mistake to believe that the judgment of the native speaker rests on perception – no matter what sort of perception it 5

Cavell (1976a), 13.

232

Matthias Haase

might be. It does indeed seem mistaken to think that the native speaker must observe his own linguistic behavior, and that of other speakers, to be able to assert S. 6 This rules out sensory perception as a source of his cognizance of the language.7 Since Cavell regards it as equally an error “to assume that the ordinary use of a word is a function of the inner state of the speaker,”8 S also cannot be grounded on that peculiar kind of inner perception that is commonly called “introspection.” Let us assume that this much is correct. Now, the claim that the judgment of the native speaker is neither a perceptual judgment nor something grounded on perceptual judgments obviously does not tell us what kind of judgment we are dealing with. In the passage cited above, Cavell initially restricts himself to a characterization of what S-type statements are not, but he does not leave it completely open what should be done to reach a positive characterization: The clue to understanding the sort of statement S is lies in appreciating the fact that ‘we’, while plural, is first personal. First person singular forms have recently come in for a great deal of attention, and they have been shown to have very significant logical-epistemological properties. The plural form has similar, and equally significant, properties; but it has been, so far as I know, neglected.9

According to this remark, statements of type S articulate a special form of self-consciousness – a self-consciousness that, in some yet-to-be-determined sense, extends beyond the individual ‘self’ to embrace a unity that, also in some yet-to-be-determined sense, comprehends a plurality of subjects. This special form of self-consciousness will become intelligible, the remark promises, if we investigate the “logical-epistemological properties” of the first person plural. If this is right, then the special character of those statements with which we say ‘what we say’ lies, not in their content, but rather in their form. 6

7 8 9

This is, of course, not uncontroversial. See for example the Cavell-critique of Jerry Fodor and Jerrold Katz; Fodor/Katz (1963). Fodor and Katz claim that the judgment of a native speaker rests on the very kind of empirical generalization as the judgments of a linguistic field researcher. Their critique of Cavell’s view of our knowledge of language rest, inter alia, on the assumption that thinking is prior to language. Cavell denies this. For the purposes of this paper I will presuppose that Cavell is right about this. This obviously does not mean that perception plays no role whatsoever. Understanding the utterances of others obviously involves hearing what is said. Cavell (1976a), 38. Cavell (1976a), 14.

Three Forms of the First Person Plural

233

Although Cavell does not immediately pursue these programmatic remarks further, it is already clear from this passage that, on his view, the philosophy of language could hardly be guilty of a greater omission than this neglect of the first person plural. For if the situation is as Cavell describes it, then only the investigation of the first person plural can clarify what it is to speak a language. Reflection on the considerations that a field linguist might carry out could contribute nothing to the clarification of this question. His judgments would prove to be strictly derivative from S-type judgments, in that they would have to already contain the latter. According to our foregoing reflections, the notes that our linguist takes in the field are descriptions of the linguistic behavior of the natives – and thereby ‘evidence’ for his theory about their language – only if his notes represent their behavior as springing from a understanding of their own language; since native speakers don’t depend on evidence of any kind in articulating how they use language, the linguist’s description must represent speakers as subjects who could themselves frame statements of the type: ‘We use E thus-and-so.’10 In a certain sense, accordingly, each speech act must already contain judgments of type S. Given that the utterance ‘Fa’ is only an assertion if the speaker knows that he thereby claims that Fa, this much follows from the generality of concepts. For an act of mind to exhibit the internal structure characteristic of thinking of an object a that it is F, the predicative element, ‘F__’ must be such as to potentially appear in infinitely many other judgments. It must be possible to deploy the predicate in other acts of affirmation or denial: in affirming or denying that Fb, Fc etc.11 The concept sets no limit as to how often it can be deployed; it reaches beyond any given act of mind in which it occurs and points to what is in principle an unlimited multiplicity of judgments each of which exemplifying how the concept figures in thinking. Given that it is possible for another thinker to deny what I affirm, the concept I deploy must be such as to potentially appear in the judgments of infinitely many other thinkers. For two subjects can only stand in a relation of agreement or disagreement in matters of truth, if the same concept can figure in both of their acts of mind. A concept is not only general in that it sets no limit as to how often it can be deployed in a judgment, 10 Native speakers, Cavell writes, “do not, in general, need evidence for what is said in the language; they are the source of such evidence” (Cavell (1976a), 4). 11 Gareth Evans calls this condition “the Generality Constraint”, see Evans (1982), 100 ff.; compare Cavell (1979), 185.

234

Matthias Haase

but also in that it sets no limit as to by how many subjects it can be deployed. It thus reaches not only beyond any given act of mind in which it is figures; it also reaches beyond any particular thinker and points to an infinite multiplicity of possible thinkers. The judging subject has to conceive of her act not only as instantiating a general pattern that is operative in her; she must conceive of her act as an instance of something that can also be exemplified by the acts of others. In understanding the concept I deploy in my judgment I thus reach beyond my current act of mind to an infinite series of other judgments by me and other thinkers. Switching to the linguistic register we can express the point like this. To be able to think ‘I assert that a is F ’, one must be able to think thoughts of the form ‘We use F thus-and-so’. The two judgments are not independent from each other; they are intrinsically connected. Or more precisely, the thought ‘I assert that Fa’ contains the thought ‘We use F thus-and-so’. One can only say ‘I’ if one can also say ‘we’, and vice-versa. Self-consciousness is always at the same time a ‘we’-consciousness, or, as Hegel would put it, it is “an I that is We and a We that is I”.12 It is often claimed that we cannot fully describe our use of words, but rather can only give examples. This may be so. But it does not affect the above point. An example is something particular that is presented as an instance of something general. To make clear that the movement that I am now performing is supposed to be an example of a tango step, I say ‘Look: this is how to dance the tango’. To understand what I am aiming at with my awkward twirling motions means to understand that my movement is supposed to exemplify how one dances the tango. Correspondingly, to use a given sentence ‘Fa’ containing the word ‘F’ as an example of the use of ‘F’ means to think a thought of the form “We use ‘F’ thusly: We say ‘Fa’; we do not say …” The program suggested by Cavell’s remark on the “logico-epistemological properties” of the first person plural will become clearer if we situate it in relation to contemporary discussion. Recently, the first person plural has received a great deal of attention. Robert Brandom, for example, begins his book Making It Explicit with the thesis that rational beings are the beings who say ‘we’. He adds, however, that this is only an empty formula waiting to be filled with content. The goal must be to “specif[y] in other terms what one must be able 12 Hegel (1987), 145.

Three Forms of the First Person Plural

235

to do in order to count as saying ‘we’”.13 Brandom holds that the first person plural is not logically simple; he thinks it needs an analysis which decomposes it into its elements. Correspondingly, he thinks that the statements that describe a linguistic practice are distinguished by their content, not, as Cavell suggests, by their form. The project of Making It Explicit is to define their content in a non-circular way – that is, without using semantic and intentional concepts – by specifying what sort of structure a social practice must have to count as a practice in which verbal utterances are assigned the significance of assertions. Brandom’s thesis is that a linguistic practice can be characterized in terms of a complex social structure of interdependent ascriptions of commitment and entitlement taking place between individuals. On this view, the ‘we’ that statements of type S bring to expression is secondary to the assessmentbehavior of particular individuals; it is instituted by their interdependent normative attitudes.14 If Cavell is right, then such a project is doomed to failure, since one cannot describe those individual attitudes without already referring to the very ‘we’ that these attitudes are supposed to institute.

3. Do We Know What We Mean? Nothing I said in the last section decides the question that came up in Section 1: whether statements of type S, when they are correct, express a special kind of knowledge or just something that underlies our knowledge. We have indeed encountered a thesis concerning how this question might be decided – namely, through an investigation of the “logicoepistemological properties” of the first person plural. But since we still remain in the dark exactly which properties these are, we have not yet moved beyond the negative characterization of this form of judgment. Now, from the fact that S-type statements do not rest on evidence, Cavell infers that they do not express knowledge about the world: “[The speaker] is not claiming something as true of the world, for which he is prepared to offer a basis […]; he is claiming something as true of himself […], for which he is offering himself, the details of 13 Compare Brandom (1994), 4. 14 Brandom (1994), 39: “Assessing, approving, and so on are all things that we do individually and grant others the right to do, and thereby a community, a ‘We’, is constituted”.

236

Matthias Haase

his feeling and conduct, as authority.”15 Now, this thesis must initially provoke astonishment, since it concerns statements in the first person plural, not the first person singular. As Cavell himself emphasizes, a statement of type S is “not […] an assertion about how a word is used by me (or ‘some given person’),” but rather “a statement about how the word is used in English.”16 At first glance, these two claims appear to stand in tension with one another. According to Cavell, this tension is resolved as soon as we recognize that to put forward a statement of this kind means “presenting [oneself] as example, as the representative of the community.”17 Thus, for Cavell, the central question to be answered in an investigation of judgments of type S is the question how I, on the basis of my “experiences,” can speak for others: “How can I, what gives me the right to, speak for the group of which I am a member?”18 Cavell’s answer has two parts, one negative and the other positive. The negative part can be summarized as follows: As we have seen in the preceding section, each speech act must, according to Cavell, already in a certain sense contain judgments of type S. In consequence of this internal relation between ‘I’ and ‘we’, judgments of type S cannot rest on comparative judgments in which I state that others speak as I do. Cavell draws the conclusion that my claim to speak for others cannot be founded on knowledge.19 The positive part of his answer then consists in the account of those statements as expressing “claims to community”: When [a speaker] ‘says what we say’, what he produces is not a generalization […], but a (supposed) instance of what we say. We may think of it as a sample. The introduction of the sample by the words ‘We say…’ is an in15 Cavell (1979), 179. In the case at issue the speaker is a philosopher articulating our ordinary usage. The cited passage thus relates specifically to statements that describe, not the use of words in a certain language, but rather the mode of employment of concepts that are common to all human languages. Since Cavell himself ultimately distinguishes the two types of judgments only in terms of the scope of their claim to generality, we can proceed on the assumption that this remark also is supposed to apply to that less embracing “we” that concerns us here. 16 Cavell (1976a), 37. 17 Cavell (1990), 72. 18 Cavell (1979), 18. 19 In the context of distinguishing different ways of understanding the question how taking oneself as one’s starting point, one can “speak for others,” Cavell writes: “if the question means ‘How do I know at all that others speak as I do?’ then the answer is, I do not” (Cavell (1976b), 67).

Three Forms of the First Person Plural

237

vitation for you to see whether you have such a sample, or can accept mine as a sound one.20

The claim that I make with a statement of type S can never be secured by me alone, because my judgment requires the agreement of the others: the thought that I express depends in its constitution on the fact that my interlocutors also think it. If my interlocutor withholds his agreement, even after lengthy argument, then “we have to conclude that”, as Cavell writes a few lines later, “on this point we are simply different; that is, we cannot here speak for one another.” In this case my judgment has turned out to be not false, but rather empty, for it has become clear that “there is no us […] to say anything about.”21 Were this to transpire not just occasionally but in general, there would be nothing left for me to say or think, since judgments of type S are, as we have seen, constitutive for the content of my other judgments. What consequences this interpretation of the grammar of S-type statements has comes out in the way in which Cavell argues for a thesis that we presupposed in the foregoing derivation of the internal nexus between “I” and “we” – namely, the claim that human thought and speech is only possible against the background of a social practice. Cavell writes: We learn and teach words in certain contexts, and then we are expected, and expect others, to be able to project them into further contexts. Nothing ensures that this projection will take place (in particular, not the grasping of universals nor the grasping of books of rules), just as nothing ensures that we will make, and understand, the same projections. That on the whole we do is a matter of our sharing routes of interest and feeling, senses of humor and of significance and of fulfillment, of what is outrageous, of what is similar to what else, what a rebuke, what forgiveness, of when an utterance is an assertion, when an appeal, when an explication – all the whirl of organism Wittgenstein calls “forms of life”. Human speech and activity, sanity and community rest upon nothing more, but nothing less, than this. It is a vision as simple as it is difficult, and as difficult as it is (and because it is) terrifying.22

What Cavell here calls “terrifying” and elsewhere calls “the uncanniness of the ordinary”,23 is the “vision” that the foundation of our thought and speech consists of nothing but the brute fact of our basic agreement 20 21 22 23

Cavell Cavell Cavell Cavell

(1979), 19. (1979), 20. (1976b), 52. (1988).

238

Matthias Haase

– our “mutual attunement”, as he expresses it in another passage.24 There is, according to Cavell, no explanation for the fact that we agree with each other in this fundamental way; it is a groundless factum that we must simply accept, and that holds only insofar as we accept it. Since Cavell finds this “terrifying” – rightly, it seems to me – he reaches the conclusion that the recognition of the condition of the possibility of our concepts is equally the source of a permanent possibility of skeptical estrangement from our concepts: If the fact that we share […] criteria is the condition under which we can think and communicate in language, then skepticism is a natural possibility of that condition; it reveals most perfectly the standing threat to thought and communication, that they are only human, nothing more than natural to us.25

This insight into the conditions of the possibility of our concepts can become a disappointment at finding no solid foundation at the basis of our practices. In this mood we begin to doubt whether we can gain knowledge about the world (among other things, about other minds) using our concepts, and thus become estranged from our “agreement” with others and consequently from our own language.26

4. Epistemological and Metaphysical Skepticism About Meaning In the long passage from “The Availability of Wittgenstein’s Later Philosophy” quoted above, Cavell arrives at the thesis of the uncanniness of the ordinary via three steps. First, he raises a problem; then, he presents the solution to this problem; finally, he refers to the shudder of terror that this solution must evoke in us. Whether the third step is inevitable evidently depends on whether what Cavell offers in the second step is in fact capable of solving the problem that he formulates in the first step. Thus the skeptical doubt that, according to Cavell’s diagnosis, can arise from our recognition of the baseless basis of our language might not relate to the problem that stands at the beginning of the passage. It is crucial to distinguish clearly between these two issues. While the skepticism of which Cavell speaks is directed toward the possibility of 24 Cavell (1979), 168. 25 Cavell (1979), 47. 26 Cf. Cavell (1979), 83 ff.

Three Forms of the First Person Plural

239

knowledge about the world, the problem from which this difficulty arises concerns the constitution of conceptual content. The suspicion that there may be no solution to the more basic problem would be a ground, not for an epistemological, but rather for a metaphysical skepticism. For what is called into question when the constitution of meaning is questioned is not how we can recognize the truth of our judgments (concerning, among other things, what others mean by their words), but rather how our judgments can have any content at all, whether true or false. Cavell tends not to distinguish sharply between these two questions.27 But to be able sensibly to pose the epistemological question, we must already have answered the metaphysical one. The ordinary can thus be “uncanny” in the sense described above only if reference to our “mutual attunement” actually makes the constitution of conceptual content intelligible. We will see that this condition is not fulfilled, and that Cavell’s position consequently leads to a metaphysical skepticism. Let us first be clear about what the problem of the constitution of meaning is. Cavell formulates it by means of the question how we are able to apply a word to new cases when we are introduced to it only in a limited number of situations. Ultimately, however, the problem is independent of the scenario of language-acquisition. What is in question is the generality of the concepts we acquire in learning a language. A concept, as we said above, can be deployed in an in-principle-unlimited series of judgments. In order for a given sequence of judgments or statements to be graspable as applications of one and the same concept to diverse objects, the sequence must belong to a practice of going on in the same way. Cavell’s question is what constitutes this sameness: What determines that going on in just this way, and not that way, counts as going on in the same way? The intuitive answer to this question is: the concept that is applied in these judgments. A concept, like a rule, divides its applications into the class that agree with it and are in this sense “correct,” and the class that do not agree with it and hence are “incorrect.” And just as to count as following a rule, one must not merely do something that conforms to the rule, but act from an understanding of the rule, so also to count as applying a concept in a judgment, one must be guided by one’s understanding of that concept. Hence, in the cases where the application accords with the concept, it is wholly explained by this concept. Assuming I am in fact following the rule, the answer to the question why I write “2002, 2004, 2006, …” when extending 27 See also Conant (2004).

240

Matthias Haase

a series of numbers according to the rule “Add 2” can only consist in this: that the rule “Add 2” requires this. Any other answer would call into question the intelligibility of the very idea of conceptual content. For only if a concept determines its applications in the sense just described does it have any content whatsoever; otherwise its application would be simply arbitrary. The difficulty is to see how we can hold onto this intuitive answer while doing philosophy. If we acquire concepts by learning a language, then the meaning of the words of our language must determine their application to particular contexts. But it seems strange to say that words stand in an explanatory and normative relation to their employment in statements. How can a word explain its application? Taken by themselves, one might think, words don’t determine how they are deployed. Consequently, it seems that there is a gap between a word and its application to particular circumstances. It might be thought that the gap between words and their employment can be bridged by introducing a mediating element, something that constitutes the relation of agreement between words and what they describe. Cavell mentions two possible candidates: “universals” and “books of rules.” As is well known, an appeal to the latter leads to an infinite regress, since these rules must themselves be applied in particular situations, and so further rules are required to determine how this should be done. No matter what linguistic or mental objects one introduces, they require determination just as much as the words to which they are supposed to supply determinacy. And so it seems as if our only hope were to appeal to objects of a special kind – Cavell calls them “universals”. But whatever such objects might be exactly, their minimal characterization already rules out their contributing to the resolution of our problem: as abstract objects they cannot causally interact with individuals in space and time and thus cannot explain the acts of individuals that are in space and time. To solve the rule-following problem, we must show how it is possible to avoid the forced choice between an infinite regress and a platonic mythology. This is possible only if we give up the idea that the relation between a word and its employment is mediated by a third element; this relation can consist in nothing other than our practice of going on in the same way. Initially, though, it is difficult to see how this can be. Our practice must be something other than the sum of our various speech acts. For what seems puzzling to us in a particular speech act must seem equally puzzling in each further speech act.

Three Forms of the First Person Plural

241

After all, our puzzle was precisely how a word can determine its applications to particular circumstances. Cavell appeals, not to a sum of speech acts, but to something that can be instantiated in an in-principle-unlimited series of particular acts: our “our sharing routes of interest and feeling, senses of humor and of significance and of fulfillment [etc.].” What is common to the items on Cavell’s list is this: they all belong to the category commonly called “disposition” – that is, to the category of thing that embraces, not dateable acts and states of individuals, but rather general tendencies that are manifested, exemplified or actualized in particular acts and states. A statement that describes such a tendency is a habitual statement. Such a statement does not say what an individual is now doing or did yesterday or has done at any particular point in time; it says what the individual does in general. It is in this sense a timeless statement; or we can also say it is “time-general.”28 Habitual statements provide us with a first clue to the solution of our problem, since they describe something general that is not abstract but actual and that explains its own instantiations. But this first step is not yet sufficient. The category of dispositions is too broad. Habits possess the sort of generality we have described, yet they do not stand in a normative relation to their manifestations. Now, one might be tempted to say: “Of course, the sort of ‘sense of …’ that we are discussing is something other than a blind disposition – let us call it a ‘self-conscious disposition’ or a ‘conceptual capacity.’” But so long as we characterize this difference only by appealing to such concepts as ‘intelligence’ and ‘normativity,’ we simply presuppose the very thing we want to understand. What we need is an account of what kind of disposition this must be, if it is to have the sort of relation to its manifestations that a concept has to its applications.29 The later Wittgenstein suggests that this issue becomes clear once we see individual dispositions against the background of a social practice, “use” or “custom”: “To understand a language means to master a technique.”30 On this view, the idea of a conceptual capacity is internally related to the concept of a social practice, inasmuch as a social practice is the very thing one “masters” when one comes to possess such a capacity. In order for this clue to lead us to a solution to the rulefollowing problem, we must understand what a social practice or “custom” is. 28 Cf. Thompson (2001). As well as Rödl (2005), 173 ff. 29 Compare Kern (2006), 184 ff. 30 Compare Wittgenstein (1984), §§198 ff.

242

Matthias Haase

Cavell attempts to elucidate the notion of practice in terms of our “mutual attunement”. He thus seems to maintain that the condition of the possibility of the relation of agreement between a rule and its application is the relation of accord between us – specifically, between our “senses” of when a rule is applicable.31 This raises the suspicion that Cavell’s picture of language is not merely “alarming” but hopeless. John McDowell expresses this suspicion as follows: “What Cavell offers […] looks like a congruence of subjectivities not grounded as it would need to be to amount to the sort of objectivity we want if we are to be convinced that we are really going on in the same way.”32 Talk of our basic agreement about when a rule has been followed cannot make intelligible the concept of a relation of agreement between a rule and its application, since such talk already presupposes this concept. Two subjects can agree or disagree in their judgments about whether a number series has been continued in the same way only if their judgments have content. And their judgments can have a determinate content only if they spring from an understanding of the rule in question. To assess whether an action is “correct” or “incorrect” according to a certain rule involves bringing oneself under the same rule one applies to the agents one assesses. The concept of a relation of agreement between the rule and its application and talk of agreement between our judgments stand in an internal relation to one another. The assumption that reference to the latter can clarify the possibility of the former in any way whatsoever makes sense only if we believe that it is possible to break out of this circle. In that case, we must presuppose that talk of our agreeing in our “sense” of when a rule has been followed is logically independent from the concept of rule-following. The consequence is that we can no longer regard the manifestations of this “sense” as judgments in the full sense; we must rather conceive of them as initially nothing more than manifestations of blind dispositions. To add the requirement that these blind dispositions of particular individuals stand in a certain relation to one another cannot alter the basic picture. For however complex the structure of these relations, a social constellation of mutually reinforced blind dispositions cannot constitute 31 “In order for there to ‘be’ such things as rules, we have to agree in our judgment that a rule has been obeyed (or not)” (Cavell (1979), 36). 32 McDowell’s remark is not meant as a critique of the passage from Cavell I quoted above, but as a rejection of a philosophical assumption that forces such a reading on the thought expressed in passage. McDowell (1998), 207.

Three Forms of the First Person Plural

243

conceptual content. On such a view, our practice no longer appears as an application of concepts that stands under an objective norm, but rather merely as the accidental concurrence of subjective habits, concerning which there can be no “correct” or “incorrect.”

5. The Distributive We Now this difficulty is not lost on Cavell. On the contrary, he criticizes the so-called “skeptical solution” that Saul Kripke ascribes to Wittgenstein using precisely the argument just sketched: “If the matching of inclinations is all Wittgenstein’s teaching leaves us with”, then this solution is “more skeptical than the problem it is designed to solve.”33 Cavell thus seems to understand his own position as been different from the view discussed in the last section. On further consideration, this is not surprising, since the picture described at the end of the last section contradicts the program that we sketched in Section 2. Cavell denies that it is possible to explain the concept of rule-following in a non-circular way; he aims to clarify this concept through an investigation of the “grammar” of judgments of type S. 34 Our argument has simply returned us to the point we reached at the end of Section 2 – the thesis that each statement made by a subject must contain judgments of type S. Given our recent reflections, we can now express this as follows: The normative attitudes of subjects cannot institute any norms, since to adopt a normative attitude means to relate oneself to an already existing norm. Since norm-platonism is unintelligible, the unity of the norms that underlies our speech acts and judgments can be nothing other than our practice itself. Statements of type S articulate this practice; hence they cannot be regarded as prescriptive statements, which express a subjective attitude that this or that should be thus-and-so. Rather, they must be understood as a special kind of descriptive statement.35 So much for our results to this point. It remains unclear what kind of descriptive judgments these must be if they are to be contained in each of our judgments while also describing something that stands in a normative and explanatory relation to our judgments. According to Cavell – if I am reading him correctly – clarifying this requires investigating the 33 Cavell (1990), 75. 34 Cavell (1990), 68. 35 Compare also Cavell (1976a), 21 – 35.

244

Matthias Haase

“logical-epistemological properties” of the first person plural. To be sure, without further explanation this suggestion can at best provoke astonishment. The fact described at the beginning of Section 1 – that first person pronouns cannot be replaced salva sensu by descriptive expressions such as “speakers of English” – is widely accepted. But how, beyond this, an investigation of the first person plural might make progress toward its designated goal is anything but clear. Imagine the following case. Suppose a prison guard remarks while on his rounds: ‘The prisoners in Cell 20 have red dots on their noses.’ Thereupon one of the celloccupants looks in the mirror and then at his sleeping cellmates and says: ‘We have red dots on our noses.’ The two statements are evidently not interchangeable salva sensu. Nevertheless, the guard and the prisoner know about the red dots in the same way: by observation. There is certainly something the prisoner can say without observation. His statement implies the statement ‘I have red dots on my nose’. To establish whether he indeed does have red dots on his nose, he relies on a glance in the mirror; but that he is the person of whom he asserts that he has red dots on his nose – this he need not and cannot discover by observation. Precisely this distinguishes the first person singular formally from demonstrative pronouns and definite descriptions.36 In both of the latter cases the judging subject always needs to make a further judgment to determine that he is identical to the object to which his judgment refers. In the first case, by contrast, the knowledge that he himself is the object of which the concept ___has red dots on the nose is predicated is not mediated by such an identity-judgment; rather, it is already contained in his original judgment. If further investigation were required here, our cell-occupant could never discover that he is speaking of himself. For each mediating judgment ‘I am this one’ would require a further identity judgment, and no matter where he looked, nothing could show him that he was identical to the object to which he was referring. The knowledge that he is predicating the concept ___has red dots on the nose of himself must therefore enter into the constitution of his judgment ‘I have red dots on the nose’. To judge in the first person singular involves knowing, in virtue of the form of one’s judgment, that one is oneself the object characterized in that judgment. Let us call judgments of this form self-predications. Perhaps this much will be granted. But doubts about the feasibility of our program cannot be waved away simply by appealing to this point. 36 See Anscombe (1981).

Three Forms of the First Person Plural

245

For how could the prisoner, in judging ‘We have red dots on our noses,’ have non-observational knowledge of the existence of his cellmates? After all, he is not identical with them! Whatever “significant logicalepistemological properties” the first person singular may have, they do not seem to reappear in the first person plural. We can gather from this consideration what must be the case if our program is to be feasible. Our red nose scenario must be unsuited to bring out the “logical-epistemological properties” of the ‘we’ that stands in the subject-position of S-type statements. That is to say, there must be another form of the first person plural. The judgment ‘We have red dots on our noses’ can be analyzed into several judgments that are independent of one another. It can be represented as a conjunction of the following singular judgments: first our prisoner notices that he himself has red dots on his nose; then he brings prisoner a under the concept ___has red dots on the nose; and finally he brings prisoner b under this concept. He then unites this series of independent singular judgments into a plural judgment, in which he refers with ‘we’ to the set of individuals of which he is an element. The plural judgment is in this sense distributive. Its truth-value is wholly determined by the truth-values of prior singular judgments. Any distributive statement will be inappropriate for our purposes, no matter what concept we put in place of ___have red dots on our noses. For if ‘we’ is distributive, then it cannot be contained in ‘I’. It is widely assumed that all uses of ‘we’ are structurally analogous to the one just discussed. I will call this view the ‘Distributive Conception of the first person plural’. That statements of type S differ from statements like ‘We have red dots on our noses’ is plain enough. While the latter is a statement about the state of individuals a, b, c at a certain time t, statement S is not an assertion about what particular speakers said at a certain time; it rather describes how speakers of English use the relevant expression in general. No one will deny this. The decisive question is how to understand this generality. Contemporary philosophy of language tends to assume that all generality is quantificational. On this assumption, there is no other alternative but to analyze statements of type S as some sort of quantification over individual speakers and their relations. However the details of such an analysis run, it will be structurally analogous to the distributive analysis, inasmuch as the truth-values of the quantified judgments are determined by the truth-values of the corresponding particular judgments. The latter must thus be regarded as prior and cannot already contain the former.

246

Matthias Haase

The understanding of the concept of social practice that results from the distributive conception leads inevitably to a picture akin to the one described at the end of Section 4. It makes no difference if one claims, like Brandom, that each articulation of the practice must be regarded as itself taking up a normative position in the practice.37 Brandom understands ‘we’ as a “recognitional quantifier.”38 That is, ‘we’ is a quantifier whose domain is defined by a certain respect in which I recognize myself in others. In the case of our sample sentence ‘We have red dots on our noses’ the respect in which I recognize myself in others is determined demonstratively. It could be put like this: ‘All of us in this room have red dots on our noses’. In other situations the range can be determined by implication or defined by an explicit clause. When our prisoner, for instance, writes a letter about the curious events in prison he might put it like this: ‘We, the prisoners of cell 20, have red dots on our noses’. Now, things get more complicated when the sentence includes verbs like ‘claim’ or ‘believe’ etc. To use Brandom’s favorite example, ‘We pragmatists believe…’: in this case the ‘recognitional quantifier’ enables me to attribute a commitment to others that I undertake myself with this very statement. Entailed in this statement is the relational predication ‘___agrees with___’, where one pole of the relation is ‘I’ and the other pole is whoever else falls into the range of my quantifier.39 On this account, then, ‘we’ is not logically simple, in that it can always be replaced salva sensu. In case all the members of the ‘we’ are the addressees of my speech act, the range of the quantifier can be defined in terms of ‘All of you and me…’. If the addressee is someone else, the relevant effect can be achieved by pairing ‘I’ either with a demonstrative or with a definite description as in ‘I and all the other pragmatists believe….’ or, to put it in the more polite form ‘All pragmatists, including me, believe…’. According to Brandom’s view, the statements of type S differ only in respect to the predicate from statements like ‘We pragmatists believe…’ or ‘We have red dots on our noses’; the subject-term appearing 37 Compare Brandom (1994), 639 ff. 38 In personal conversation. 39 In the case of statements of type S the quantifier would have to range over every speaker of the language. Accordingly, ‘We use E in such-and-such a way’ would read as a normative statement in which I ascribe to every speaker of English a commitment (namely the commitment/entitlement to use expression E in such-and-such a way) that I thereby undertake myself. This includes a relational predication ‘___attributes commitment C to___’.

Three Forms of the First Person Plural

247

in these statements plays the same logical role each time. But while the sketched analysis seems, as far as it goes, adequate for statements of the latter kind, a paradox arises if one tries to apply it to statements of type S which are supposed to articulate the norm that constitutes the content of my thinking. For, since ‘we’ is supposed to be analyzed into the components ‘I’ and ‘all the other participants of the practice’, it cannot be contained in ‘I’. In consequence, I have to conceive of the normative attitude that I express with this statement as logically prior to the norm under which I bring myself by adopting such an attitude. Since this is impossible, I ultimately have to give up conceiving of my attitude as an act of bringing myself under a norm. What remains is to conceive of it as the manifestation of a mere impulse or ‘blind disposition’. Cavell’s deep insight is that a judgment of type S cannot be a “generalization,”40 since the corresponding singular judgments must already contain it. It must, therefore, involve some other form of the first person plural. This discovery will furnish us with understanding, however, only if we can also say what form this is. On closer inspection, there are various types of plural judgment that resist distributive analysis.41 The sentence ‘Prisoners a, b, and c are cellmates’, for instance, cannot be analyzed either as a conjunction of singular predication or as a conjunction of relational predications. What is true if it is true is not something that holds of all three because it holds of each one individually; it can only hold of them all together. ‘We are cellmates’ is thus an example of a non-distributive use of ‘we’. But since our prisoner is evidently not identical to his cellmates, this example does nothing to alter the impression that his reference to the other two can only be demonstrative.42 It is obviously not sufficient to say that the ‘we’ in our judgments of type S is not distributive; we need a positive account of what it is instead.

40 Cavell (1979), 19. 41 See McKay (2006). 42 It is significant for the following reflections that the prisoner in both cases addresses his utterance, not to the other cellmates, but rather to the guard or some other person, someone who is not contained in the relevant “we.” Even though one refers with “we” only to persons – that is, to objects one could address as “you” – not every “we” contains an implicit address to a “you.”

248

Matthias Haase

6. The Communal We Cavell attempts to demonstrate the difference between his view and the picture described at the end of Section 4 by distinguishing different interpretations of the concept of agreement. In the controversy with Kripke he introduces his program with the question “whether the communal agreement Kripke finds in Wittgenstein reaches the depth of confidence Wittgenstein places in our ‘agreement’.”43 His diagnosis is that Kripke’s mistake lies in a “contractualizing or conventionalizing” of the idea of agreement. For the relevant agreement is a matter of “our natural reactions.”44 In The Claim of Reason, Cavell formulates the contrast as follows: The idea of agreement here is not that of coming to or arriving at an agreement on a given occasion, but of being in agreement throughout, being in harmony, like pitches or tones, or clocks, or weighing scales, or columns of figures. That a group of human beings stimmen in their language berein says, so to speak, that they are mutually voiced with respect to it, mutually attuned top to bottom.45

What Cavell rejects is the idea of elements that are determinate prior to the relation of agreement and that are “coming to agreement on a given occasion”. Rather, there is supposed to be an agreement among elements, which, like our cellmates, are what they are only as elements of this relation – only in their “mutual attunement”. But this distinction alone cannot make the possibility of judgments of type S intelligible: the bees in a beehive are mutually attuned, yet they have no concept of the beehive. We retain a relevant difference from the picture described in Section 4 only if we introduce the further condition that the “attunement to one another” be self-conscious. If we do this, then the relation of agreement that constitutes the elements as such must be a relation that only obtains if it is thought by its elements to obtain. That is to say, subjects can judge only if their dispositions that they instantiate in so doing stand in a relation of agreement to one another; and they stand in such a relation only if these subjects think they stand in this relation. In judging, the subjects must thus understand themselves as “mutually attuned.” This is exactly the interpretation of the internal relation between “I” and “we” that we encountered in Section 3. Cavell’s thesis 43 Cavell (1990), 67. 44 Ibid., 94. 45 Cavell (1979), 32.

Three Forms of the First Person Plural

249

is consequently this: In thinking ‘I assert that p’ I implicitly think ‘We use E thus-and-so’; and thinking this latter thought, I think a thought that depends in its very constitution on the fact that my fellow speakers think it as well. Now it may seem doubtful whether there are any thoughts at all which depend in their very constitution on the fact that others also think them. But this doubt would be misplaced. Communal actions provide an example. Imagine the inmates decide to revolt against the inhuman conditions in prison. The ‘we’ of their revolutionary act doesn’t refer to a mere sum of agents. For, this act they can only perform together. Similarly the statement ‘We are dancing a tango with each other’ cannot be analyzed distributively. The concept of dancing a tango cannot be predicated in singular judgments; only a plural subject can fall under this concept. For, I can only dance a tango with you if you dance a tango with me. It follows that our activity is a case of dancing a tango only if we both think that we are dancing a tango with each other. If this is right, then we cannot discover by observation that we are doing so. For that would require that each of us grasps the thought ‘We are dancing a tango with each other’ independently of the other. But dancing a tango is not something one does to another person; it is something one does together with another person. My judgment ‘I’m dancing a tango with you’ must therefore already contain your corresponding judgment that you are dancing with me. Otherwise our turn on the dance floor would be something that we each do in relation to the other, not that harmonious movement which two can only perform together. We can put this as follows: my relation to you, on the basis of which you can appear in my thinking as my partner in action, consists in the fact that you think of me as your partner in action. If you do not think this, then my judgment remains incomplete and is in the end empty.46 Since, for this reason, perception cannot be the source of my awareness of the other, the relevant ‘we’ cannot be constituted from a first person singular and a demonstrative reference to another person. That you are the one my judgment is about is not something I can first discover through further investigation; this knowledge must enter into the constitution of my judgment. Consequently we cannot think the thought ‘We are dancing a tango with each other’ independently of one another; we can only think it together. If we do this, then we do not make two judgments with the same content; we make the 46 On this logical structure see also Rödl (2006).

250

Matthias Haase

very same judgment. Our judgment is a communal self-predication – that is, a judgment in which you and I have, in virtue of the form of our judgment, an awareness of the unity that we together constitute. Each of us is in this sense an “I that is We and a We that is I.” If this admittedly somewhat sketchy line of thought is right, then we have identified a form of the first person plural that is distinct from the distributive form. Let us call it the Communal We. Whether the sort of “we-intentionality” just described is irreducible is a matter of controversy in the literature. We need not enter into this debate here. For only if it is irreducible can the talk of our “accord with one another” be distinguished from the picture described in Section 4. Now, Cavell certainly does not claim that a natural language is a communal action. He propounds the following thesis: A natural language is the communal disposition of its speakers. One must defend such a thesis if one thinks that the first person plural in statements of type S stands for the community of speakers of the relevant language. For if the use of ‘we’ in statements of type S is the same as the one in statements articulating communal actions, then the difference between the two forms of statements can only consist in their temporality. Statements of type S must consequently be a special sort of habitual statement. They express judgments in which a communal subject is brought under a concept in a time-general manner, and indeed in such a manner that a thought is grasped only if this judgment is made together by the members of the community – that is, only if the judgment is an act of a communal subject. Judgments of type S would accordingly be habitual communal self-predications. Only if this is so do we have the two features that Cavell takes to characterize such statements: (i) Native speakers are in a position to frame these statements without receiving any evidence whatsoever (see Section 2); and (ii) The judgment depends on the agreement of the other members of the linguistic community (see Section 3). I do not wish to dispute that there are judgments that exhibit these logical properties. If the category of habitual communal self-predications turned out to be a grammatical fiction, much that we commonly take to be real would turn out to be mere appearance. For the sentence ‘We play in an orchestra’ seems to be an example. ‘We are having a love affair’ is another, ‘We fight for justice’ a third one. In the present context, however, the only relevant question is whether appealing to this form of judgment supplies us with the right interpretation of the grammar of our statements of type S.

Three Forms of the First Person Plural

251

One of the considerations that leads Cavell to this claim is the observation that communicating is a “cooperative activity.”47 This thesis seems quite plausible. For I can tell you something only if you think that I am telling you something, and not making a joke or talking with somebody else. But it is another question altogether whether the language that we speak when we communicate is the disposition of a communal subject. Our brief reflection on communal actions at the beginning of this section suffices to make clear why it cannot be. The unity of multiple subjects in a communal subject or a ‘community’ must have a ground or source. If two people are dancing a tango with each other, then it can be no accident that they both think that they are dancing a tango with each other; they must make one judgment together. To say this is in the first instance merely to describe what communal actions are; it does not yet explain how they are possible. To make their possibility intelligible, we must explain what unites individuals into a communal subject of judgment. In our example, the following answer suggests itself. What makes the harmony of our movements on the dance floor possible is the fact that we both have learned to dance the tango. In other words, our “attunement” is constituted by the fact that our thinking has the same source: the practice of tango-dancing. Our unity in action is secondary to the practice of which our action is an actualization. No doubt, there is not for each communal action some established practice that underlies it. But however a general theory of collectivity might look, the ground of the unity of a community cannot consist, in the basic case, of relations between its members. Rather, there must be something that enables their attitudes and dispositions to come together in this peculiar way. Their unity in thought must have an underlying source. On reflection it is clear that this can only be the concept they employ in their common judgment. The idea of the Communal We thus presupposes the possibility of following a rule, and therefore can contribute nothing to the clarification of the latter. As soon as we advert to the Communal We to clarify the relation of agreement between a concept and its employment, we rob ourselves of all hope of explaining how it can be anything but an accident that our movements on the dance floor harmonize so well. The resulting picture cannot, in the final analysis, be distinguished from the picture of Section 4. For to say that there is no explanation of our agreement 47 Cavell (1976a), 33.

252

Matthias Haase

in our “sense” of when a rule has been followed is an expression of metaphysical skepticism. There is something that explains our agreement: the rule under which we bring ourselves whenever we assess whether that rule has been followed in a given case. On shouldn’t think that we have overlooked an option. If one begins the investigation of judgments of type S with the question ‘How and with what right can I speak for the group of which I am a member?’, then only two possibilities remain: either one understands ‘we’ distributively, and consequently cannot explain how a native speaker can be justified in this judgment without relying on evidence; or else one understands ‘we’ as the expression of a self-conscious community, and consequently can no longer explain what constitutes this community.

7. The Generic We In the remainder of this paper I want to suggest Wittgenstein rejects both, the Distributive and the Communal conception of the social; and that he has good reason to do so. What the Distributive and the Communal Conception of the social share is the idea that the relevant notion of a social practice is to be elucidated by talk about the relations between their speakers – their agreement or disagreement. Cavell realizes that the Distributive Conception of the social leads to disaster. But he thinks, as we have seen, that we will find a way out of trouble if we focus on a different notion of agreement – namely, one that does justice to the “depth of confidence Wittgenstein places in our ‘agreement’”.48 At least as far as Wittgenstein is concerned, this proposal seems to be misguided. For, on closer inspection, it turns out that Wittgenstein doesn’t put much confidence in the notion of agreement at all – at least not in the context of the rule-following problem: It is no use […] to go back to the concept of agreement, because it is no more certain that one action is in agreement with another, than that it happened in accordance with a rule.49

If the invocation of the notion of a social practice is to be of help in the context of the rule-following problem, then it cannot be elucidated in terms of a relation between subjects – no matter how one conceives of 48 Cavell (1990), 67. 49 Wittgenstein (1989), Part VII, § 26.

Three Forms of the First Person Plural

253

such a relation. For, such an approach presents us with a dilemma. If we conceive of the relation in terms that are logically independent of the notion of conceptual content, then we end up with a picture akin to the one sketched in Section 4. And if the relation is of the right kind – namely, one of agreement or “attunement” between the conceptual contents of the acts or attitudes of these individuals – then the appeal to it will presuppose precisely what we want to understand – namely, how the acts and states of individuals can have conceptual content. Wittgenstein is very clear about this. This suggests that he must be employing a different notion of a social practice when he appeals to it in the context of the rule-following problem. In Bemerkungen ber die Grundlagen der Mathematik Wittgenstein writes: “In order to describe the phenomenon of language, one has to describe a practice, not a singular event, no matter of what kind.” And he adds: “This is a very difficult insight.”50 One of the difficulties consists in seeing how there can be any alternative to the description of a singular event besides the description of an ordered set of events. As long as we remain in the logical space defined by these alternatives, we must conceive of the relation between the particular and the general as the relation between the real and the abstract. And then we will think that the description of a natural language can only consist of statements one arrives at by abstracting from linguistic processes. The result is that it becomes puzzling how something general can stand to its instances in an explanatory relation. The employment of a concept in a judgment or a statement can then no longer be explained by the concept, but rather only by reference to other processes. In such an explanation, the relevant process obviously can no longer be taken to be one that is either “correct” or “incorrect.” It must lie on another level altogether. That a judgment stands under a norm then appears as something that does not enter into the constitution of judging, but arrives only with the interpretation of another judger. And it appears that the philosophical clarification of the normativity of language can only consist in an investigation of the relations and interactions between speakers. Cavell sees that judgments of type S cannot be “generalizations,” since we are warranted in making them without receiving evidence of any kind, and since they consequently must already be contained in each of our judgments. To turn this insight into a solution to the rule-following problem, we must investigate exactly what form judgments of type 50 Ibid., Part VI, §34.

254

Matthias Haase

S have, such that they can be contained in each of our judgments. Here we have focused on the subject-position of S-type judgments. In so doing, we encountered the analogue of the assumption that the alternative to describing a singular event is describing an ordered set of events – the idea that the alternative to describing an isolated speaker is to describe a group of speakers. Both versions of this idea have turned out to lead to meaning-skepticism. If the appeal to the concept of practice is to lead to a solution to the rule-following problem, there must thus be a third form of the first person plural that is more fundamental than the Distributive and the Communal We. Developing this form properly would require a separate investigation, but I think our reflections justify us in drawing the following conclusion. Alluding to Wittgenstein, we can express it like this. To describe the phenomenon of language we must describe a practice and not a group of individuals, no matter of what kind. This negative characterization already contains a positive one. For judgments that refer neither to a particular individual nor to a group of individuals, and nevertheless describe them, are judgments that are not only time-general but also have a general subject. They are generic judgments.51 Judgments of type S are judgments of this sort. With them I describe, not a set of utterances of which my utterance is an element, but rather the practice that I exemplify with this statement. If I do this, then I speak, not about or for a group, but rather articulate the general form whose exemplar or ‘bearer’ I am. Cavell presupposes that the existence of a natural language is bound to the existence of several speakers of this language, since one can only acquire conceptual capacities by being initiated into a practice. I don’t want to deny this. To say that something general is not abstract but actual is to say that it exists only as what Aristotle would call the ‘form’ of its instantiations. A practice is a ‘form’ of a specific kind; it exists only as the form of a multiplicity of bearers. But from the fact that the practice I actualize in my statements must also be exemplified by others, it does not follow that I relate myself to these others in articulating the practice. Cavell confuses generic predication with plural predication. The difference between generic and plural predication is orthogonal to the difference between singular and plural predication. A generic statement can occur in the singular or the plural. In a description of chess, generic statements must appear in the plural, since this is a game that people 51 Compare Thompson (1998); as well as Rödl (2005), 188 ff.

Three Forms of the First Person Plural

255

can only play together. By contrast, if we explain how Patience is played, our description will consist of generic statements in the singular. For “One plays Patience alone” – as we knew even before Wittgenstein made an issue of it. This line of thought does not yet explain in what way judgments of type S express knowledge. We have only seen that it must be so. And we have identified an assumption that makes it impossible to comprehend how it can be so – namely, the assumption that the relevant ‘we’ relates me to a group of which I am a member. If the employment of ‘we’ in statements of type S can be neither Distributive nor Communal, then it must involve a Generic We. Judgments of type S are generic self-predications. To make a judgment of this kind means to understand oneself, in virtue of the form of one’s judgment, as an exemplar of something general, of which there are also other exemplars. And even if one is the last human being left on earth: as long as one thinks, one thinks thoughts of the form ‘We use A thus-and-so’ and thereby understands oneself as an exemplar of a practice of which other exemplars once existed.

References Anscombe (1981): Gertrude Elizabeth Margaret Anscombe, “The First Person”, in: G.E.M. Anscombe, Metaphysics and the Philosophy of Mind, Oxford, 21 – 36. Brandom (1994): Robert Brandom, Making It Explicit. Reasoning, Representing, and Discursive Commitment, Cambridge/MA. Cavell (1976a): Stanley Cavell, “Must We Mean What We Say?”, in: S. Cavell, Must We Mean What We Say? A Book of Essays, Cambridge, 1 – 43. Cavell (1976b): Stanley Cavell, “The Availability of Wittgenstein’s Later Philosophy”, in: S. Cavell, Must We Mean What We Say?, 44 – 72. Cavell (1979): Stanley Cavell, The Claim of Reason, Wittgenstein, Skepticism, Morality, and Tragedy, Oxford. Cavell (1988): Stanley Cavell, “The Uncanniness of the Ordinary”, in: S. Cavell, In Quest of the Ordinary. Lines of Skepticism and Romanticism, Chicago, 153 – 178. Cavell (1990): Stanley Cavell, “The Argument of the Ordinary. Scenes of Instruction in Wittgenstein and Kripke”, in: S. Cavell, Conditions Handsome and Unhandsome. The Constitution of Emersonian Perfectionism, Chicago, 64 – 100. Conant (2004): James Conant, “Varieties of Skepticism”, in: Denis McManus (ed.), Wittgenstein and Skepticism, London/New York, 97 – 136. Evans (1982): Gareth Evans, Varieties of Reference, Oxford.

256

Matthias Haase

Frege (1994): Gottlob Frege, “Über Sinn und Bedeutung”, in: G. Frege, Funktion, Begriff, Bedeutung, Göttingen, 40 – 65. Fodor/Katz (1963): Jerry Fodor/Jerrold Katz, “The Availability of What We Say”, in: The Philosophical Review 72 (1), 57 – 71. Haase (2007) Matthias Haase, “Drei Formen der Ersten Person Plural”, in: Deutsche Zeitschrift fr Philosophie 55 (2), 225 – 243. Hegel (1987): Georg Wilhelm Friedrich Hegel, Phnomenologie des Geistes, Hamburg. Kern (2006): Andrea Kern, Quellen des Wissens. Zum Begriff vernnftiger Erkenntnisfhigkeiten, Frankfurt/Main. McDowell (1998): John McDowell, “Non-cognitivism and Rule-Following”, in: J. McDowell, Mind, Value, and Reality, Cambridge/MA, 198 – 218. McKay (2006): Thomas J. McKay, Plural Predication, Oxford. Rödl (2005): Sebastian Rödl, Kategorien des Zeitlichen, Frankfurt/Main. Rödl (2006): Sebastian Rödl, “Is the Second Person Concept a Logical Concept?”, in: R. Bubner/G. Hindrichs (eds.) Von der Logik zur Sprache, Stuttgart 2006, 582 – 593. Thompson (1998): Michael Thompson, “The Representation of Life”, in: R. Hursthouse/G. Lawrence/W. Quinn (eds.), Virtues and Reasons. Philippa Foot and Moral Theory, Oxford, 247 – 296. Thompson (2001): Michael Thompson, “Two Forms of Practical Generality”, in: A. Ripstein/C. Morris (eds.), Practical Reasoning and Preference, Cambridge, 121 – 152. Wittgenstein (1984): Ludwig Wittgenstein, Philosophische Untersuchungen, Frankfurt/Main. Wittgenstein (1989): Ludwig Wittgenstein, Bemerkungen ber die Grundlagen der Mathematik, Frankfurt/Main.

Testimony, Address, and the Second Person* Benjamin McMyler I. Introduction A great deal of what we know and believe is acquired from the spoken and written word, from being told things by people we trust and treat as authorities concerning what to believe. Testimony, as epistemologists call it, is one very common source of knowledge and justification, and learning from testimony is an epistemic capacity that deserves to be classed alongside perception, memory, and inference as one of the paradigmatic ways in which we acquire knowledge and justification. There is one important respect, however, in which the capacity for learning from testimony is very different from the other paradigmatic epistemic capacities. As Thomas Reid puts it, the capacity for learning from testimony is “a social operation of mind”1. Learning from testimony is an epistemic capacity the exercise of which necessarily takes two— a speaker and a hearer. In contrast, perception, memory, and inference are fundamentally solitary capacities. They are capacities the exercise of which takes only one. In acquiring knowledge through perception, memory, or inference, I am utilizing a cognitive capacity the exercise of which is something I do for myself. In a slogan, I am coming to my own conclusion about things. When I acquire knowledge by learning from testimony, however, there is a basic and fundamental sense in which I am not coming to my own conclusion about things. I am not blindly adopting the conclusion of someone else, but neither am I relying solely on my own cognitive resources. Instead, I am exercising a cognitive capacity that is essentially social or interpersonal. But how exactly? How is it that in believing something on the basis of a speaker’s testimony, I am exercising a cognitive capacity that is es* 1

Expanded version of a paper originally published in Synthese (2011). With kind permission from Springer Science+Business Media: © Springer Science+Business Media B.V. 2011. Reid (2002), 68.

258

Benjamin McMyler

sentially social? After all, a speaker’s utterances are ordinary empirical events the observation of which can give me a reason for believing something in just the same way as can my observation of other ordinary objects and events such as a footprint in the snow or a person running from the scene of a crime. So how does testimony, a speaker’s telling an audience that p, differ from something like a footprint in the snow? What is distinctively interpersonal about learning from testimony? Reid holds that the social operations of mind cannot be reduced to or explained in terms of the solitary operations.2 The social character of these social operations cannot be reduced to some combination of the operation of the solitary capacities. Unfortunately, however, he doesn’t have much to say about the precise sense in which these social operations are irreducibly social. There are in fact a variety of ways in which the solitary capacities for inference and perception may require “an intercourse with some other intelligent being.”3 As 20th-Century developments in semantics and philosophy of mind have shown, the reference of our terms and the very content of our representational mental states may depend heavily on what Hillary Putnam has called a “linguistic division of labor.”4 If this is true, then the determination of the content of our beliefs may be said to be “social” in the sense of being constitutively dependent on our linguistic community.5 Additionally, it is plausible that what we are in a position to perceive or infer from evidence often depends heavily on the society in which we have been raised. The kinds of things of which we are perceptually aware and the kinds of inductive generalizations that we draw are plausibly affected by what we have been taught and by the kinds of awareness and inference that we have experienced in our fellows. Surely, if we had no intercourse with other intelligent beings, our beliefs based on perception and inference would be very different. If the kind of sociality characteristic of learning from testimony is supposed to distinguish learning from testimony from the capacities for inference and perception, then it must be different from any of the kinds of sociality mentioned above. C. A. J. Coady has claimed that what distinguishes the social operations of mind is that they are es2 3 4 5

Cf. Reid (2002), 69. Reid (2002), 69. Putnam (1975). Cf. Burge (1979).

Testimony, Address, and the Second Person

259

sentially “interactive”6, and I think that this is on the right track. It is important to recognize that Reid is here concerned with social operations of mind—with social mental capacities, powers, or abilities—meaning that the sociality he has in mind pertains to the actual exercise of the mental capacities themselves, not merely to the output of these capacities. In this sense, Reid holds that it is the capacity for learning from testimony itself that is irreducibly social, not merely the beliefs that result from the exercise of this capacity. Learning from testimony is irreducibly social in that the very exercise of the capacity is somehow interactive or co-operative and so couldn’t exist without intercourse with other intelligent beings. Reid doesn’t have much to say about the precise sense in which the exercise of social capacities is interactive or cooperative, but he does make an intriguing reference to issues concerning what we might call the second person. In all languages, the second person of verbs, the pronoun of the second person, and the vocative case in nouns, are appropriated to the expression of social operations of mind, and could never have had place in language but for this purpose: Nor is it a good argument against this observation, that, by a rhetorical figure, we sometimes address persons that are absent, or even inanimated beings, in the second person. For it ought to be remembered, that all figurative ways of using words or phrases suppose a natural and literal meaning of them.7

Reid here claims that, when expressed in language, the social operations of mind often utilize second-personal grammatical devices. For example, the speech acts of commanding, promising, and telling are often performed by using grammatical constructions involving the second person pronoun—“I command you to fall in line,” “I promise you that I will pick up the kids,” “I tell you: the weather here has sure been hot lately.” Second-personal grammatical devices are markers of address. If the social operations of mind typically utilize these markers of address when they are expressed in language, then this might be reason to believe that the operation of these capacities itself requires an addresser and an addressee. Promises, commands, and testimony all must be addressed to an audience, and they all call for their audience to recognize them as being so addressed. The sense in which the social operations of mind are interactive or cooperative might thus have to do with 6 7

Coady (2004), 192. Reid (2002), 70.

260

Benjamin McMyler

the way in which they necessarily require not only two individuals interacting but two individuals interacting in a particular way, as addresser to addressee.8 In this paper I argue that we can make some headway in understanding the distinctively interpersonal character of learning from testimony by considering the epistemic significance of address, by considering the way in which a speaker’s addressing a hearer with her testimony makes a genuine difference with respect to the epistemic credentials of the hearer’s belief. This is something that is denied by nearly all epistemologists working on testimony. The overwhelming consensus amongst epistemologists is that the communicative intentions involved in a speaker’s addressing her testimony to a hearer make no difference to the epistemic credentials of the hearer’s belief.9 Hence, if we divide a hearer’s testimonial beliefs into beliefs based on testimony that is addressed to the hearer—call these address-based beliefs—and beliefs based on testimony that is not addressed to the hearer—call these non-address-based beliefs—there is no epistemological difference between these two classes of belief. From an epistemological perspective, address-based and non-address-based beliefs are of a piece. I will argue that this overwhelming epistemological consensus is mistaken. There is a genuine epistemological distinction between address-based and non-address-based beliefs. More specifically, I will argue that a hearer’s epistemic responsibilities with respect to addressbased beliefs are very different from a hearer’s epistemic responsibilities with respect to non-address-based beliefs. A hearer can discharge her epistemic responsibilities with respect to her address-based beliefs by passing the epistemic buck or deferring justificatory responsibility for her beliefs back to the testimonial speaker, but she cannot do this with respect to her non-address-based beliefs. This suggests that when a speaker addresses a hearer with her testimony, the audience acquires a distinctive kind of reason for belief, a reason for belief that depends essentially on the speaker and hearer relating to one another as addresser to addressee. Stephen Darwall has recently argued that there is a distinctive class of reasons for action that depend in this way on individuals relating to one another as addresser to addressee. He calls these reasons for action second-personal reasons. However, he 8 9

For a more detailed account of the historical significance of Reid’s account of testimony, see McMyler (2011a), Chapter 1. A notable exception is Hinchman (2005).

Testimony, Address, and the Second Person

261

explicitly denies that there are any genuinely second-personal reasons for belief.10 I will argue that a speaker’s testimony provides an addressee with a reason for belief that genuinely parallels the kind of reason for action provided by one paradigmatic kind of second-personal reason for action, the reason for action provided by a speaker’s command. If this is right, then the philosophical significance of address and of the second-person actually spans the divide between theoretical and practical reason. The interpersonal relations involved in addressing authoritative reasons to a hearer play a genuine role in both theoretical and practical rationality.

II. Strong and Weak Epistemic Insignificance Regardless of one’s particular views concerning the nature of knowledge and justification, it seems plainly true that both address-based and non-address-based beliefs can be justified and can amount to knowledge. Whether a hearer’s belief is address-based or not, it can still be the result of a reliable belief-forming process, based on sufficient evidence, responsibly formed, sensitive, safe, apt, etc. Depending on the circumstances, both address-based and non-address-based beliefs can be perfectly well grounded, meaning that if there is a genuine epistemological distinction between address-based and non-address-based beliefs, this cannot be a distinction with respect to whether the addressed or non-addressed testimony provides a hearer with good grounds for belief. The recognition that both addressed and non-addressed testimony can provide perfectly good grounds for belief may be part of what has motivated so many philosophers to deny that address is of any proper epistemological significance. However, to say that addressed and non-addressed testimony can both provide perfectly good grounds for belief is not yet to say that those grounds are the same. Both my perception of a deer walking through my backyard and my perception of a footprint left in the snow can provide me with perfectly good grounds for believing that there is a deer in the vicinity, but there is arguably a genuine epistemological distinction between the grounds provided. In the one case my grounds are perceptual, while in the other they are inferential. The fact that both address-based and non-addressbased beliefs can be well grounded thus doesn’t foreclose the possibility 10 Darwall (2006).

262

Benjamin McMyler

that there is a genuine epistemological distinction between addressbased and non-address-based beliefs. Though most epistemologists share the conviction that there isn’t a genuine epistemological distinction between addressed-based and nonaddress-based beliefs, the denial of the epistemic significance of address can take stronger and weaker forms. Some epistemologists accept the strong claim that whether an instance of testimony is addressed to another person at all makes no difference to the epistemic credentials of a hearer’s beliefs based on the testimony. We might call this claim Strong Epistemic Insignificance. According to Strong Epistemic Insignificance, if we divide a hearer’s testimonial beliefs into beliefs based on testimony that is addressed to someone and beliefs based on testimony that isn’t addressed to anyone at all, then there is no distinction between the epistemic credentials of these two types of belief. Importantly, Strong Epistemic Insignificance requires a particular conception of the kind of speech act relevant to the epistemology of testimony. If Strong Epistemic Insignificance is true, then we must be able to identify instances of testimony that are not in fact addressed to anyone. A speaker must be able to perform the speech act paradigmatically relevant to the epistemology of testimony absent the communicative intentions, whatever exactly they are, typically involved in addressing an utterance to a hearer. Along these lines, Jennifer Lackey has argued that a statement that is not intended to communicate information to anyone can nevertheless amount to testimony as long as it is intended to express “communicable content.”11 Lackey thus holds that the kind of non-addressed statements found in private soliloquies and entries in a private diary genuinely amount to testimony, to what she calls “hearer testimony” (or “h-testimony”).12 Hearer testimony amounts to an act of communication that can serve to justify a hearer’s belief in the content of the testimony despite the fact that the testimony is not addressed to anyone and hence is not intended to communicate information. A speaker may h-testify even if she positively intends for her statement or thought never to be taken as conveying information to anyone. For example, suppose Lucy’s private journal is found by the police. Suppose further 11 Lackey (2008). 12 Lackey argues for a view of the nature of testimony according to which testimony includes both “speaker testimony” and “hearer testimony”. Speaker testimony requires that a speaker intend to convey information, but hearer testimony does not; Lackey (2008), 35 f.

Testimony, Address, and the Second Person

263

that she wrote that her husband had committed the crime for which he is suspected, though she intended such thoughts only for herself—in fact, she even encrypted the entries in what she (mistakenly) thought was an unbreakable code and, thus, in no way offered such thoughts as conveying information regarding her husband’s guilt. Nonetheless, according to the current proposal, Lucy’s journal will qualify as h-testimony since it is taken as conveying information by the police.13

On this kind of view, oral and written statements that are made in private and are not intended to communicate information to anyone nevertheless put a hearer in a position to acquire properly testimonial belief. Many epistemologists object to the conception of the speech act of testimony required by Strong Epistemic Insignificance while still downplaying the epistemic significance of address. For example, Elizabeth Fricker has recently claimed that testimony, tellings, and assertions all must have an intended audience; they all must be addressed to someone.14 Without this they cannot amount to instances of the relevant speech acts. Nevertheless, she still holds that whether the speaker’s testimony is addressed to the particular hearer in question makes no difference to the epistemic credentials of the hearer’s testimonial belief. Though Fricker accepts the idea that address is necessary for a speech act to so much as amount to an instance of testimony, she nevertheless holds that there is no salient epistemological distinction between addressees of a genuine instance of testimony and non-addressees. [A]lthough a telling must have an intended audience, it is not only the intended audience who can avail herself of the reason to believe made available by the act of telling. An eavesdropper can witness the same act and, providing she has an epistemic basis properly to trust the teller, may equally acquire knowledge at second hand, as the intended audience does. Moreover, her grounds or basis of entitlement is essentially the same. Unlike some, I do not think the audience/eavesdropper distinction is directly epistemically relevant. There is no special entitlement to believe (nor requirement, epistemic or moral) made available to the intended audience of the telling, which is not available to anyone else. The entitlement to believe made available by a telling, or indeed any assertion, is what economists call a public good: it is in the nature of my speech act, that I cannot, 13 Lackey (2008), 31 f. 14 Fricker claims that these speech acts must have “an intended audience.” Arguably, a speech act’s having an intended audience is insufficient for the speech act’s being addressed to the audience. I might intend you to be an (overhearing) audience of my soliloquy without thereby addressing the soliloquy to you. Nevertheless, I am here reading Fricker as taking “having an intended audience” and “having an addressee” to be coextensive; cf. Fricker (2006).

264

Benjamin McMyler

even should I want to, make available an entitlement to believe to you, without simultaneously making it available to anyone who hears and understands me doing so. Would it were so simple to restrict one’s sharing of information! No, if I want to tell you that P, and I don’t want anyone else in earshot to know (not just believe, but know!), it is not enough that I address you, and not the others. I have to whisper, or someotherwise make sure they do not hear what I say to you! 15

On Fricker’s view, even though a statement must be addressed to a speaker in order to so much as amount to an instance of testimony, the communicative intentions embodied in the act of addressing an utterance to a hearer do not grant the addressee any particular kind of epistemic entitlement that can’t be acquired by hearers other than the addressee. Addressees and mere eavesdroppers are thus in an equivalent epistemic position with respect to the epistemic entitlement made available by a speaker’s testimony. We might call this claim Weak Epistemic Insignificance. According to Weak Epistemic Insignificance, if we divide a hearer’s testimonial beliefs into beliefs that are based on testimony that is addressed to the hearer and beliefs that are based on testimony that is not addressed to the hearer (but that is nevertheless addressed to someone else), then there is no distinction between the epistemic credentials of these two types of belief. I will argue that even Weak Epistemic Insignificance is mistaken. There is a genuine epistemological distinction between addressees and mere eavesdroppers, and thus the epistemic credentials of addressbased and non-address-based beliefs are different. In making this argument I will be expanding on a point provisionally suggested by Richard Moran. Moran writes: When all goes well in testimony, a speaker gives his audience a reason to believe something, but unlike other ways of influencing the beliefs of others, in this case the reason the audience is provided is seen by both parties as dependent on the speaker’s making himself accountable, conferring a right of complaint on his audience should his claim be false.16

Moran then goes on to note that this “right of complaint” is acquired only by addressees of a speaker’s testimony and not by mere overhearers. If one person gives his word on something to another, whether as promise or assertion, someone overhearing this may derive a sufficient reason to believe, say, that the speaker will in fact do what he promised or that what he 15 Fricker (2006), 598. 16 Moran (2005), 21, my emphasis.

Testimony, Address, and the Second Person

265

asserted is true. And the overhearer improves his epistemic situation in this way without entering into the altered normative relationship of the two parties involved in giving and accepting of words. He has not himself been told anything, much less promised anything, and no right of complaint has been conferred on him.17

Moran is concerned to insist that mere overhearers can acquire a reason for belief from a speaker’s testimony only given the fact that the testimony has been addressed to someone. Like Fricker, he wants to insist that addressive relations are ineliminably involved in constituting an utterance as an instance of testimony. However, unlike Fricker, he seems to accept that there is a salient difference between addressees and mere over-hearers. While addressees acquire a right of complaint if a speaker’s testimony turns out to be false, mere overhearers do not. Unfortunately, Moran doesn’t have much to say about what exactly this right of complaint involves. A speaker’s testimony can violate all sorts of moral and prudential norms—even if the testimony is true, it can be rude, imprudent, inappropriate, etc.—and overhearers may have just as much right to complain about this as addressees. Moreover, as Lackey has argued18, it isn’t clear how the fact that a speaker’s testimony can violate such moral, prudential, or conventional norms makes it the case that her testimony provides a hearer with a reason for belief that differs from other ways of influencing the beliefs of others. It isn’t clear that such moral, prudential, or conventional norms are genuinely epistemically significant.19 I will argue (1) that there is in fact a distinctive right of complaint that is only acquired by addressees and (2) that this right of complaint is genuinely epistemically significant.

17 Moran (2005), 22. 18 Cf. Lackey (2008). 19 As I read Moran, he does think that the “right of complaint” distinctively generated by a speaker’s testimony is genuinely epistemically significant. Since he doesn’t have much to say about what this right of complaint involves, however, it can be difficult to see how this is so. As I argue in McMyler (2011a), this leaves Moran’s Assurance View insufficiently supported and opens him to objections like those raised by Lackey.

266

Benjamin McMyler

III. Passing the Epistemic Buck Imagine a police detective using a wiretap to listen in on a telephone conversation between two criminals. The detective hears criminal A tell criminal B that an extremely large shipment of narcotics is arriving on Wednesday. On the basis of A’s testimony, both criminal B and the detective form the belief that an extremely large shipment of narcotics is arriving on Wednesday. B’s belief is based on being addressed with A’s testimony and so is an address-based belief. The detective’s belief is not based on being addressed with A’s testimony—the detective cannot be considered an addressee of A’s testimony in even the widest possible sense—and so the detective’s belief is a non-address-based belief. Our question, then, is this: Is there a salient distinction between the epistemic credentials of B’s address-based belief and the epistemic credentials of the detective’s non-address-based belief ? Provided that both B and the detective have sufficient reason to judge that A is reliable in this particular case, both B and the detective appear to be in a perfectly adequate position to acquire justification for believing that an extremely large shipment of narcotics is arriving on Wednesday on the basis of A’s testimony. Moreover, if it is true that an extremely large shipment of narcotics is arriving on Wednesday, then both B and the detective appear to be in a position to acquire knowledge of this fact. Hence, if there is a salient epistemological distinction between B’s address-based belief and the detective’s non-address-based belief, this cannot be a matter of whether the relevant beliefs are justified or amount to knowledge. Nevertheless, there may still be a difference in the way in which their respective beliefs are justified, and this can be illustrated by considering the way in which their epistemic responsibilities for their respective beliefs diverge. Imagine that after forming their beliefs on the basis of A’s testimony, both B and the detective are challenged by third parties to defend their beliefs. B is challenged by a fellow criminal who states that no one has ever before secured a shipment of narcotics of this size, and the detective is challenged by a fellow detective who states the same. Faced with such a challenge, it would be natural for B to respond by saying, “Don’t ask me. A told me that this shipment is arriving on Wednesday.” It would not be natural for the detective to respond in the same way. The detective has not been told that the shipment is arriving on Wednesday. She has merely overheard A tell B that the shipment is arriving on Wednesday. Hence, it would be natural for the detective to respond to the chal-

Testimony, Address, and the Second Person

267

lenge by saying, “Well, that is what A said” or “That is what A told B,” but the detective should not respond by saying that she herself has been told that the shipment is arriving on Wednesday. To count as having been told that p by a speaker, a hearer must be an addressee of the speaker’s telling. If a hearer is not an addressee of a speaker’s telling, then the hearer merely counts as having heard the speaker say (or tell someone else) that p. Importantly, however, being told by a speaker that p puts one in a distinctive epistemic position. Note that the force of criminal B’s response to the third-party’s challenge is quite different from the force of the detective’s response to the same challenge. In responding to the challenge by saying, “Don’t ask me. A told me that the shipment is arriving on Wednesday,” B is not merely informing the challenger of her reason for believing that the shipment is arriving on Wednesday. Instead, B is actually deferring responsibility for meeting this challenge to her belief back to A. She is passing the epistemic buck. Moreover, in doing so, B is not simply pointing out that there happens to be another person, criminal A, who may have the epistemic resources to meet this challenge. Rather, she is actively holding A responsible for meeting it. B is asserting that she is rationally entitled to maintain her belief despite her inability to meet the challenge herself due to the fact that, in addressing her with a telling, A has assumed justificatory responsibility for her belief. The detective’s response to the challenge does not have the force of passing the epistemic buck back to A. In responding to the challenge by saying, “Well, that is what A said” or “That is what A told B”, the detective is doing little more than re-iterating her reason for belief. The detective is in a position to appeal to what A has said as evidence that counts in favor of her belief, but since she hasn’t been told by A that the shipment is arriving on Wednesday, she is not entitled to defer responsibility for meeting this epistemic challenge to her belief back to A. Since A didn’t tell the detective anything, A has not assumed responsibility for meeting epistemic challenges to the detective’s belief, and hence justificatory responsibility for the detective’s belief lies solely with the detective.20 The same can be said for any case in which a hearer 20 One might object to my reading of the detective case by arguing that either (1) criminal B is not genuinely entitled to pass the epistemic buck in the way that I have described or (2) the detective is entitled to pass the epistemic buck in the way that I have described. As I construe epistemic buck-passing, passing the

268

Benjamin McMyler

merely overhears a speaker’s testimony. If I merely overhear a speaker outside my closed office door tell someone else that p, then even though this may give me a very good reason to believe that p, I have not been told by the speaker that p, and hence I am not entitled to defer justificatory responsibility for meeting epistemic challenges to my belief that p back to the testimonial speaker.21 epistemic buck involves deferring justificatory responsibility for one’s belief to another person. Buck-passing is thus very different from, for example, pointing to the fact that one’s reason for belief is another person’s statement, pointing to the fact that another person likely possess the resources to meet an epistemic challenge to one’s belief, or even confronting another person with an epistemic challenge in the hopes of eliciting a response that might give one additional reasons for belief. These, I contend, are all insufficient for holding the other person responsible for the justification of one’s own belief. Passing the epistemic buck is thus very different from querying after another person’s reasons for belief, and while the detective is certainly in a position to query after criminal A’s reasons for belief, I think it is clear that the detective is not in a position to pass the epistemic buck as here construed. I therefore think that (2) is implausible. However, one might very well question my construal of epistemic buck-passing. Perhaps buck-passing doesn’t amount to anything more than querying after another’s reasons for belief. If this is the case, then (1) is true—criminal B is not entitled to pass the epistemic buck in the way that I have described—and moreover, what B is entitled to do, namely query after A’s reasons for belief, may be no different than what the detective is entitled to do. I think that this line of thought is the more serious threat to my argument, but I think that this too is implausible. As I argue in McMyler (2007) and McMyler (2011a), there appears to be something that I am entitled to do when I take someone’s word for something or take something on the authority of another that I am not entitled to do when I treat what someone says as ordinary evidence from which I draw my own conclusion. When I take someone’s word for something, and when I am challenged by a third-party to defend my belief, I am entitled to defer responsibility for meeting certain epistemic challenges to my belief back to the testimonial speaker. I am rationally entitled to maintain my belief despite my own inability to meet the challenge by instead deferring to the testimonial speaker. I am not so entitled when I treat what the speaker says as ordinary evidence from which I draw my own conclusion, and as I have argued here, I am not so entitled when I merely overhear the speaker’s testimony. 21 I take it that the distinction that I have illustrated between the way in which criminal B and the detective are in a position to respond to epistemic challenges is a result of the fact that criminal B has been addressed by criminal A with a telling while the detective has not. The distinction is not a result of other differences between B and the detective, for instance, that the detective’s confronting A with the third-party’s challenge might be prudentially undesirable insofar as it might have the effect of blowing the detective’s cover. Admittedly, the case of the detective involves characters that are related to one another in pragmatically

Testimony, Address, and the Second Person

269

This suggests that there is a genuine epistemological distinction between address-based and non-address-based beliefs. Moran holds that addressees of a speaker’s telling acquire a “right of complaint” that mere overhearers do not.22 We are now in a position to see that addressees of a speaker’s testimony acquire the more specific right to pass the epistemic buck, while mere overhearers do not. When a hearer is addressed with a speaker’s testimony that p, the hearer counts as being told by a speaker that p, and being told by a speaker that p puts the hearer in a distinctive epistemic position. If a hearer is told by a speaker that p, and if the hearer accepts the speaker’s telling as such, then the hearer is entitled to pass the epistemic buck, to defer justificatory responsibility for meeting epistemic challenges to her belief back to the testimonial speaker. If a hearer has not been addressed with a speaker’s testimony, if she does not count as having been told by the speaker that p, then even though her overhearing of the speaker’s testimony may give her very good reason to believe that p, it does not put her in a position to pass the epistemic buck. The speaker has not assumed responsibility for meeting epistemic challenges to the mere overhearer’s belief, and hence the mere overhearer is not rationally entitled to maintain her belief in the face of epistemic challenges by deferring to the testimonial speaker. Importantly, a hearer’s acquiring the entitlement to pass the epistemic buck requires that she accept the speaker’s testimony in a particular idiosyncratic ways. However, I don’t think that the distinction that I have illustrated is generated by these particular idiosyncrasies. I use this case because it involves a clear distinction between an addressee and a mere overhearer. As I argue below, there are many cases of what might be described as overhearing where the overhearer in question can be construed as a kind of indirect addressee. There is no danger of this in the case of the detective. There is no sense in which the detective can be construed as an addressee of A’s testimony in even the widest possible sense. The case thus involves a clear distinction between address-based and non-address-based belief. I think the distinction that I have illustrated between the address-based and non-address-based beliefs in this case will be present in any case in which there is a clear distinction between address-based and non-address-based beliefs. In the case of my merely overhearing a conversation outside my closed office door, for example, there may be no respect in which my deferring a challenge to the testimonial speaker is prudentially undesirable, but it still appears to be the case that I am not entitled to pass the epistemic buck in the way that I have described. We can thus abstract away from the particular pragmatic idiosyncrasies present in the case of the detective. 22 Moran (2005).

270

Benjamin McMyler

way. The hearer must “take the speaker’s word for p” or “take p on the authority of the speaker”, meaning she must accept the speaker’s assumption of epistemic responsibility for her belief by being appropriately disposed to defer epistemic challenges back to the speaker. If the hearer does not accept the speaker’s testimony in this way—if, for example, the hearer treats the speaker’s telling as nothing more than ordinary inductive evidence—then even though the hearer has been addressed with a telling, she is not entitled to pass the epistemic buck. In order to acquire an entitlement to pass the epistemic buck, a hearer’s belief must be based on a speaker’s addressing the hearer with a telling in a way that involves the audience’s accepting the speaker’s telling as the kind of assumption of epistemic responsibility that it is. If the hearer comes to believe what the speaker says by treating the speaker’s telling as ordinary inductive evidence, then even though the hearer’s belief may be justified and may amount to knowledge, her belief is based on the wrong kind of reason to generate an entitlement to pass the epistemic buck. The responsibilities involved in epistemic buck-passing are genuinely epistemic responsibilities. They are responsibilities that pertain to the way in which cognitive agents are distinctively answerable for their beliefs. Belief, I take it, is a psychological attitude that involves a subject’s committing herself to or settling on an answer to a particular question. To believe that p is to commit oneself to or settle on a positive answer to the question whether p. As a result, a believer is answerable for being so committed. She is open to criticism that bears on the content of her commitment, criticism that bears on the question whether p.23 Plausibly, one aspect of the way in which believers are thus answerable for their beliefs concerns their epistemic conduct in the face of reasonable challenges to their beliefs, where reasonable challenges to their beliefs involve the presentation of evidence that counts against their beliefs. Typically, when confronted with such a challenge, a rational epistemic agent ought to either find some way to meet the challenge—some basis upon which to rationally discount the evidence presented—or else give up her belief. When it comes to beliefs that are based on being told something by a speaker, however, an epistemic agent is entitled to maintain her belief without meeting the challenge herself by instead passing the epistemic buck back to the speaker. If an audience comes to believe that p on the basis of a speaker’s testimony, and if a third party challenges the au23 For an account of responsibility for belief along these lines, see Hieronymi (2008).

Testimony, Address, and the Second Person

271

dience’s belief by producing evidence that counts against p, the audience is entitled to defer responsibility for meeting the challenge back to the original speaker, whereupon the original speaker is responsible for meeting the challenge.24 An audience isn’t entitled to defer just any epistemic challenge back to the speaker. If the challenge involves the production of evidence that tells against the trustworthiness of the testimonial speaker—say, that the speaker is an inveterate liar—rather than against the content of the speaker’s testimony, then the audience is not entitled to defer the challenge back to the speaker.25 Nevertheless, the audience does appear to be entitled to defer challenges like that involved in the case of the detective, challenges that involve the production of evidence that tells against the content of the speaker’s testimony. This entitlement to pass the epistemic buck is a distinguishing feature of a particular class of beliefs. It doesn’t apply to beliefs based on perception, memory, or inference. If I perceive that p, remember that p, or infer that p, then I am not entitled to pass the epistemic buck should the content of my belief be challenged. I have argued elsewhere that the class of testimonial beliefs ought to be identified with the class of beliefs that involve an entitlement to pass the epistemic buck.26 Though we can plausibly acquire knowledge and justified belief from treating a speaker’s testimony as, for example, ordinary inductive evidence, insofar as such belief does not involve an entitlement to defer challenges, such belief ought to be excluded from the category of properly testimonial belief. Such belief does not involve “taking a speaker’s word for something” or “taking something on the authority of a speaker,” and as such it looks more akin to the kind of belief acquired from ordinary instruments. I have also argued elsewhere that the standard reductionist and antireductionist accounts of the epistemology of testimony available in the literature are in a poor position to explain the phenomenon of epistemic buck-passing.27 Reductionists about testimony hold that a hearer’s testimonial belief that p is justified by the strength of an inference from the speaker’s testimony that p, through independently available considerations of the speaker’s trustworthiness, to the conclusion that p. Reductionists thus construe testimony as kind of ordinary inductive evidence. 24 25 26 27

Cf. Cf. Cf. Cf.

Brandom (1983) and (1994), Goldberg (2006). McMyler (2007) and (2011a). McMyler (2011a). ibid.

272

Benjamin McMyler

But if a subject infers that p from ordinary inductive evidence, she is not thereby entitled to pass the epistemic buck. Reductionism is thus straightforwardly inconsistent with the phenomenon of epistemic buck-passing. Something similar is true of many standard anti-reductionist accounts of testimony. Anti-reductionists about testimony reject the idea that the justification relevant to testimonial belief is inferential. Instead, they typically model their account of the epistemology of testimony on the epistemology of perception. Anti-reductionists about testimony thus typically hold that comprehension of the force and content of a speaker’s testimony provides a hearer with a prima facie reason for belief analogous to the kind of prima facie reason for belief provided by perceptual representation. However, a subject clearly isn’t entitled to pass the epistemic buck with respect to her perceptual beliefs, and so standard anti-reductionist accounts are in a poor position to explain how testimonial belief differs from perceptual belief in this important respect. I won’t try to defend all of this here. Instead, I simply want to argue that the case of the detective shows that the entitlement to pass the epistemic buck only accrues to address-based beliefs. The detective is certainly in a position to treat A’s testimony as a good reason for belief, but given the fact that she has not been addressed by A she is not entitled to pass the epistemic buck back to A should her non-address-based belief be challenged. Addressees and mere overhearers are thus in a different epistemic position with respect to the content of their beliefs acquired from a speaker’s testimony, and so Weak Epistemic Insignificance is mistaken. The very same considerations that tell against Weak Epistemic Insignificance also tell against Strong Epistemic Insignificance. Whereas Weak Epistemic Insignificance denies that there is a genuine epistemological distinction between beliefs based on testimony that is addressed to a particular hearer and beliefs based on testimony that is not addressed to the hearer, Strong Epistemic Insignificance denies that there is a genuine distinction between beliefs based on testimony that is addressed to a hearer and beliefs based on testimony that is not addressed at all (assuming for the sake of argument that unaddressed statements can actually amount to instances of testimony). However, if there is a genuine epistemological distinction between the beliefs of criminal B and of the detective, then there must also be a genuine epistemological distinction between the beliefs of criminal B and the beliefs of someone overhearing a private soliloquy. If the detective isn’t in the position to defer chal-

Testimony, Address, and the Second Person

273

lenges to his perfectly well justified beliefs based on criminal A’s testimony, then surely the overhearer of the private soliloquy isn’t in a position to defer challenges either. If an utterance is not addressed to anyone, then an audience of the utterance must be in the position of a mere overhearer and so cannot acquire an epistemic right of deferral. There is thus a genuine epistemological distinction between beliefs that are based on testimony that is addressed to the believer and beliefs that are based on testimony that is not addressed at all, and so Strong Epistemic Insignificance is mistaken.28

IV. Addressees and Overhearers I have offered an example of what I take to be a clear case of the distinction between addressees and mere overhearers. Criminal B is clearly an addressee of criminal A’s testimony, and given that, unbeknownst to the criminals, the detective is listening in on their conversation over a wiretap, the detective clearly cannot be considered an addressee of A’s testimony in any relevant sense. I have thus referred to the detective as a mere overhearer. Importantly, however, not all overhearers are mere overhearers. In many cases the distinction between addressees and overhearers is difficult to discern, and as a result conversational participants that might legitimately be referred to as overhearers are nevertheless sometimes in a position to acquire an entitlement to pass the epistemic buck. Imagine that as students are filing out of a lecture hall after class, student A asks the professor about the due date for the final paper, and the professor tells A that the due date will be pushed back a week from what is indicated on the syllabus. While leaving the room, student B overhears what the professor says and on this basis forms the belief that the due date will be pushed back a week. B later relates this fact to another student in the class who challenges her by pointing out that pushing the due date back a week will give the professor very little time to 28 Is there an epistemic distinction between beliefs based on testimony that is not addressed at all (Lackey’s hearer testimony) and beliefs based on testimony that is addressed to someone other than the believer in question? It doesn’t seem to me that there is. Here I disagree with Moran (2005). It just isn’t clear to me that, from a hearer’s perspective, the kind of reason for belief provided by a speaker’s private soliloquy is relevantly different from the kind of reason for belief provided by a speaker’s addressed testimony overheard on a wiretap.

274

Benjamin McMyler

mark the papers before final grades are due. In this case, despite the fact that she was not a direct addressee of the speaker’s testimony, B still appears to be in a position to pass the epistemic buck back to the professor. B can fulfill her epistemic responsibilities with respect to the proffered challenge by deferring it back to the testimonial speaker. Student B can legitimately be described as an overhearer of the professor’s testimony, and nevertheless she appears to be entitled to pass the epistemic buck. In this respect, she is in a different epistemic position from the detective listening in over a wiretap. She is not a mere overhearer. In fact, the context in which the professor’s testimony is offered is one in which the professor might be conventionally understood to be addressing everyone within earshot. Clearly, there is a direct addressee of the professor’s remark—student A, who might even be picked out by use of a vocative, as in “Well Sally, the due date for the paper will actually be pushed back a week”—but given the context in which the remark occurs, all of the remaining students in the room can legitimately be taken to be indirect addressees of the professor’s remark. In this sense, while student B can be referred to as an overhearer in the sense that she is not the direct addressee of the speaker’s testimony, she can nevertheless also be understood to be an indirect addressee. This clearly distinguishes student B from the detective. Given that the detective is listening in on criminal A’s testimony over a wiretap, there is no sense in which the detective can be understood to be an indirect addressee. The detective cannot be understood to be an addressee of criminal A’s testimony in even the widest possible sense. In contrast, student B can be understood to be an addressee of the professor’s testimony in a wide, indirect sense, and this explains why, unlike the detective, she is in a position to pass the epistemic buck. The epistemic distinction that I have identified between addressbased and non-addressed based belief is thus one that pertains to a wide sense of address. It is a distinction between the beliefs of addressees in a wide sense that includes student B and the beliefs of mere overhearers like the detective. In order to mark this distinction between addressees and mere overhearers, I propose the following provisional account of address: (ADD) In performing communicative act C, person A addresses person B just in case A intends B to believe that A intends B to understand C.

There are two things to note about (ADD). First, address is here characterized as a feature of communicative acts, but I intend the category of

Testimony, Address, and the Second Person

275

communicative acts to be broader than the category of illocutionary acts. One can address an audience with a wink, a nod, or a gesture, none of which involve performing an illocutionary act.29 (ADD) is thus a very broad account of address designed to capture the way in which communicative action can be directed from one person to another.30 Second, according to (ADD), addressing an audience with a communicative act involves performing the act with a particular intention. Importantly, however, the content of this intention is not that the audience understands the communicative act. Intending an audience to understand the communicative act is neither sufficient nor necessary for addressing the audience. On the one hand, a speaker can clearly intend an audience to understand the force and content of her utterance without addressing the audience. Imagine someone discussing with a friend what she would like for her birthday while knowing full well that her spouse will overhear her. The speaker might explicitly intend her utterance to be overheard and understood, even while she does not intend to be telling her spouse what she wants, perhaps thinking that such openness would be inappropriate or undesirable. (ADD) is in a position to account for this. Though the speaker intends her spouse to understand her communicative act, she does not intend her spouse to believe that she intends her spouse to understand her communicative act. 31 On the other hand, a speaker can succeed in addressing an audience without actually intending the audience to understand the force and content of her utterance. Consider the following case presented, for rather different purposes, by John Searle: Suppose that I am an American soldier in the Second World War and that I am captured by Italian troops. And suppose also that I wish to get these troops to believe that I am a German soldier in order to get them to release 29 I have therefore formulated (ADD) in terms of persons rather than speakers and hearers. Addressors needn’t be speakers, and addressees needn’t be hearers. As my particular concern is with the epistemic significance of address for the beliefs of hearers, I will generally leave this complication aside in what follows. 30 Can the notion of address apply to things that are not communicative acts, such as thoughts or utterances said under one’s breath? Such things can make use of the second-person pronoun—for example, thinking “Screw you” or saying it under one’s breath. 31 This case might be modified so that the speaker does have something like this intention toward her spouse. However, to the extent that this is so, the spouse looks more and more like an indirect addressee, ceremony notwithstanding.

276

Benjamin McMyler

me. What I would like to do is to tell them in German or Italian that I am a German soldier. But let us suppose I don’t know enough German or Italian to do that. So I, as it were, attempt to put on a show of telling them that I am a German soldier by reciting those few bits of German I know, trusting that they don’t know enough German to see through my plan. Let us suppose I know only one line of German which I remember from a poem I had to memorize in a high school German course. Therefore, I, a captured American, address my Italian captors with the following sentence: Kennst du das Land wo die Zitronen blhen? 32

Here the American soldier clearly addresses his Italian captors with his utterance. Equally clearly, however, he does not intend his captors to understand the force and content of his utterance. He positively intends that they will not understand what he says and that they will thereby conclude that he is a German soldier. Nevertheless, the American soldier does intend that his Italian captors believe that he intends that they understand the force and content of his utterance, and so (ADD) is able to account for this case. According to (ADD), the addressees of a speaker’s testimony are those hearers the speaker intends to believe that she intends them to understand her testimony. Mere overhearers of a speaker’s testimony are those hearers the speaker does not intend to believe that she intends them to understand her testimony. Returning to our wiretapping case, the detective is clearly a mere overhearer. Criminal A does not intend the detective to believe that she intends the detective to understand her testimony. On the other hand, criminal B is clearly an addressee. Criminal A does intend criminal B to believe that she intends criminal B to understand her testimony. (ADD) also helps us to understand what is at issue in the more difficult case of the professor. The professor clearly intends student A to believe that she intends student A to understand her testimony. But given the broader context and the professor’s conventional standing with respect to the rest of the students in the class, it is plausible that the professor should also be understood to intend that all of the students in the class, including student B, believe that she intends them to understand her testimony. On this reading, all of the students in the class can be understood to be addressees of the professor’s testimony in the sense specified by (ADD).

32 Searle (1969), 44.

Testimony, Address, and the Second Person

277

V. Testimony as a Second-Personal Reason for Belief I have argued that the epistemic significance of address lies in the fact that there is an entitlement to pass the epistemic buck that is possessed by addressees of a speaker’s testimony but not by mere overhearers. The fact that address-based belief differs in this respect from non-addressbased belief suggests, I think, that when a speaker’s testimony is addressed to an audience, the speaker’s testimony provides the audience with a distinctive kind of reason for belief. Unlike the reason for belief provided by other kinds of impersonal evidence, a speaker’s testimony provides an audience with a reason for belief that entitles the audience to pass the epistemic buck, a reason for belief that depends essentially on the speaker and the audience relating to one another as addresser to addressee. In The Second-Person Standpoint, Stephen Darwall argues that there is an important class of reasons for action that also depend essentially on individuals relating to one another as addresser to addressee. He calls these reasons for action second-personal reasons. 33 However, in making his case for the existence of this class of distinctively second-personal reasons for action, Darwall explicitly denies that there are any genuinely or irreducibly second-personal reasons for belief. In the remainder of this paper, I will argue that the kind of reason for belief provided by a speaker’s testimony is genuinely second-personal in a way that parallels Darwall’s characterization of second-personal reasons for action.34 Darwall distinguishes between what he calls second-personal and third-personal reasons for action. Both second and third personal reasons for action can be addressed from a speaker to an audience. What distinguishes second-personal reasons is that they derive their justificatory force from the interpersonal relations involved in the speaker’s addressing the reason to the audience.35 We can see how this is so by consid33 Cf. Darwall (2006). 34 Much of the argument in this section follows that in McMyler (2011a), Chapter 5. 35 Importantly, Darwall (2006) doesn’t hold that second-personal reasons must be explicitly addressed from an addresser to an addressee. Second-personal reasons for action can be generated by relations of authority between persons even where nothing is said from one person to another. Hence, in one of Darwall’s favorite recurring examples, I can have a second-personal reason to get off of your foot simply in virtue of my recognizing you as a person. You needn’t tell me to get off of your foot. The fact that I recognize you as a person and

278

Benjamin McMyler

ering Thomas Hobbes’s distinction between command and counsel. 36 Counseling or advising a person to ¦ justifies the person’s ¦-ing by providing the person with the materials to come to her own conclusion about whether to ¦. The reason for action provided by counsel or advice thus gains its justificatory force not from the act of counseling or advising itself but rather from the particular considerations that happen to be presented by the counsel or advice. These considerations exist and tell in favor of ¦-ing completely independently of the speaker’s addressing them to the audience, and hence the reason for action that they provide is, according to Darwall, third-personal. Commands are very different. Commanding a person to ¦ does not involve attempting to provide the person with the materials to come to her own conclusion about whether to ¦. A command doesn’t present materials or considerations that are themselves reasons to ¦. Rather, the command is itself the reason to ¦. To obey the command to ¦ is to ¦ on the basis of the speaker’s having addressed one with the command to ¦ and not on the basis of having come to one’s own conclusion that ¦-ing is the thing to do.37 One must still come to one’s own conclusion about whether to treat the speaker as an authority, but this is different from coming to one’s own conclusion that ¦-ing is the thing to do. If one were to ¦ on the basis of coming to one’s own conclusion that ¦-ing is the thing to do, then one would not be obeying the command—one would be ¦-ing for the wrong kind of reason. The kind of reason for action provided by a speaker’s command is thus, according to Darwall, genuinely second-personal. It is a reason that wouldn’t exist but for the speaker’s addressing the reason to the audience.38 thereby as one standing in interpersonal relations with me is enough to itself generate a reason for me to get off of your foot, a reason that would not exist but for our standing in these interpersonal relations. 36 Cf. Hobbes (1996), 176. This is roughly the distinction in terms of which Darwall introduces his conception of second-personal reasons in The Second-Person Standpoint (2006), 5. In this sense, it seems to function for Darwall as a paradigmatic case of the distinction between second and third-personal reasons. 37 As Hart (1990) puts it, the reason for action provided by a speaker’s command is both content-independent and peremptory. It is content independent in that the reason derives from the will of the speaker independently of the specific character of the action to be performed, and it is peremptory in that it is intended to preclude or cut off the agent’s acting on her own private assessment of the merits of the action. 38 The distinction between second and third-personal reasons for action may often be quite fine. On many occasions there may be little if any difference between

Testimony, Address, and the Second Person

279

As I’ve said, Darwall denies that there are any genuinely second-personal reasons for belief. He thinks that there is something essentially practical about second-personal reasons. I think that this is a mistake, and we can begin to see how this is so by noting that there is an epistemic analogue to the practical distinction between command and counsel. Indeed, this is just what we should expect if there is in fact a genuine distinction in the epistemic realm between second and thirdpersonal reasons. The distinction between the kind of reason for action provided by counsel and the kind of reason for action provided by command is very much like the distinction between the kind of reason for belief provided by arguing and the kind of reason for belief provided by telling (by testifying). In fact, the practical distinction between counsel and command can itself be construed (and is often construed by Darwall himself) as a distinction between cases in which a speaker argues for a given course of action and cases in which a speaker comes out and tells an audience to perform a given course of action. As we’ve seen, in counseling or advising a course of action, a speaker is presenting an audience with considerations that tell in favor of the course of action that the audience is itself charged with coming to its own conclusion about. Similarly, in arguing that p a speaker is presenting the audience with considerations that tell in favor of p that the audience is itself charged with coming to its own conclusion about. The reason for belief provided by a speaker’s argument thus gains its justificatory force not from the act of arguing itself but rather from the particular considerations that happen to be presented by the argument. These considerations exist and tell in favor of belief in the conclusion of the argument completely independently of the speaker’s act of addressing them to the audience. After all, the whole point of providing argument is to help the audience to see things for itself, to come to believe the conclusion of the argument on the basis of the audience’s own evaluation and assessment of the strength of the argument. The kind of reason for belief provided by argument is therefore third-personal. telling a person to ¦ and telling a person she should ¦ (telling a person that, all things considered, the reasons count in favor of her ¦-ing). In this sense, much that goes under the label of advice, especially between friends and family, is often quite second-personal. Still, there is a genuine distinction between clearly third-personal cases of advice and clearly second-personal cases of command, and this is all that we need in order to get the distinction between second and third-personal reasons on the table.

280

Benjamin McMyler

Telling (or testifying) is very different. Like commanding a person to ¦, telling an audience that p does not involve attempting to provide the

audience with materials to come to her own conclusion about p. Unlike arguing, telling doesn’t present independent materials or considerations that are themselves reasons to believe that p. Rather, the telling is itself the reason to believe that p. To believe that p on the basis of the speaker’s telling is to believe that p on the basis of the speaker’s having addressed one with the testimony that p and not on the basis of having come to one’s own conclusion about p.39 One must still come to one’s own conclusion about whether to treat the speaker as a theoretical authority, but this is different from coming to one’s own conclusion about p. If one were to believe that p on the basis of coming to one’s own conclusion about p, perhaps by treating the speaker’s testimony as good inductive evidence for p, then one would not be in the position of believing the speaker, of accepting the speaker’s testimony—one would be believing that p for the wrong kind of reason. The kind of reason for belief provided by a speaker’s testimony thus appears to be genuinely second-personal. It appears to be a reason for belief that would not exist but for the speaker’s addressing the reason to the audience.40 The analogy here between commanding and telling is further supported by the fact that, as in the case of testimony, there is a distinction between the kind of reason for action available to addressees and mere overhearers of a speaker’s command. I’ve argued that the distinctive kind of reason for belief generated by a speaker’s testimony can only be acquired by addressees of a speaker’s testimony. Mere overhearers may acquire a perfectly good reason for belief from overhearing a speaker’s testimony, but insofar as they have not been told anything by the speaker, they are not in the position to pass the epistemic buck should their beliefs be challenged. While addressees of a speaker’s testimony acquire a distinctively second-personal reason for belief, mere overhearers can only acquire a third-personal reason for belief. There is no such distinction between addressees and overhearers when it comes to the reason for belief generated by a speaker’s arguing. 39 For this reason, Hart claims that testimony provides a reason for belief that is both content-independent and peremptory; Hart (1990), 107. 40 Friedman argues for a similar parallel between authority over action and authority over belief, claiming that both actions and beliefs based on authority involve “the surrender of private judgment” (Friedman (1990), 63 – 68).

Testimony, Address, and the Second Person

281

Since arguing presents considerations that count in favor of a belief completely independently of the speaker’s addressing these considerations to a hearer, a mere overhearer of the speaker’s argument can acquire the very same kind of reason for belief from the argument as can an addressee. Both addressees and overhearers can only acquire third-personal reasons for belief. Imagine, for example, our detective overhearing criminal A present criminal B with an argument to a particular conclusion. In such a case it looks like the detective and criminal B are in a position to acquire the very same kind of reason for belief, a third-personal reason. The same seems true of advice. Since advice presents considerations that tell in favor of a particular course of action completely independently of the speaker’s addressing the advice to a hearer, a mere overhearer of a speaker’s advice can acquire the very same kind of reason for action from the advice as can an addressee. Again, imagine our detective overhearing criminal A advise criminal B to ¦. If the detective is engaged in a similar project to criminal B, the detective might take the considerations that A presents to be a perfectly good reason to ¦. Both the detective and criminal B thus appear to be in a position to acquire the very same kind of third-personal reason. This is not the case with respect to a speaker’s commands. Like mere overhearers of a speaker’s testimony, mere overhearers of a speaker’s command cannot acquire the very same kind of reason for action as addressees of the command. Mere overhearers might very well come to their own conclusion that the speaker’s command is a reliable indicator of what to do in the situation, but they have not themselves been commanded, and hence they cannot acquire the distinctively second-personal kind of reason for action available to addressees. In particular, they cannot acquire a reason for action that entitles them to defer justificatory responsibility for their actions to someone else. Addressees of a speaker’s commands, like addressees of a speaker’s testimony, appear to be ordinarily entitled to defer justificatory responsibility for their actions based on the speaker’s commands. If I ¦ on the basis of having been addressed with a speaker’s command to ¦, and if a third-party challenges my action by arguing that, in the circumstances, ¦-ing was not the thing to do, then I am ordinarily entitled to defer the challenge back to the commander. This is not the case for action based on advice. If I ¦ on the basis of accepting a speaker’s advice, and if a third-party challenges my action by arguing that, in the circumstances, ¦-ing was not the thing to do, then I am not entitled to defer the challenge back to the advisor.

282

Benjamin McMyler

In ¦-ing on the basis of a speaker’s advice, I am coming to my own conclusion that ¦-ing is the thing to do, and so I am solely responsible for the justification of my action.41 The same is true for action based on merely overhearing a speaker’s command. If I proceed to ¦ on the basis of overhearing a speaker commanding someone else to ¦, perhaps judging that the speaker’s command is a reliable indicator of what to do in the situation, and if a third-party challenges my action, then I am not entitled to defer the challenge back to the commander. I have not been commanded to do anything, and hence the commander has not assumed any responsibility for the justification of my action. The fact that justificatory responsibility for belief and action can be parceled out between addressor and addressee in the way that I have described is, I think, an ordinary feature of relations of legitimate authority. Authoritative directives, be they either theoretical, in the case of testimony, or practical, in the case of commands, provide reasons for belief or action that parcel out responsibility for the belief or action between the authority and the subject of the authority. If a legitimate practical authority tells a subject to ¦, and if the subject proceeds to ¦ on this basis, then the authority is partially responsible for the subject’s action. If a third-party challenges the subject’s action, the subject is entitled to defer the challenge back to the authority. Analogously, if a legitimate theoretical authority tells a subject that p, and if the subject proceeds to believe that p on this basis, then the authority is partially responsible for the subject’s belief. If a third-party challenges the subject’s belief, the subject is entitled to defer the challenge back to the authority. The kind of reason for belief provided by testimony thus appears to genuinely parallel the kind of reason for action provided by command. Both testimony and commands provide distinctively second personal reasons that are only acquired by addressees and not by mere overhearers. In drawing this parallel between telling and commanding, I am not claiming that testimony is a form of command, that telling a person that p amounts to commanding her to believe that p. Arguably, belief, along

41 This is consistent with the fact that, in practice, the giving of advice often involves testimony—testifying to various premises and then drawing particular practical conclusions from these premises. In such cases, an audience is not in the position of coming to its own conclusion about the speaker’s testimony, but it is still in the position of coming to its own practical conclusion from the testimony offered in the advice.

Testimony, Address, and the Second Person

283

with a host of other attitudes, simply cannot be commanded. In Zettel Wittgenstein writes, Can one order someone to understand a sentence? Why can’t one tell someone: “Understand that!” Couldn’t I obey the order “Understand this Greek sentence” by learning Greek?—Similarly: One can say “Produce pain in yourself”, but not “Have pain”. One says “Bring yourself into this condition” but not “Be in this condition”.42

Commands aim to direct the will of an audience, and therefore only things that can be willed, namely actions, can be fit objects of command. Belief, like understanding, is not (directly) subject to the will—it is not an action—and therefore belief cannot be (directly) commanded.43 But to say that belief cannot be (directly) commanded is not yet to say that there cannot be reasons for belief that serve to justify beliefs in a way analogous to that in which commands serve to justify actions. Telling an audience that p does not involve commanding an audience to believe that p. Testifying, unlike commanding, aims to direct belief. Nevertheless, this is completely consistent with the idea that the way in which testifying aims to direct belief is analogous to the way in which commanding aims to direct the will. As I have argued, both telling and commanding aim to direct their respective objects second-personally.

VI. Conclusion I have argued that the communicative intentions involved in a speaker’s addressing her testimony to a hearer are genuinely epistemically significant. Addressees of a speaker’s testimony acquire an entitlement to pass the epistemic buck while mere overhearers do not. In this respect, the kind of reason for belief provided by a speaker’s testimony is analogous 42 Wittgenstein (1967), 11. 43 Note that insofar as there is a sense in which belief can be indirectly willed, there is a corresponding sense in which belief can be indirectly commanded. Belief can be indirectly willed in the sense that one can will to perform actions intended to bring about a desired belief, actions like taking a pill or collecting more evidence. These actions are fit objects of command, and so there is nothing incoherent about commanding an audience to perform them. Belief can thus be indirectly commanded in the sense that one can command an audience to perform actions intended to bring about a desired belief. For a related discussion of the various ways in which belief can and cannot be coerced, see McMyler (2011b).

284

Benjamin McMyler

to the kind of reason for action provided by a speaker’s command. Both testifying and commanding provide addressees with a distinctively second-personal reason for belief or action that serves to parcel out justificatory responsibility for the belief or action between the addresser and the addressee. Mere overhearers of a speaker’s testimony or commands do not acquire such reasons. Mere overhearers are in a position similar to the audience of a speaker’s argument or advice—they are in the position of coming to their own conclusion about what to believe or what to do—and hence they can only acquire third-personal reasons for belief or action. If this is right, then pace Darwall, the philosophical significance of the second person actually spans the divide between theoretical and practical reason. The interpersonal relations involved in a speaker’s addressing authoritative reasons to an audience play a genuine role in both practical and theoretical reasoning. This suggests that relations of authority and responsibility between persons play a role in the acquisition of knowledge and justification that has been largely unappreciated by epistemologists. While nearly all contemporary epistemologists accept that a speaker’s testimony is capable of providing a hearer with a perfectly genuine reason for belief, few accept that testimony is capable of providing a hearer with a distinctively authoritative reason for belief. Few accept that testimony is capable of providing a reason for belief that serves to justify belief in a way analogous to that in which authoritative practical directives serve to justify actions. This has the effect of shielding from view the extent to which epistemic agency is an irreducibly social affair dependent upon individuals sharing out responsibility for their beliefs interpersonally. I hope that what I have said here goes at least some distance towards bringing this irreducibly social dimension of epistemic agency back into view.

References Brandom (1983): Robert Brandom, “Asserting”, in: Nous 17, 637 – 650. Brandom (1994): Robert Brandom, Making It Explicit, Cambridge/MA. Burge (1979): Tyler Burge, “Individualism and the Mental”, in: Midwest Studies in Philosophy 4, 73 – 122. Coady (2004): C. A. J. Coady, “Reid and the Social Operations of Mind”, in: T. Cuneo/R. Van Woudenberg (eds.), The Cambridge Companion to Thomas Reid, Cambridge/MA, 180 – 203.

Testimony, Address, and the Second Person

285

Darwall (2006): Stephen Darwall, The Second-Person Standpoint, Cambridge/ MA. Friedman (1990): Richard Friedman, “On the Concept of Authority in Political Philosophy”, in: J. Raz (ed.), Authority, New York, 56 – 91. Fricker (2006): Elisabeth Fricker, “Second-Hand Knowledge”, in: Philosophy and Phenomenological Research 73, 592 – 618. Goldberg (2006): Sanford Goldberg, “Reductionism and the Distinctiveness of Testimonial Knowledge”, in: J. Lackey/E. Sosa (eds.), The Epistemology of Testimony, Oxford, 127 – 144. Hart (1990): Herbert Lionel Adolphus Hart, “Commands and Authoritative Legal Reasons” in: J. Raz (ed.), Authority, New York, 92 – 114. Hieronymi (2008): Pamela Hieronymi, “Responsibility for Believing”, in: Synthese 161, 357 – 373. Hinchman (2005): Edward Hinchman “Telling as Inviting to Trust”, in: Philosophy and Phenomenological Research 70, 562 – 587. Hobbes (1996): Thomas Hobbes, Leviathan, Cambridge/MA. Lackey (2008): Jennifer Lackey, Learning from Words: Testimony as a Source of Knowledge, Oxford. McMyler (2007): Benjamin McMyler, “Knowing at Second Hand”, in: Inquiry 50, 511 – 540. McMyler (2011a): Benjamin McMyler, Testimony, Trust, and Authority, Oxford. McMyler (2011b): Benjamin McMyler, “Doxastic Coercion”, in: Philosophical Quarterly 61, 537 – 557. Moran (2005): Richard Moran, “Getting Told and Being Believed”, in: Philosopher’s Imprint 5, 1 – 29. Putnam (1975): Hillary Putnam, “The Meaning of ‘Meaning’”, in: K. Gunderson (ed.), Language, Mind and Knowledge, Minnesota Studies in the Philosophy of Science 7, 131 – 193. Reid (2002): Thomas Reid, Essays on the Intellectual Powers of Man, Penn State. Searle (1969): John Searle, Speech Acts, Cambridge/MA. Wittgenstein (1967): Ludwig Wittgenstein, Zettel, Berkeley/CA.

II. Knowledge and the Problem of Dualism

Tyler Burge on Disjunctivism*1 John McDowell 1. According to Burge, disjunctivism claims that there is never an explanatorily relevant mental state type in common between (and specific to) a veridical perception and a referential perceptual illusion. And it claims that there is never a mental state type in common between (and specific to) perception of an object and perception of a would-be duplicate substitute for the object that would be, in the context, perceptually indiscernible to the perceiver. The same claims are made with respect to corresponding perceptual beliefs. Disjunctivism makes these claims because it holds that the particular environmental objects (or lack of objects) that are involved in perception are essential to type-identifying all explanatorily relevant perceptual state types and perceptual belief types.2

He dismisses these claims as plainly inconsistent with what is known about how perceptual systems work. Who holds this position? In the body of the essay I have quoted from, there is almost no identification of supposed culprits. (The exception is a mention of Gareth Evans and me.3) In the body of the essay Burge does not even purport to show that his target is not a straw man. In an Appendix, he identifies several people, including me, whom he reads as proponents of the position he attacks. (The mention of Evans and me in the body of the essay is a forward reference to the Appendix.) Now I cannot speak for the others who figure in Burge’s Appendix. (I will make a partial exception for Evans, who cannot speak for himself: §10 below.) But for my part, I do not recognize the doctrine Burge labels “disjunctivism” as a view I hold. His discussion of some of my work * 1

2 3

Originally published in Philosophical Explorations 13 (2010), 243 – 255. Reprinted by permission of the publisher (Taylor & Francis Group, http://www.informaworld.com). Tyler Burge has responded to this essay in Burge (2011), arguing that it leaves his original case against disjunctivism unthreatened. No doubt unsurprisingly, I do not believe that Burge’s response works, and I will try to explain this elsewhere. Meanwhile, readers should know that the conversation is continuing. Burge (2005), 25. Burge (2005), 29.

290

John McDowell

in his Appendix4 makes no case for reading me as a disjunctivist in his sense. He assumes that the work he considers is “motivated by” disjunctivism in his sense, but he cites no statement of that position from me. The fact is that the position is not there in my work. 2. It is true that I have recommended a conception of perceptual experience that is well expressed by a disjunction. But the position I have recommended is not the position Burge attacks. Explaining this will take a while. I will start by considering the allegation that I recognize no state in common between veridical perceptions and perceptual illusions. To begin with, I will ignore Burge’s focus on illusions that are specifically referential. Limiting attention to referential illusions merely obscures the point of the approach to perceptual experience I have recommended. I will come to considerations about reference later. (See §10 below.) I will start by recapitulating how, in the paper Burge considers, I arrive at a disjunctive conception of perceptual experience.5 I build on a suggestion by M. F. Burnyeat about a respect in which Cartesian scepticism differs from ancient scepticism. In Burnyeat’s reading, ancient sceptics acknowledged that perceptual appearances are not open to question, but they did not conceive that as an acknowledgment that a subject’s judgments about how things appear to her are guaranteed to be true; they did not conceive those judgments as bearers of truthvalues, which is to say that in a certain sense they did not conceive them as judgments at all. According to Burnyeat, Descartes went beyond the ancient sceptics’ acknowledgment that appearances are not open to question, and on just that point. Descartes extended the notion of truth to a subject’s judgments about how things appear to her. By doing that, he transformed the acknowledgment that appearances are not open to question into an idea on these lines: there is a region of reality, a subject’s inner world, whose elements include states of affairs consisting in its appearing to the subject that things are a certain way; and the layout of a subject’s inner world, at least in respect of the appearances it is seen as containing, is available to a capacity for knowledge on the part of the subject that, in the transformed version of the thought that appearances are not open to question, is regarded as infallible. 4 5

Burge (2005), 43 – 50. McDowell (1986). The paper is reprinted, with small improvements of wording, in McDowell (1998a), 228 – 259.

Tyler Burge on Disjunctivism

291

In my paper I urge that this move does not yet suffice for a distinctively Cartesian conception of the inner world, in so far as the inner world is in view as having appearances among its elements. For a distinctively Cartesian conception of appearances, we need the idea that how things appear to a subject on some occasion—which is available to an infallible capacity for self-knowledge, according to the transformation of the ancient sceptics’ acknowledgment—is the whole truth about the relevant inner state of affairs. I bring that idea into relief by contrasting it with the idea that appearances that things are thus and so come in two sorts, so that the whole truth about an appearance would need to include not only a specification of how things appear to the subject, but also a determination of which of the two sorts it belongs to. We can express the idea with a disjunction: an appearance is either a case of things being thus and so in a way that is manifest to the subject or a case of its merely seeming to the subject that that is how things are. If we go on regarding appearances as elements in a subject’s inner world, this disjunctive conception embodies a recognizably non-Cartesian conception of that world. When a state of affairs that conforms to the first of those two disjuncts is an element in a subject’s inner world, how things are in that world cannot be fully specified without a commitment as to how things are in the subject’s environment. On this conception, a subject’s inner world does not have the characteristic Cartesian independence from the outer world. I take it that for something to be in a certain state is for an expression with state grammar (as opposed to, say, episode or process grammar) to be truly predicable of it.6 And it should be clear, from the sketch I have just given of my disjunctive conception of experience, that according to me there is a state type in common between those disjuncts: an appearance, a state of a subject that consists in her having it appear to her that things are a certain way. That is surely, on any reasonable view, a mental state type. Is it, perhaps, not explanatorily relevant? We shall have to consider the question: relevant to what explanatory project? (I will come to this in §7 below.)

6

Spelling this out would require saying something about the various grammars. So far as I can see, Burge takes the idea of a state as clear enough without further explanation, and, apart from the remark in the text, I will follow him in this.

292

John McDowell

3. Burge is deaf to the point of the disjunctive conception I have just sketched. He can get no closer to the idea that figures in my formulation of the “good” disjunct—the idea of things being thus and so in a way that is manifest to a subject—than the idea of an experience that is veridical, an experience such that things are as the subject seems to perceive them to be. Burge’s thinking has no room for the idea of an experience in which some aspect of objective reality is there for a subject, perceptually present to her. That is a more demanding condition than an experience’s being merely veridical. I think the idea that experience at its best makes aspects of objective reality present to us is completely natural and intuitive. But the very possibility of such a thing is absent from Burge’s thinking. To have an experience describable in those terms is to have an indefeasible warrant for believing that things are as the experience is revealing them to be. If an aspect of objective reality is perceptually present to someone, there is no possibility, compatibly with her experience’s being as it is, that she might be wrong in believing that things are the way her experience is revealing them to be (as we can put it: not just “is representing them as being”). But perceptual capacities are fallible. And Burge thinks it follows that experience cannot provide indefeasible warrant for belief about the environment. Even when an experience is veridical—that is, even when things are as the subject of the experience seems to be perceiving them to be—the warrant provided by the experience, as Burge sees things, cannot be better than defeasible.7 Since perceptual presence of an aspect of objective reality to a subject would be an indefeasible warrant for belief, Burge cannot even contemplate the idea of perceptual presence of aspects of objective reality to subjects. Burge scolds disjunctivists for failing to understand fallibility.8 But on this topic the failure is all his. Fallibility is a property that attaches to capacities. It is a confusion to think the idea of fallibility can intelligibly carry over to exercises of fallible capacities, as in Burge’s remark: “I believe that all perceptual representations apply fallibly to their referents, in 7

8

This view can be found, with some difficulty, in Burge (2005), 30 f. (The difficulty stems from the fact that Burge is there focusing on fallibility in respect of a perceptual state’s possession of a referent rather than fallibility in respect of, more generally, representing things as they are.) The idea that the fallibility of perceptual capacities shows that perceptual warrants can only be defeasible is explicit in Burge (2003), 535 f. See Burge (2005), 30 f.

Tyler Burge on Disjunctivism

293

any given instance.”9 And it is not much of an improvement if, in the case of fallible capacities to achieve warrant, the confused idea that it makes sense to attribute fallibility to their exercises is replaced by the idea that their exercises cannot yield better than defeasible warrant. A perceptual capacity, in the sense that matters for the disjunctive conception I have sketched, is a capacity—of course fallible—to get into positions in which one has indefeasible warrant for certain beliefs. That is what the capacity is a capacity to do, and that is what one does in non-defective exercises of it, exercises in which its acknowledged fallibility does not kick in. For instance, a capacity to tell whether things in one’s field of vision are green is a capacity—of course fallible—to get into positions in which the greenness of things is visibly there for one, so that one has indefeasible warrant for believing that they are green. That the capacity is fallible means that a possessor of it can be fooled; for instance, if the light is unsuitable for telling the colours of things, one can take something to be green when it is not. It is wrong to think it follows that even when one is not fooled in an exercise of this capacity, one’s position must fall short of having the greenness of something visibly present to one, and thereby having an indefeasible warrant for believing the thing to be green. Fallibility is an imperfection in cognitive capacities. But the mistake I am pointing out is easier to recognize if we consider its analogue in application to other sorts of imperfection in capacities. Some people have a capacity to throw a basketball through the hoop from the freethrow line. Any instantiation of such a capacity is imperfect; even the best players do not make all their free throws. Burge thinks there cannot be a fallible capacity in whose non-defective exercises one gets to have indefeasible warrant for certain beliefs. One might as well think there cannot be a capacity—of course not guaranteed success on all occasions—in whose non-defective exercises one actually makes free throws. 4. It can seem that this way of accommodating fallibility cannot be right. A possessor of a fallible capacity can be fooled; that is just what it means to say that the capacity is fallible. Even with perfect care and attention, one cannot discriminate, at least at the time, a case in which one is being fooled from a case in which one’s capacity is working perfectly. And it can seem to follow that on any occasion on which one’s experience seems to present things as being a certain way, the experience does 9

Burge (2005), 30, my emphasis.

294

John McDowell

not put one in a position to know that this is not one of the cases, ex hypothesi indistinguishable, in which one would be fooled. But if an experience yielded an indefeasible warrant for a belief about one’s environment, the experience would have to put one in a position to know that one was not being fooled. And now it seems to follow that experience cannot yield indefeasible warrants for belief. But this is just another application of that mistake about what follows from fallibility. In a non-defective exercise of a perceptual capacity of the relevant sort, an aspect of objective reality is perceptually there for one. For that to be so is for one to be in a position that constitutes having an indefeasible warrant for believing that things are a certain way. And one cannot be in such a position without being in a position to know that that is one’s position. The capacity—of course fallible—to know on certain occasions that one’s experience is revealing to one that things are a certain way is just an aspect of the capacity—of course fallible—to know through experience that things are that way. It is a single capacity, self-consciously possessed and exercised. And we need to avoid that faulty inference from its fallibility not only in connection with its guise as a capacity for knowledge about one’s environment, but also in connection with its guise as a capacity for self-knowledge, knowledge that one’s experience is revealing an aspect of objective reality to one. Granted, one cannot discriminate cases in which one is being fooled from cases in which one is not being fooled, at least at the time; if one could, one would not be fooled. But it is a form of that same mistake about fallibility to infer that when one is not being fooled one’s experience does not put one in a position to know that one is not being fooled. One’s experience does put one in a position to know that, in an exercise of one’s admittedly fallible capacity. When the capacity stumbles, as—being fallible—it can, one is being fooled, of course without knowing it. But when the capacity does not stumble and one is not being fooled, one is in a position to know, in an exercise of one’s fallible capacity, that one is not being fooled—to know that one’s experience is revealing things to be a certain way, thereby providing one with an indefeasible warrant to believe that things are that way. My disjunctive approach to experience is a way of expressing this rejection of a faulty inference from fallibility. The disjunctive formulation states the point positively: of experiences that seem to reveal to one some aspect of how things are in one’s environment, some make that aspect of reality perceptually present to one, whereas the others only

Tyler Burge on Disjunctivism

295

seem to do that. Making something perceptually present to one goes beyond being merely veridical, the best condition Burge can attribute to exercises of perceptual capacities. Having something perceptually present to one is having an indefeasible warrant for believing that things are a certain way, and it is part of having such a warrant that one is in a position to know that that is one’s position. The capacity to get into such positions is fallible. It does not follow that that cannot really be what it is a capacity to do. 5. This brings me to another of the features of Burge’s thinking that make him deaf to the point of my disjunctive approach to perceptual experience. I have been elaborating a conception of perceptual capacities according to which in a non-defective exercise of such a capacity its possessor is self-consciously equipped with an indefeasible, and so knowledge-constituting, warrant for belief about the environment. This has implications about the knowledge yielded by such perceptual capacities. When a belief owes its status as knowledge to a self-consciously possessed warrant, rationality is at work in the self-conscious possession of the warrant, and the knowledge is—to echo Wilfrid Sellars—a standing in the space of reasons.10 Perceptual capacities, as I am conceiving them, belong to their possessor’s rationality. Now Burge thinks any such conception “hyper-intellectualizes” perceptual knowledge. He thinks such a conception is bound to falsify the perceptual knowledge even of rational animals, subjects whose belief-formation can be responsive to reasons.11 But this accusation merely reflects Burge’s failure to understand the conception. Burge thinks that if one conceives some capacities for perceptual knowledge as rational capacities, capacities whose non-defective exercise yields knowledge of a sort that is a standing in the space of reasons, one is committed to envisaging, in the exercise of such capacities, a normative transition, from perception to belief, that is a piece of reasoning. And he insists, rightly enough, that it would be a gross falsification of perceptual knowledge, even as possessed by rational animals, if one took it to be based on reasoning. But the conception I have been elaborating does not represent the perceptual knowledge of rational animals as based on reasoning. The 10 See Sellars (1997), 76. 11 See Burge (2003), 503 ff., 526 – 30.

296

John McDowell

perceptual capacities of rational animals are among the rational powers of their possessors. Reason—a capacity—is operative in their exercises. That is, reason is operative in a subject’s having aspects of objective reality perceptually present to her. When an aspect of objective reality is perceptually present to a subject, in an exercise of a rational perceptual capacity, there is no need of a normative transition, something on the lines of an inference, for the subject to equip herself with the beliefs whose indefeasible warrant by that perceptual presence qualifies them as knowledgeable. There is no step of reasoning in coming to believe, knowledgeably, that things are as one’s experience is revealing them to be.12 The damaging effect of Burge’s mishandling of fallibility extends into this context. He thinks that to grasp the rational bearing on a perceptual belief of a claim like “I saw such and such” would require more sophistication than it is plausible to attribute to ordinary people.13 But that reflects his thinking the perceptual state invoked by such a claim cannot provide better than defeasible warrant for beliefs. Accordingly, he thinks that to justify describing the state as a seeing, with the factive character that imports, would require working with sophisticated concepts like that of defeating conditions. But if we refuse to let that mistake about fallibility restrict our understanding of the possibilities for perceptual states, we can recapture what I believe is a bit of common sense: that one can conceive oneself as, for instance, having knowledge rationally grounded in exercises of a capacity to recognize instances of greenness when one sees them, without needing a level of sophistication beyond what might be possessed by someone who was only minimally articulate. There is no “hyper-intellectualization” in a conception of a capacity, of course fallible, for knowledge that is grounded, and known by its possessor to be ground12 Burge is distracted here by the fact that in McDowell (1994a), I attributed propositional content to experiences. That was inessential for the fundamental point, that reason is operative in having aspects of objective reality perceptually present to a rational perceiver. (For a shift away from the idea that experiences have propositional content, see McDowell (2009a)). And even if we leave in place the idea that rational perceptual capacities yield experiences with propositional content, there is no implication of a quasi-inferential step from experience to knowledgeable belief. Coming to accept that things are a certain way when one perceives them to be that way is not a piece of reasoning. It happens without any need for rational activity. 13 See Burge (2003), 529.

Tyler Burge on Disjunctivism

297

ed, in the perceptual presence to her of objective states of affairs. An ordinary rational perceiver might not put it in those terms, but that is what she would mean if she said something like “I can tell a green thing when I see one”. It does not require much sophistication to have such thoughts about oneself. 6. Part of Burge’s accusation of “hyper-intellectualization” is the claim that if we conceive the rationality of rational animals—sufficiently mature human beings—as informing their perceptual capacities, we cut off our understanding of human cognitive capacities from their animal roots.14 This too is misplaced. In the conception I have been elaborating, the rationality of rational animals informs their perceptual capacities. That is perfectly consistent with accepting that perceptual capacities are a genus that has rational perceptual capacities as a species, and the genus belongs to animal life as such. Exercising rational perceptual capacities is the form taken by perceiving, which is generically animal, in the special case of rational animals. But properly acknowledging the animal roots of human perception, as Burge sees things, requires more than this abstract acknowledgment that human perception is a species of something generically animal. The capacity to perceive surface colours (for example) is present in some rational and some non-rational animals. On both sides of the divide between the rational and the non-rational, we find cognitive responsiveness, of course fallible, to colours that are uniform over a surface. But the light that reaches an animal’s retinas from different regions of a uniformly coloured surface is not uniform. That an animal, rational or not, is cognitively responsive to uniform surface colours would be unintelligible if the animal did not have a visual system that is capable of extracting information, of course fallibly, about uniform colours of surfaces from that non-uniform retinal input. That much is clear to common-sense reflection, but there is impressive detailed science about how such visual systems work. Animals on both sides of the divide have visual systems that can do that kind of thing. Acknowledging the animal roots of human perception, as Burge sees things, requires giving proper credit to the science that explains how the perceptual sys14 See Burge (2003), 503 f.

298

John McDowell

tems of human beings work, which is typically just a case of how the perceptual systems of a wider range of animals work. Here we come to what infuriates Burge, understandably, about the position he attacks under the label “disjunctivism”. An animal’s capacity to respond cognitively to different colours is fallible, whether the response is a standing in the space of reasons or an instance of some non-rational mode of cognition. Consider occasions on which if we knew how an animal’s visual system works we would be able to explain cognitive responses, by the animal, of a type whose instances are correct if, say, some surface in the animal’s field of vision is green. The state that the visual system gets into on these occasions, in its fallible extracting of information about surface colours from retinal input, is the same whether the animal’s response is correct or incorrect. That is, there are state types of visual systems in common between occasions on which animals have veridical perceptions of colour and occasions on which their colour perception is erroneous. The science that informs us about these common state types is a branch of psychology, so they are psychological state types. Psychology is the science of the mental, so they are surely, in some sense, mental state types. And they are explanatorily relevant; they matter for explaining the fact that animals with the relevant kind of visual systems can perceive uniform surface colours even though the retinal input from different regions of a uniformly coloured surface varies greatly. But disjunctivism as Burge defines it denies that there are any explanatorily relevant mental states in common between veridical perceptions (of anything: surface colour has figured here only an example) and illusions. That is: disjunctivism in Burge’s sense is flatly inconsistent with the science that explains in detail how perceptual systems work. Burge finds it scandalous that philosophers might set themselves in this kind of opposition to an impressive body of science.15 I think he should have devoted more effort to wondering whether he is misreading the philosophers. 7. I mentioned the question: relevant to what explanatory project? (§2 above.) The common state types that figure in my disjunctive conception—states of having it appear to one that things are thus and so—are relevant to understanding how we can accommodate fallibility in an 15 See, e. g., his comparison with Hegel on the number of planets: Burge (2005), 29.

Tyler Burge on Disjunctivism

299

epistemology according to which experience at its best reveals aspects of objective reality to subjects. The common state types whose existence disjunctivism, in Burge’s sense, denies are relevant to understanding how, although sensory input underdetermines its distal aetiology, perceptual systems can account for cognition, of course fallible, of aspects of an animal’s distal environment. These explanatory projects are strikingly different. The project to which the common state types that figure in my conception are relevant is one that Burge, as I have explained, does not understand, since his mistake about fallibility precludes him from contemplating the idea that experience at its best makes aspects of objective reality present to us. The explanatory relevance of the common state types that figure in my conception is certainly not the explanatory relevance of the common state types whose existence disjunctivism, in Burge’s sense, denies. But that is no ground for thinking I am a disjunctivist in Burge’s sense. In fact it seems obvious to me that there must be state types of perceptual systems in common between occasions on which perception enables perceivers to get something right and occasions on which their experience misleads them. As I said in connection with the example of colour constancy, that much is clear to common-sense reflection. It is for empirical work to spell out the details, and Burge describes some fascinating examples of its results. Why should I deny any of that? My purposes in putting forward my disjunctive conception of experience did not require me to say anything about the workings of perceptual systems. But not having occasion to affirm something is not the same as denying it. My disjunctive conception of experience is not a disjunctivism in Burge’s sense. 8. There is a complication here because of the issue I discuss in the other paper of mine that figures in Burge’s Appendix.16 For my present purposes, this is largely a distraction, and I will not respond in detail to Burge’s often point-missing treatment of that paper. That issue I consider there is the relation between the conceptual apparatus that figures in those descriptions of how perceptual systems work, on the one hand, and the conceptual apparatus that figures in 16 McDowell (1994b). This paper is reprinted, with minor improvements of wording, in McDowell (1998b), 341 – 358.

300

John McDowell

talking about exercises of perceptual capacities on the part of perceivers, with rational perceivers as a special case, on the other. It is natural to say, as I have done, that perceptual systems extract information about features of the environment from sensory input. And in the theories Burge cites, perceptual systems are described as moving from registration of arrays of sensory stimulation to states with representational content pertaining to the relevant environmental features: those to which the perceivers whose systems are described by the theories are perceptually sensitive in the way the theories explain. Descriptions of perceptual systems in such terms are shot through with metaphor, in ways Burge partly acknowledges.17 A perceptual system as it were solves a problem posed by the fact that sensory input underdetermines distal states of affairs. It does that by as it were making certain assumptions. But Burge angrily rejects any suggestion that there is metaphor, or at any rate an extended use of language, in attributing representational content to the states of perceptual systems. He assumes that when a good theory of a perceptual system credits it with a state whose representational content coincides with content attributable to an experience of the animal whose perceptual system it is, there is only one perceptual state, which is indifferently attributable to the perceptual system and to the perceiver.18 The implicit principle here—same content, same state—seems remarkably insensitive to the possibility that it might matter who or what is in a state. Experiences are states of perceivers; the states that perceptual systems get into when they as it were solve the problem of moving from sensory input to representations of the environment (which is not something perceivers do even in a metaphorical sense) are states of perceptual systems. The conceptual framework in which talk of perceivers operates is in many ways very different from the conceptual framework in which talk of perceptual systems operates. Burge exploits his assumption that if the content is the same there is only one state to reject any suggestion that the concept of content functions differently in theories of perceptual states from the way it functions in talk of the perceiving that per-

17 See, e. g., Burge (2005), 13. 18 See, e. g., Burge (2005), 50: “Many of the representations and representational states attributed to the perceptual system … are equally perceptual representations or representational states of or for the whole animal”.

Tyler Burge on Disjunctivism

301

ceivers do. But it is not clear why the argument should not go in the opposite direction. In my paper I urge that a good theory of the workings of a perceptual system yields accounts of what enables a perceiver to get into perceptual states, not accounts of what it is for a perceiver to be in those states. Burge scoffs at this. But he does not understand its point. He thinks, quite wrongly, that it is aimed at protecting a prior commitment to disjunctivism in his sense. His assumption that the perceptual states of perceivers just are some of the states of their perceptual systems stands in a mutually supporting relation with his inability to contemplate the idea that exercises of fallible perceptual capacities could provide indefeasible warrant for belief. It makes no sense to suppose a perceptual system might be in a state with respect to which there is no possibility that things might not be as it represents them as being. As I have urged, however, the perceptual capacities of rational perceivers are capacities—as always, of course fallible—to get into states that leave no possibility that things are not as an experience is revealing them to be. If a state of a perceiver is properly described in those terms, it cannot be right to identify it with a state of her perceptual system. A state of a perceptual system cannot have the epistemic significance of a perceptual experience that consists in having an aspect of objective reality perceptually present to one. I would not credit non-rational animals with perceptual capacities whose exercises are, at their best, cases of having aspects of objective reality perceptually present to them. But with non-rational animals there is an analogue of the point I have just made about rational perceivers. In a normal environment, an environment of the sort in which its perceptual systems evolved, exercises of an animal’s perceptual capacities are, when things go well, cases of awareness of features of its environment. It does not make sense to conceive an animal’s perceptual systems as being aware of features of the animal’s environment. Here too, a state of a perceptual system cannot have the cognitive significance of an animal’s exercise of a perceptual capacity. All this is perfectly consistent with acknowledging that a good theory of the perceptual systems of perceiving animals, rational or not, provides an illuminating explanation of their perceptual capacities. It seems exactly right to say that the explanation is enabling, not constitutive; the states of perceptual systems that do that explanatory work are not the states of perceivers that they explain.

302

John McDowell

To claim that the explanation is not constitutive is not to deprecate its power. Burge fails to distinguish suggesting that a certain use of language is metaphorical or extended from debunking its power to illuminate. There need be nothing wrong with extended uses of language. Claiming that some of the language they use is metaphorical or extended does not prevent me from applauding theories that say in detail how perceptual systems move from sensory arrays to states that, in the sense that is appropriate for describing states of perceptual systems, represent features of the environment. In my paper I make it as clear as I know how that I have no intention of deprecating the explanatory power of theories of perceptual systems in such terms. I raise a conceptual question about how we should understand the language used in such theories. Burge responds, bizarrely, as if to a case of empirical ignorance.19 9. At one point in “Disjunctivism and Perceptual Psychology” Burge entertains the possibility of a position that, without denying the existence of common state types in the way that defines disjunctivism in his sense, classifies states more finely than by the types that figure in the position he sets against disjunctivism.20 He envisages, for instance, that someone might count veridical perceptual states as a type. For my purposes it would be more to the point to consider a type whose instances are states in which experience makes aspects of objective reality present to a perceiver. About a position with this structure, Burge says: “Whether this flexible approach leads to insight is an interesting issue, but it is not our issue. Our issue is whether the disjunctivist denial of common explanatorily relevant kinds of states across the three sorts of cases is correct.”21 Remarkably enough, this has the effect of excluding from “our issue” any consideration of my disjunctive conception of experience. This is just one more indication that Burge does not understand the point of the conception. After noting the possibility that one might count being in suitable relations to objects as states of individuals, Burge adds an endnote in 19 Not just in his response to me. See Burge (2005), 70, fn. 20, where he responds to Frances Egan’s obviously conceptual queries about the sense in which theories of psychological subsystems deal with content in Egan (1995) by claiming that she is “misinformed”. 20 See Burge (2005), 26 f. 21 Burge (2005), 27.

Tyler Burge on Disjunctivism

303

which he discourages such “flexible” approaches by criticizing Timothy Williamson’s argument that seeing, remembering, and knowing are states of mind.22 Williamson equates the claim that “knowing is a mental state in every reasonable sense of that term” with the claim that “there is no more restrictive but still reasonable sense of ‘mental’ in which knowing can be factored like believing truly into a combination of mental states with non-mental conditions”. Burge responds: “These are definitely not equivalent. There could be no natural ‘factor’ constituting the nonmental conditions, so that knowledge is a primitive kind, but knowledge could still entail truth and belief, where belief is construed as a mental state in a sense that knowledge is not.” But this does not address Williamson’s point. Williamson need not dispute that knowledge entails belief. But to say that knowledge entails belief is not to say that a state of knowledge can be factored into some non-mental conditions (which need not be separately specifiable) and a belief conceived as being the belief it is independently of the non-mental conditions. Such factoring fails, consistently with the entailment, if the state is the belief it is only because it is the knowledge it is. Of course belief is a mental state, but if factoring fails in that way, we have cases in which a belief is the mental state it is only because it is a state of knowledge. And that seems indistinguishable from saying that knowledge is a mental state in any reasonable sense. Analogously, it is part of the point of my disjunctive conception of experience that having an aspect of objective reality perceptually present to one entails having it appear to one that things are a certain way. But that is not to say that having an aspect of objective reality perceptually present to one can be factored into some non-mental conditions and an appearance conceived as being the mental state it is independently of the non-mental conditions. The factoring fails; the state is the appearance it is only because it is a state of having something perceptually present to one.

22 See Burge (2005), 72, fn. 28. There is a shift between the text and the endnote. The position Burge contemplates in the text reflects his characterization of disjunctivism as a position about referential substitutions and referential illusions. Williamson’s point is about knowledge in general, not specifically knowledge that relates the knower to determinate objects. Williamson’s argument is in Williamson (1995).

304

John McDowell

10. So far I have elaborated my disjunctive conception of experience without needing to broach the topic of singular thought. As I said (§2 above), that is only a distraction, so far as the conception of experience is concerned. My paper “Singular Thought and the Extent of Inner Space” is (of course) about singular thought. Burge seems to think I use a disjunctive conception of singular thought to argue for a disjunctive conception of experience. But the order is the reverse: having set out the disjunctive conception of experience that I have been elaborating here, I go on to exploit an analogy with it in order to bring out features of the conception of singular thought that I recommend. On that conception, a thinker of a singular thought self-consciously stands in a certain relation to a determinate object, a relation of directing her thought at it (under a mode of presentation).23 A state of directing one’s thought at a different object is, in an obvious sense, a different state. And if there is no object to which the thinker stands in the relation in which she takes herself to stand to the supposed object of her thought, there is no state consisting in her standing in that relation to an object. There can be a state that consists in the obtaining of a relation only if there are relata for the relation. As with my conception of experience, this conception is not a form of disjunctivism in Burge’s sense. I do not deny that there is a state type in common between the cases. On the contrary, it is part of the conception that there is a state type in common between them: the state of seeming to stand in a specific intentional relation to an object. The conception does not imply anything about perceptual systems. In particular, it does not include a denial that a good theory of a thinker’s perceptual systems would find a state type in common between actually standing in a specific intentional relation to an object and merely seeming to. In fact I think it is obvious that any plausible theory of perceptual systems would posit a state type in common between such states. In the case of experience, the corresponding commonality at the level of perceptual systems serves an explanatory purpose, but it is irrelevant to 23 “Mode of presentation” is a bit of Fregean apparatus. It is common for philosophers not to see how its presence in the conception can be compatible with the thinking’s being a matter of standing in a relation to the object. I explain how a feature of Burge’s thinking underlies his version of blindness to this possibility in McDowell (1998c).

Tyler Burge on Disjunctivism

305

the question how states of perceivers should be classified. Burge’s assumption that states of perceivers just are states of their perceptual systems, together with his mistake about fallibility, blinds him to the possibility of characterizing the “good” disjunct in terms of an experience that makes things manifest. Similarly with singular thought. The commonality I take to be obvious at the level of theories of perceptual systems serves an explanatory purpose, but it is irrelevant to the question how we should classify states of a thinker, as opposed to states of the perceptual systems that enable her to stand in intentional relations to objects. Here Burge’s assumption that the state types that would be recognized in a good theory of perceptual systems are all the state types we need to recognize prevents him from being able to see the point of the conception of singular thought. The counterpart to his not being able to contemplate the idea of an experience that provides indefeasible warrant for a belief is that he cannot contemplate the idea of an intentional directedness of a thought at an object that is, at the most fundamental level of classification, a relation between the thinker and the object. Burge makes a partial concession to object-dependent classifications of mental states.24 He reconciles this with the basic thrust of his essay by exploiting what is in effect a form of the factoring strategy that, in a different context, Williamson undermines. (See §9 above.) He sees objectdependent classifications of perceptions or experience-based beliefs as more fine-grained partitionings, useful for certain explanatory purposes, of state types that figure at a more fundamental level of classification, state types that are not object-dependent but still supposedly involve content with a singular element. I do not believe this factoring move is any more persuasive than the ones I considered in §9.25 The conception of singular thought that I recommend in my paper and have sketched here is in its essentials due to Gareth Evans. Burge dismisses Evans’s views as “idiosyncratic and quaint”.26 He bases this assessment on a discussion of an argument about information-based communication in chapter 9 of The Varieties of Reference, and a brief com24 Burge (2005), 31 – 40. 25 It would take me too far afield to go into this in detail. An important point is that the supposed singular contents of the states into which, along with extramental conditions, object-dependent states are supposed to be factored can only be content-schemata, not contents proper. Evans makes the point (in a slightly different context) in Evans (1982), 202. 26 Burge (2005), 51.

306

John McDowell

ment on a line of thought that Evans sets out, in an anticipatory spirit, at the end of chapter 5. I think the argument in chapter 9 is indeed vulnerable. But it is not central to Evans’s position. Burge does not understand Evans; what shows this is that he does not see that to dismiss the conception—especially the application of it that is in question here—he would need to undermine the neo-Fregean account of perceptually demonstrative thought that Evans gives in chapter 6. 11. In “Singular Thought and the Extent of Inner Space”, I criticize views according to which the layout of a subject’s inner world “would be exactly as it is however things stood outside it”.27 I argue that such views make it unintelligible how experience can be anything but dark, how it can purport to be revelatory of the world. Burge reads this as an attempt to recommend disjunctivism in his sense; as I have explained, it is no such thing. He speculates that when I object to the idea that inner states as such would be as they are however things were in a subject’s environment, I am giving expression to a parochially British concern with an outmoded problem, that of overcoming an empiricistic veil-of-ideas scepticism.28 Let me end by trying to indicate how wildly off target this speculation is. The idea of inner states that would be as they are however things stood outside the inner world is clearly apt in the connection in which I introduce it, when I am describing the fully Cartesian conception of the inner world that serves as a foil to my conception of experience. (See §2 above.) But states of perceptual systems, as Burge perfectly reasonably conceives them, would equally be, in the relevant sense, exactly as they are however things stood outside them. That is not to deny what Burge insists on, that the types instantiated by such states are individuated, in a good theory of the states, by causal relations between perceptual systems and aspects of the normal environment of animals with those systems. But any instance of the perceptual-system state type that accounts for an animal’s capacity to perceive, say, that something is green would be exactly as it is—it would instantiate the type—whether or not there was a green thing perceptually confronting the animal. That is just a way of expressing Burge’s own claim, which I accept, that the state type is common between occasions on which there is a green thing perceptually confronting the animal and occasions on which there is 27 McDowell (1986), 151. 28 Burge (2005), 74, fn. 44 and see Burge (2005), 73, fn. 40.

Tyler Burge on Disjunctivism

307

not. This point about state types in perceptual systems, in the context of Burge’s equating states of perceivers with states of their perceptual systems, conspires with his mishandling of fallibility to prevent him from contemplating the idea of experiences that reveal things to be a certain way, experiences that provide indefeasible warrant for belief. Many philosophers think we can make room for something genuinely recognizable as experience-based knowledge while assuming that experience cannot provide indefeasible warrant for belief. I think that is a fantasy. If it did not seem that there is no alternative, no one would take seriously the idea that one can know something about one’s environment on the basis of an experience such that it is possible, compatibly with the experience’s being as it is, that things are not as one supposedly knows them to be.29 And something similar holds in the case of capacities for non-rational awareness. A state of an animal cannot be genuinely recognizable as awareness of a feature of the environment if the state would be just as it is even if the feature of the environment was not there. The point is not just epistemological, and its wider significance is further brought out by the counterpart conception of singular thought. I believe that if we do not countenance capacities in whose non-defective exercise objective states of affairs are perceptually present to subjects and capacities in whose non-defective exercise subjects self-consciously stand in intentional relations to objects, we can only pretend to make sense of the idea of subjects for whom there is a world, as opposed to beings that merely live in the world. And the less demanding counterpart for non-rational perceivers is analogously necessary if we are to make sense of the idea of beings for which there is an environment. I will resist the temptation to speculate about whether there is something deeper than a failure of imagination about fallibility that prevents Burge from comprehending philosophical thoughts on these lines, let alone engaging with them.

References Burge (2003): Tyler Burge, “Perceptual Entitlement”, in: Philosophy and Phenomenological Research 67, 503 – 548. 29 See the devastating treatment of “fallibilist” epistemologies for perceptual knowledge in chapter 5 of Rödl (2007).

308

John McDowell

Burge (2005): Tyler Burge, “Disjunctivism and Perceptual Psychology”, in: Philosophical Topics 33 (1), 1 – 78. Burge (2011): Tyler Burge, “Disjunctivism Again”, in: Philosophical Explorations 14, 43 – 80. Egan (1995): Frances Egan, “Computation and Content”, in: Philosophical Review 105, 181 – 203. Evans (1982): Gareth Evans, The Varieties of Reference, Oxford. McDowell (1986): John McDowell, “Singular Thought and the Extent of Inner Space”, in: P. Pettit/J. McDowell (eds.), Subject, Thought, and Context, Oxford, 137 – 68. McDowell (1994a): John McDowell, Mind and World, Cambridge/MA. McDowell (1994b): John McDowell, “The Content of Perceptual Experience”, in: Philosophical Quarterly 44, 190 – 205. McDowell (1998a): John McDowell, Meaning, Knowledge, and Reality, Cambridge/MA. McDowell (1998b): John McDowell, Mind, Value, and Reality, Cambridge/ MA. McDowell (1998c): John McDowell, “De Re Senses”, in: McDowell (1998a), 214 – 227. McDowell (2009a): John McDowell, “Avoiding the Myth of the Given”, in: McDowell (2009b), 256 – 272. McDowell (2009b): John McDowell, Having the World in View. Essays on Kant, Hegel, and Sellars, Cambridge/MA. Rödl (2007): Sebastian Rödl, Self-Consciousness, Cambridge/MA. Sellars (1997): Wilfrid Sellars, Empiricism and the Philosophy of Mind, Cambridge/ MA. Williamson (1995): Timothy Williamson, “Is Knowing a State of Mind?”, in: Mind 104, 533 – 565.

Motivating Disjunctivism Thomas Lockhart Disjunctivism in the philosophy of perception claims that: (*) in order to maintain the idea that perceptions allow us to gain knowledge of the world, we have to abandon the idea that a subject currently having a veridical perceptual experience would be having the same perceptual experience if she were instead having an indistinguishable hallucination.

In this paper, I argue that both M.G.F. Martin and John McDowell subscribe to this formulation of disjunctivism, but I argue that their implementations of it are incompatible. Further, I argue that the objection from perceptual content, which some critics have brought against disjunctivism, applies only to the variety of disjunctivism that Martin defends. I conclude that McDowell’s disjunctivism represents a stronger candidate for a successful implementation of (*). More importantly, I argue that a construal of (*) along the lines proposed by Martin robs disjunctivism of its initial promise.

1. Introduction According to the disjunctivist, the history of epistemology has shown how difficult it is to hold together the following two thoughts: 1. In virtue of perception, we can come to know about the world around us. 2. A subject, currently having a veridical perception, is having an experience which is exactly the same as the experience she would be having were she instead having an indistinguishable hallucination. The disjunctivist claims that the reason philosophers have found it so hard to reconcile these two claims is that it is impossible to do so. In this, disjunctivism agrees with scepticism. The sceptic abandons the first thought. Instead, disjunctivism urges that if we want to maintain the first thought, we must defend a conception of perception which rejects (a common understanding of) the second.

310

Thomas Lockhart

My focus in this paper will be on the work of the two most influential contemporary proponents of disjunctivism: M. G. F. Martin and John McDowell. My first goal is both to explain the significant differences between their views and correctly to situate these views so that we can see their common core motivation. I’ll start by articulating what I take to be the core commitments of disjunctivism to which they both subscribe. I’ll argue that the difference between Martin and McDowell can rather fruitfully be understood as a disagreement about the implications of this core view, and thus a disagreement about what—beyond the common core—the disjunctivist is required to defend. Once the arguments of Martin and McDowell are clearly separated, I’ll make the case that their two positions are, far from being complementary, essentially incompatible. Disjunctivism has attracted a lot of attention and criticism. But the relationship between McDowell and Martin has been misunderstood and therefore this attention and criticism is often misdirected. Sometimes, philosophers assume that Martin and McDowell must be defending exactly the same view, and so they think that criticisms of the view Martin defends must apply also against McDowell.1 In this paper, I’ll consider an objection which has been brought against Martin’s disjunctivism, and show why it does not cut against the version of disjunctivism which McDowell defends. I’m not the first to notice that confusion can be avoided if we carefully separate Martin’s disjunctivism from that of McDowell. For example, Byrne and Logue notice the significant differences between the two authors. But as Byrne and Logue characterize the views of the two, they are compatible and (in one direction) independent: Epistemological disjunctivism [which they associate with McDowell] is not a rival to metaphysical disjunctivism [Martin]; in fact … the latter leads naturally if not inexorably to the former. However, epistemological disjunctivism is quite compatible with the denial of metaphysical disjunctivism.2 1

2

A good example of this phenomenon is Burge (2005). McDowell, in his McDowell (2010) (reprinted in this volume), rightly disowns the view Burge ascribes to him. Another good example of this confusion can be found in Travis (2005). Travis assumes that the only viable form of disjunctivism is the form Martin defends. He notes that this form of disjunctivism is inconsistent with views that McDowell elsewhere holds, and so he concludes that, in so far as McDowell claims to be a disjunctivist, he is guilty of inconsistency. By the end of this paper we’ll be in a position to see how this charge is wrong. Byrne/Logue (2008), 67.

Motivating Disjunctivism

311

Byrne and Logue, however, fail to recognize (a) that there is a critical sense in which Martin and McDowell are engaged in the same project and (b) that when we recognize what they have in common, we can clearly see that their views are incompatible.3 It is precisely this incompatibility which I articulate in this paper. In §2 I argue that at the heart of disjunctivism is the following thought: (*) in order to maintain the idea that perceptions allow us to gain knowledge of the world, we have to abandon the idea that a subject currently having a veridical perception would be having the same perceptual experience if she were instead to be having an indistinguishable hallucination.

Martin and McDowell have very different arguments for (*). In §3, I argue that Martin believes that in order to defend (*), the disjunctivist will have to insist that there is no non-trivial sense in which veridical and non-veridical perceptions have anything ‘in common’, and I rehearse his argument for this claim. In §4, I show that McDowell’s implementation of (*) does not require such a claim. In §5, I argue for the following three claims: (a) that Martin will have to reject any view according to which perceptions have any kind of representational or conceptual content, (b) that McDowell’s disjunctivism can and must have

3

Byrne and Logue see Martin’s position as the more interesting contender for the title ‘disjunctivism’ because it enters a bold metaphysical claim about the nature of perceptual experience, whereas they see McDowell as merely making a point about the epistemological status of veridical and non-veridical perceptual experience. But if I am right, McDowell is no less interested in the nature of veridical perceptual experience—indeed, his claim about the epistemological credentials of veridical perceptual experience follows from a claim about what a veridical perceptual experience is (Byrne and Logue seem to lose track of this in their eagerness to mark a distinction between the two authors). Haddock and Macpherson also distinguish between McDowell’s disjunctivism and that of Martin, and follow Byrne and Logue in labelling McDowell’s disjunctivism ‘epistemological’ (Haddock and Macpherson are somewhat more nuanced in their discussion of the relative implications between the views—see Haddock/Macpherson (2008b), 12 f. and 17 f.). It is certainly open to these authors to discriminate and label the forms of disjunctivism as they do. My point in this paper is that this kind of labelling runs the risk of losing track of the sense in which Martin and McDowell are incompatibly attempting to implement the same core idea. Excessive attention to the differences between the various species of disjunctivism runs the risk of overlooking or trivializing the sense in which they are all species of the same genus.

312

Thomas Lockhart

room for the idea that perceptions themselves have conceptual content,4 and (c) that McDowell has a ready way to reject Martin’s argument that there is no non-trivial sense in which veridical and non-veridical perceptions have ‘something in common’. I’ll conclude in §6 by suggesting that this means that McDowell’s is a more promising implementation of the basic idea of disjunctivism.

2. The Basic Claim of Disjunctivism Before beginning to see the differences between the two philosophers I consider in this paper, I start by offering a characterization of the basic insight at the heart of disjunctivism—an insight which I take Martin and McDowell both to want to be able to claim for their respective versions of disjunctivism.5 It is (or it should be) a truism that in virtue of perceptual experience we can have access to the world around us. One of the tasks facing the philosopher of perception is to articulate a conception of perception which will allow us to maintain this truism in the face of arguments which seem to deprive us of the resources we need to defend it. Consider three perceptual experiences Smith might have. First, she might see a lemon6 in her kitchen. Second, she might see a perfect wax copy of a lemon in her kitchen. Third, she might hallucinate a lemon in her kitchen. If the illusion or the hallucination is good enough, then Smith won’t be able to distinguish it from a veridical perceptual experience of a lemon; that is, we can imagine a situation in which Smith 4

5

6

McDowell’s disjunctivism can make room for the idea that experiences themselves have propositional content. However, McDowell’s own current view about the content of perceptual experience is weaker—perceptual experiences ‘draw on’ conceptual capacities. See McDowell (2008a). As we’ll see, even this view would be incompatible with Martin’s disjunctivism. I don’t argue directly for the claim that the presentation of the basic idea of disjunctivism I offer in this section is correct—instead, the definition is supposed to prove its worth by its fruitfulness in helping us to see a common background against which emerges the sense in which McDowell and Martin are attempting to implement the same project in incompatible ways. Finding a properly generic characterization of disjunctivism matters not only in order to clearly identify the specific difference which distinguishes Martin’s variant from McDowell’s but also in order to frame the genus in such a way that one can see how it applies to other areas of philosophy. To use a canonical example.

Motivating Disjunctivism

313

cannot tell whether she is having a veridical perceptual experience of a lemon or a non-veridical perception of a lemon. This has led philosophers to conclude that we are forced to accede to some version of an indistinguishability thesis regarding the relation between veridical and non-veridical experiences. So let us, for the moment, accept some version of the following, intentionally vaguely-worded, version of such a thesis: IND Suppose a subject is currently having a veridical perception. If, instead, she were to be having a sufficiently impressive non-veridical perception, her experience would be, to her, indistinguishable from the experience she is currently having.

One familiar tendency amongst epistemologists is to take this observation to show that there is some common item that figures in the subject’s experience in both cases. I will call this way of further specifying the implicit commitments involved in such an indistinguishability thesis, via the postulation of a common experiential component, the traditional epistemologist’s conception of an apparent perception. Thus we arrive at the following way of specifying the concept of an apparent perception: TAP An apparent perception of v is the component in a subject’s experience which figures in both a veridical and a non-veridical perceptual experience of v.

Once we thus introduce the notion of an apparent perception, it is natural to ask what it is that distinguishes a veridical perception from a nonveridical perception by asking what it is that a veridical perception has, over and above being an apparent perception, that makes it a veridical perception. Let’s call the following the traditional epistemologist’s equation: TEE Veridical perception of v = Apparent perception of v + x.

The resulting project (which has preoccupied much of the last few centuries of epistemology) might be summed up by the following question: Can we find a non-trivial solution to TEE? Let us say that a philosopher who attempts to find a non-trivial solution to TEE subscribes to the conjunctive conception of perceptual experience (or conjunctivism, for short)—that is, a conception of experience according to which a veridical perception is an apparent perception (conceived of as the common factor between a veridical and non-veridical perception) plus something else. The conjunctivist starts by identifying something which he calls Smith’s ‘experience’ in terms which abstract from the question whether her perception is veridical or not. If we accept this analysis of the notion

314

Thomas Lockhart

of an apparent perception (that is, in terms of a TAP), it makes sense to ask what additional feature is present in the case of a veridical perception over and above this common experience, and thus attempt to analyze a veridical perception into a TAP plus x (for some x). According to the disjunctivist, if we accept (with the conjunctivist) that Smith has a perceptual experience in the good case which would be, in the required sense, exactly the same in the bad case, then we will have to give up on the idea that we can have access to the world around us in virtue of our perceptual experience. It is important to see that a rejection of this way of analyzing the concept of an apparent perception does not in itself amount to a rejection of the familiar fact that if Smith were hallucinating she might not be able to tell that she was hallucinating (IND). All the disjunctivist claims is that we need not—and should not—conclude from the bare fact of IND to the adequacy of TAP as the analysis of the concept of an apparent perception. We can perhaps start to see more clearly what the disjunctivist is rejecting by first noting the rather different and philosophically less committal manner in which he seeks to account for the fact that non-veridical perceptions can be, from the point of view of the perceiving subject, indiscriminable from veridical perceptions. His mode of accounting for this fact turns on the idea that the philosopher of perception ought to start with the case of veridical perception, working out from there to an understanding of the case of non-veridical perception through an analysis of the manner in which it falls short of or is an otherwise defective instance of a properly veridical perception. This leads him to a philosophically less committal conception of an apparent perception than the one a traditional epistemologist employs, in which an apparent perception is conceived of (not as a genus which includes both veridical and non-veridical perceptions as its species, but rather) as a deficient mode of perception. Let us call this disjunctivist concept of an apparent perception a merely apparent perception, and define it as follows: MAP A merely apparent perception is a perception which is in fact not veridical, but which turns out, as a further matter of fact, for a given subject in a given circumstance, to be indistinguishable from a veridical perception.

Whereas the notion of a TAP is sufficiently philosophically committal so as to be one the disjunctivist is obliged to reject, the notion of a MAP is sufficiently philosophically non-committal so as to be one which both a conjunctivist and a disjunctivist can happily employ; they will, however, give very different analyses of what is involved in

Motivating Disjunctivism

315

such cases of merely apparent perception. (A conjunctivist will insist on employing the notion of a TAP to spell out what is involved in the case of a MAP.) To see how the disjunctivist analysis differs radically from the conjunctivist one, it will help first to introduce a further notion—namely that of the genus which subsumes both deficient and non-deficient modes of perception as its species. I will call this an ostensible perception: OP An ostensible perception of v is either a veridical perception of v or a merely apparent perception of v.

The notion of an OP provides us with a different model for how to accommodate the truth of IND. With OP in hand, nothing forces us to define a MAP in terms of a TAP. It is open to us to analyze the notion of a MAP in such a way that it can be made to depend essentially on the notion of a veridical perception (for example, if we take a MAP to be a defective case of veridical perception). Indeed, the disjunctivist will want to urge a stronger claim here: namely, that we cannot correctly account for the truth of IND without properly appreciating the asymmetry between the two disjuncts here, and acknowledging the priority of the veridical over the non-veridical case. Consequently, from the disjunctivist point of view, the notion of a TAP builds too much into its description of what is involved in the case of a MAP. TAP and OP are equally attempts to introduce a comparatively inclusive concept of an apparent perception—that is, a concept of an apparent perception which is able to encompass both cases of veridical perception and non-veridical perception as its instances. TAP and OP involve different ways of spelling out what the two cases ‘have in common’. A TAP is supposed to be a sort of experience which is a common factor between a veridical and a non-veridical perception; whereas an OP is not an experience which is a common factor, but rather a concept of what two specifically different instances of a concept have generically in common. For many philosophers, conjunctivism has seemed compulsory. This is in part because, in their employment of the ordinary language expression ‘apparent perception’, traditional epistemologists have tended to equivocate between the three different ways of understanding what an apparent perception is encapsulated in TAP, MAP, and OP respectively. Part of the interest of disjunctivism lies in the mere observation that once we clearly disambiguate these three different ways of understanding what is at issue in a case of apparent perception, conjunctivism is no longer compulsory. Once we come to see that disjunctivism itself is an

316

Thomas Lockhart

option, then the conjunctivist, in order to defend the intelligibility of his project of solving TEE, will now be obliged to show that the bare concept of a MAP itself must be analyzed in terms of a TAP. Since we are now faced with two possible ways of accounting for a MAP, either in terms of a TAP or an OP, the traditional epistemologist is no longer able simply to slide from the philosophically non-committal notion of a MAP to the philosophically committal notion of a TAP with no further argument. The ambition of the disjunctivist, however, is not merely to shift the burden from himself (about how to analyze the concept of an apparent perception) back onto the traditional epistemologist, but rather to show that the traditional epistemologist’s analysis of that concept, in effect, concedes to the epistemological sceptic that which he most desires—namely that our perception always falls short of taking in what in the case. The disjunctivist therefore is concerned to mount an argument (as we shall see) to the effect that we will never arrive at a satisfactory account of veridical perception—one which allows us to maintain the idea that in virtue of perception we can come to know the world around us—if we insist on a conjunctivist schema for analyzing what is involved in a veridical perception. So, on this account of what is generically constitutive of disjunctivism, what different species of disjunctivist will all share is the idea that we will not be able to arrive at a satisfactory account of veridical perception if we insist upon analyzing that which a veridical and a non-veridical perception ‘have in common’ in terms of the concept of a TAP. This leaves room for different sorts of disjunctivist to offer differing accounts of what veridical and non-veridical perceptions do ‘have in common’, while all equally eschewing a form of analysis in terms of a common experiential constituent. This is precisely the question over which the two disjunctivists I consider in this paper differ. However, before we are in a position to see precisely how Martin and McDowell differ in this regard, first we need to uncover a further layer of difference between them—one which arises in their more specific reasons for holding that the conjunctivist will never be able to arrive at a satisfactory account of the fact that veridical perception allows us to come to know about the world. In the next section, §3, I will explore Martin’s disjunctivist critique of the traditional epistemologist. In the following section, §4, I will highlight some respects in which McDowell’s disjunctivist critique of the traditional epistemologist differs from that of Martin. Finally, in §5, we will be in a position to specify how Martin and McDowell differ

Motivating Disjunctivism

317

on the particular question of what it is that veridical and non-veridical perceptual experiences ‘have in common’, and thus to see why there is an important difference in the species of disjunctivism they respectively advocate.

3. Martin’s Disjunctivism For Martin, disjunctivism arises as a result of an attempt to defend what he calls ‘Naive Realism’. What is Naive Realism? One might have thought that a maximally philosophically non-committal way of specifying what Naive Realism in the philosophy of perception comes to would be something like the following: “Perceiving is a way of knowing the world”. Or perhaps: “Perception discloses what is the case”. Martin’s ways of specifying what Naive Realism comes to tend to be terminologically more heavy-handed. At one point, he tells us that Naive Realism is the claim that “veridical perception is a relation of awareness to mind-independent objects”.7 On the face of it, even this formulation might still seem to be fairly innocuous—so much so that one might plausibly take it as a constraint on any philosophy of perception that it be able to endorse a claim which can be expressed using this form of words. As he goes on, however, Martin tends to further burden his specifications of that to which the Naive Realist is committed, so that they come to feel less and less naive. Thus, for example, Martin takes the formula to be amenable to the following gloss: So-Called Naive Realism “That is to say, taking experiences to be episodes or events, the naive realist supposes that some such episodes have as constituents mind-independent objects.”8

This gloss ought to strike one as a little stronger than that which it glosses, and when I use ‘So-Called Naive Realism’ in this paper I will intend it in this stronger way. The crucial point for the moment is simply that Martin takes achieving clarity about that to which the Naive Realist is committed to be the proper starting point for a disjunctivist line of thought.9 7 8 9

Martin (2006), 357. Ibid. It is worth noting that already here we encounter a significant difference between Martin and McDowell: namely in their respective accounts of what the philosophical starting point for a disjunctivist line of thought ought to

318

Thomas Lockhart

Martin takes So-Called Naive Realism to express the thesis that best describes our pre-reflective common-sense attitude towards perception; and he thinks, therefore, that we should abandon this thesis only if we are forced to by a compelling argument. Furthermore, Martin thinks that unless we can defend So-Called Naive Realism, we will find ourselves in a position where we lack knowledge of the empirical world “because we lack the kind of perceptual access to it we supposed ourselves to have”.10 For these allied reasons, Martin thinks that we should do what we can to defend So-Called Naive Realism. Martin thinks that a defence of So-Called Naive Realism will require us to adopt disjunctivism. For that reason, let us provisionally suppose he is right that we should defend So-Called Naive Realism, and ask why he thinks it will require disjunctivism. As Martin summarizes things: “Disjunctivism … is reactive: it blocks a line of argument which would threaten to show we have no knowledge of the empirical world because we lack the kind of perceptual access to it we supposed ourselves to have”.11 The attack on Naive Realism which disjunctivism repels is what Martin calls the Argument From Hallucination. The Argument From Hallucination is, he writes, “best underst[ood] … as a form of reductio against Naive Realism”.12 be. Martin starts by announcing that the aim of disjunctivism is to vindicate our entitlement to a particular epistemological thesis—one which he identifies with Naive Realism. Thus, for Martin, in order to understand what disjunctivism is, we first need an account of what Naive Realism is, what it commits us to, and what would be involved in defending it. Martin’s aim is to argue for a philosophical thesis which he takes to be true, and against one which he takes to be false. This is very different from McDowell’s self-understanding. McDowell takes the task of disjunctivism to be to make perspicuous how any analysis of what is involved in an exercise of a perceptual capacity must allow it to come intelligibly into view as an exercise of a capacity. McDowell’s criticism of the traditional epistemologist is that his analysis fails to do this. The criticism turns on the idea that (in the traditional epistemologist’s account of the relation between a successful and an unsuccessful exercise of our perceptual capacity) he no longer operates with a coherent conception of a cognitive capacity. On this account, the problem with the traditional epistemological concept of an apparent perception is that it is incoherent. Such a critique simply returns to us the idea that our perceptual capacity is a capacity which admits of successful and unsuccessful exercises. It is not meant to furnish a preliminary move toward a defence of some particular substansive philosophical thesis in epistemology. 10 Martin (2006), 355. 11 Ibid. 12 Ibid., 357.

Motivating Disjunctivism

319

In order to allow the crucial differences between Martin and McDowell which will concern us later to come into view, it will be helpful initially to summarize the overall shape of Martin’s argument for disjunctivism in the following schematic manner: 1. The Argument From Hallucination constitutes a reductio of SoCalled Naive Realism. 2. Disjunctivism is the best way to avoid this reductio. 3. We should defend So-Called Naive Realism because it represents our natural pre-theoretic conception of perception without which we would be forced to conclude that we lack knowledge of the empirical world “because we lack the kind of perceptual access to it we supposed ourselves to have”. 4. By 2, disjunctivism is the best way to defend So-Called Naive Realism. 5. So, assuming that we want to hold onto the idea that we have knowledge of the empirical world, by 3 and 4, we should accept disjunctivism.

3.1 First Objection and Response: Argument from Hallucination The aim of the present section is to assess the credentials of steps 1 and 2 in the above argument. This requires addressing the following two questions: ‘How does the Argument From Hallucination constitute a reductio against So-Called Naive Realism?’, and, ‘How does disjunctivism thwart the Argument From Hallucination?’ We’ll address the first question first. The Argument From Hallucination, as Martin understands it, has two steps: 1. one aimed at establishing what Martin calls the Common Kind Assumption, 2. another aimed at showing that the Common Kind Assumption is inconsistent with So-Called Naive Realism. The Common Kind Assumption is, for Martin, an assumption. His choice of terminology is designed to reflect his claim that there is no good argument for the Common Kind Assumption, and so the disjunctivist can feel free to refuse to accept it. The Common Kind Assumption is Martin’s way of formulating the conjunctivist claim that we must ac-

320

Thomas Lockhart

count for IND by reference to the notion of a TAP.13 Recall (from §2) that all IND requires us to hold is that Smith, as a matter of fact, cannot distinguish between a sufficiently impressive hallucination of a lemon and a veridical perception of one. As we saw, the conjunctivist is moved on this basis to conclude that we need to postulate the existence of a TAP to account for that which a case of a veridical and a non-veridical perception ‘have in common’. Rather than introduce something like our notion of a TAP, in his own account of the crucial misstep of the traditional epistemologist, Martin charges him instead with the error of postulating what he calls a ‘Common Kind’. We could try to perhaps offer a provisional notion of a common kind as follows: Common Kind Although Smith is currently having a veridical perception of a lemon, she would be having just the same kind of experience were she hallucinating a lemon.

Martin puts the conclusion that I have labelled ‘Common Kind’ this way: “whatever kind of mental, or more narrowly experiential, event occurs when one perceives, the very same kind of event could occur were one hallucinating”.14 With the Common Kind Assumption in hand, the second step in the Argument From Hallucination is to show that the Common Kind Assumption is inconsistent with SoCalled Naive Realism. Recall that according to So-Called Naive Realism, “[t]aking experiences to be episodes or events, … some such episodes have as constituents mind-independent objects”.15 The argument that we cannot hold both the Common Kind Assumption and SoCalled Naive Realism goes this way: suppose we “bring about a hallucinatory experience through suitable manipulation of brain and mind”.16 By the Common Kind Assumption, whatever kind of mental or experiential event occurs when one perceives also occurs in that hallucination. Since a hallucination does not constitutively involve a relation between a subject and a mind-independent object (it either involves a relation between the subject and a mind-dependent object, or no relation between the subject and an object at all) and hallucinations are the very 13 14 15 16

See above, §2. Martin (2006), 357. Ibid. Ibid., 358. That this is possible is a further assumption which I will not dispute. Martin guarantees it through what he calls Experiential Naturalism: “Our sense experiences are themselves part of the natural causal order, subject to broadly physical and psychological causes” (Martin (2006), 357).

Motivating Disjunctivism

321

same kind of things as veridical perceptions, it follows that veridical perceptions do not constitutively involve mind-independent objects. This contradicts So-Called Naive Realism. To see more clearly how Martin thinks this argument works, it helps to imagine a So-Called Naive Realist trying to argue that the Common Kind Assumption is not inconsistent with So-Called Naive Realism: According to the Common Kind Assumption, the kind of event that occurs when one has a veridical perception could occur were one instead to be having a hallucination: for example, whether she is having a veridical perception or not, Smith is having an experience as of a lemon in front of her (she is also, in both cases, having an apparent perceptual experience, and, in both cases, having a Tuesday-apparent-perceptual experience, and so on). But if she were having a veridical perceptual experience of a lemon, her experience would be one of those kinds of experiences which has, as component, a lemon (and thereby one of those kinds of experiences which has, as constituent, an external object). By contrast, if she were having a hallucination, she would not be having an experience which falls under those kinds.

To make sense of the Argument From Hallucination as Martin understands it, the Common Kind Assumption needs to be understood against the background of a particular metaphysical presupposition: Kind Presupposition There is a most specific answer to the question ‘What is it?’ asked of Smith’s veridical perception of a lemon, and this most specific answer is the fundamental kind of her veridical perception (call it K1). There is also a most specific kind to which an indistinguishable hallucination would belongs (call it K2).17

With the Kind Presupposition in hand, Martin goes on to elaborate what he calls the Common Kind Assumption. It can be put this way: CKA The fundamental kind to which a veridical perceptual experience belongs is the same fundamental kind to which an indistinguishable apparent perception belongs; that is, K1=K2.

Since the fundamental kind to which a mental episode belongs “tells us what essentially the event or episode is”, we can see more clearly why Martin thinks the Common Kind Assumption is inconsistent with So17 Martin puts the Kind Presupposition this way: “[E]ntities (both objects and events) can be classified by species and genus; for all such entities there is a most specific answer to the question, ‘What is it?’ In relation to the mental, and to perception in particular, I will assume that for mental episodes or states there is a unique answer to this question which gives its most specific kind; it tells us what essentially the event or episode is” (Martin (2006), 361).

322

Thomas Lockhart

Called Naive Realism. The defender of the Common Kind Assumption can concede that there might well be kinds to which veridical perceptions belong but indistinguishable hallucinations do not. But if the Common Kind Assumption is correct, veridical perceptions and indistingushable hallucinations belong to the same fundamental kind, and so share their essential features. Since it is not essential to a hallucination to involve a mind-independent object, it cannot be of the essence of a veridical perception to involve a mind-independent object. And this contradicts So-Called Naive Realism. It is also worth highlighting that for Martin, the Common Kind Assumption is the claim that a veridical perception and an indiscriminable hallucination belong to the same fundamental mental kind of event or episode. It is not a claim about the brain-state kinds which are involved in such experiences. Martin flags this with a further term-of-art: ‘phenomenal’. The Common Kind Assumption is a claim about the phenomenal, rather than, say, the physiological, kind to which veridical perceptions and hallucination belong.18 What has emerged is the following: the Argument From Hallucination, according to Martin, will only have teeth against So-Called Naive Realism if we take the Common Kind Assumption to amount to the claim that veridical perceptions and indistinguishable hallucinations belong to the same fundamental phenomenal kind. Disjunctivism, according to Martin, blocks the Argument From Hallucination by adumbrating a conception of perception according to which the move from IND to CKA would be an error. It does so by insisting that “veridical perceptual experience is of a distinct kind from hallucination”19, where, as we have seen, what this comes to is the following claim: veridical perceptual experience is of a distinct, fundamental, phenomenal kind from hallucination. So a disjunctivist, according to Martin, claims: (I) No instance of the specific fundamental phenomenal kind of experience I have now, when seeing the white picket fence for what it is, could occur were I not to perceive such a mind-independent object as this.20 18 See, for example, Martin (2006), 367. 19 Ibid., 354. 20 Martin’s actual formulation of this claim runs as follows: “No instance of the specific kind of experience I have now, when seeing the white picket fence for what it is, could occur were I not to perceive such a mind-independent object as this” (Martin (2006), 357). This formulation, however, as it stands, is sufficiently vague as to allow it to appear to coincide with something McDowell

Motivating Disjunctivism

323

Before we continue, it is worth noting that Martin’s account not only satisfies the constraints on a generically disjunctivist view adumbrated in §2, namely by rejecting the traditional epistemologist’s highly philosophically committal way of analysing what it is that veridical and nonveridical perceptions ‘have in common’ but he has offered his own alternative account of this analysis which turns on our willingness to buy into a great deal of additional metaphysics regarding the differences between fundamental and non-fundamental kinds, their respective relations to essences, and between phenomenal and non-phenomenal kinds; so that it now appears that a successful argument against the traditional epistemologist turns on the proper deployment of all these additional notions.

3.2 Second Objection and Response: The Explanation Objection We started from the fact that Martin thinks that in order to hold onto the idea that perception allows us to gain knowledge of the world, we will have to defend a conception of perception according to which veridical perceptions have as constituents mind-independent objects. He thinks that a defence of this claim will lead to a rejection of the Common Kind Assumption and an acceptance of (I). Since Martin is interested in defending So-Called Naive Realism from attack, he focuses on the objections which might be raised against it. So far we have seen him claim that in order to defend So-Called Naive Realism, we will not be able to accept any view according to which veridical perceptions and indistinguishable hallucinations are of fundamentally the same kind. This led Martin to formulate (I) as a central claim of disjunctivism. But Martin thinks that there is a powerful variant of the Argument From Hallucination which also casts doubt on So-Called Naive Realism. I dub this further objection the Explanation Objection. The stout disjunctivist defender of So-Called Naive Realism needs more than (I) to defeat the Explanation Objection. Once one begins to see how and why Martin feels the need to further strengthen disjunctivism in order to provide it with additional resources holds. It is only once we see that Martin is committed to an analysis of ‘the common kind’ here at issue in terms of kinds that are both fundamental and phenomenal in his sense that his divergence from McDowell can come into view.

324

Thomas Lockhart

for meeting the Explanation Objection, it becomes even more apparent how radically his disjunctivism differs from that of McDowell. For this reason, I will linger for a moment over the Explanation Objection and the materials Martin furnishes the disjunctivist to respond to it. Remember that, according to Martin, the Argument From Hallucination causes trouble for So-Called Naive Realism in virtue of an application of the Common Kind Assumption (CKA). To defend So-Called Naive Realism, Martin thinks the disjunctivist has to block CKA by accepting (I). The Explanation Objection causes trouble for So-Called Naive Realism by working with a variant of CKA which concedes (I). I’m going to give this variant of CKA—whose ultimate aim is again to make trouble for So-Called Naive Realism—a name: CKA Redux. Recall that the problem with CKA was that it claimed that veridical perceptions and indistinguishable hallucinations were of essentially the same kind—so veridical perceptions could not feature mind-independent objects as essential constituents. CKA Redux concedes that veridical perceptions have mind-independent objects as essential constituents, but threatens to deprive this feature of veridical perceptions of any explanatory significance. That is, the proponent of CKA Redux insists that although veridical perceptions and indistinguishable hallucinations belong to different most specific kinds, they both belong to a less specific common kind. And it is this common, less-specific kind which does all the explaining. The proponent of CKA Redux issues a challenge to the So-Called Naive Realist: Take some fact the explanation of which you think involves So-Called Naive Realist veridical perceptions. I’ll show you that the explanation would be just as good if, instead of appealing to veridical perceptions, we appealed to that kind to which an indistinguishable hallucination would belong. So, to the extent that we are concerned with generating adequate explanations, there is no need to invoke veridical perceptions, understood as the So-Called Naive Realist wishes to do.

It might help to lay this argument out step-by-step. We’ll start with: CKA Redux Although Smith’s veridical perception of a lemon belongs to a different most specific phenomenal kind than would an indistinguishable hallucination, nevertheless there is a less specific phenomenal kind (i) to which the veridical perception belongs, and (ii) to which the hallucination (were she to be having it instead) would belong.

Motivating Disjunctivism

325

The Explanation Objection starts with CKA Redux, which is consistent with (I).21 Let us introduce another kind of experience that Smith might have—a Seeming-Lemon-Seeing. This kind is supposed to meet the characterization introduced in CKA Redux: Smith’s hallucination of a lemon is of fundamental kind Seeming-Lemon-Seeing, and her veridical perception of a lemon is also of this kind.22 To see how the objection proceeds, imagine the following two scenarios: Situation 1 Sitting in a London pub, Smith has a veridical perception of a lemon—an experience of fundamental kind Lemon-Seeing. By So-Called Naive Realism, the lemon she sees is a constituent of her experience. It is summer, and seeing the lemon awakens in her a craving for a gin and tonic, so she orders one. Situation 2 Sitting in a London pub, Smith has a hallucination of a lemon, indistinguishable from a veridical perception of a lemon. Her experience is of fundamental kind Seeming-Lemon-Seeing, a kind different (by (I)) from Lemon-Seeing. It is summer, and apparently seeing the lemon awakens in her a craving for a gin and tonic, so she orders one.

According to the Explanation Objection, if the disjunctivist admits the existence of the kind Seeming-Lemon-Seeing which (a) embraces both veridical perceptions of lemons and lemon hallucinations and (b) explains Smith’s behaviour, then the disjunctivist will have thereby “made the kind of episode which is unique to the perceptual situation redundant in an account of consciousness and the mind”.23 We could dub this conclusion: Explanatory Redundancy Even if there were So-Called Naive Realist veridical perceptions, they would be explanatorily redundant in an account of consciousness and mind.

The argument goes this way: 21 I’m not here going to rehearse Martin’s reconstruction of an argument for CKA Redux (Martin (2003), 53 f.), for our purposes it’ll do merely to observe that (I) is consistent with CKA Redux. 22 We’ll have to think of a Seeming-Lemon-Seeing as different from an OP. From Martin’s point of view, the notion of an OP is formed by disjoining two kinds, and thus does not give the most specific answer to a ‘What is it?’ question. A Seeming-Lemon-Seeing is supposed to give the fundamental answer to the question: ‘What is it?’ asked of a lemon hallucination, although not the most specific answer to this question asked of a veridical perception of a lemon (see fn. 9). 23 Martin (2003), 61.

326

Thomas Lockhart

1. In both Situation 1 and Situation 2, we take Smith’s action of ordering the G&T to be a “causal (and possibly rational)” consequence of the experience she has had.24 2. In Situation 2, the explanation of Smith’s ordering-behaviour will derive from an appeal to the kind of experience she had—a SeemingLemon-Seeing.25 3. By CKA Redux, the kind of event Seeming-Lemon-Seeing also occurs in Situation 1 (as does the kind Lemon-Seeing). 4. Since the drink-ordering-behaviour is the same in Situations 1 and 2, and Smith’s having an experience of the kind Seeming-Lemon-Seeing explains her ordering-behaviour in Situation 2, and that very same kind of event occurs in Situation 1, it is natural to conclude that what explains her ordering-behaviour in Situation 1 must also be the fact that she is having an experience of kind Seeming-Lemon-Seeing.26 5. So the fact that Smith has a veridical perception of a lemon in Situation 1 is redundant in an explanation of her behaviour in that situation.27

Martin notes that this argument is not deductive—the conclusion is simply that it is far more natural to think that what does the explaining is the kind common to veridical perceptions and hallucinations. But Martin thinks that the onus is squarely on the disjunctivist to show why we should not conclude Explanatory Redundancy. That is, the disjunctivist who wants to admit CKA Redux must say something to defend his claim that “there is a distinctive role for only veridical perceptual experiences to play”.28 Martin puts the challenge for the disjunctivist this way: “[W]hat shows that what is relevant to the explanations we 24 Hallucinations and perceptions alike are “liable to coerce our beliefs and move us to action” and naturally enough we take these beliefs and actions to be “causal (and possibly rational) consequences” of those experiences (Martin (2003), 61). 25 “[I]n the case of hallucination, to the extent that these phenomena do have any explanation, that explanation will derive from appeal to the kind of experience that the subject is then having” (Martin (2003), 61). 26 “The phenomena which are in common between the hallucination and the perception are accompanied by a common kind of occurrence in both situations. So, the objector suggests, those phenomena will have a common explanation in the two situations, namely the occurrence of a kind of experience common to both perception and hallucination”, “the common kind of event between hallucination and perception seems better correlated with these common phenomena than the kind of event unique to perception and so seems to screen off the purely perceptual kind of event from giving us an explanation” (Martin (2003), 61 f.). 27 “[A]nd the kind of event which is unique to perceptual situations will be explanatorily redundant” (Martin (2003), 62). 28 Martin (2003), 63.

Motivating Disjunctivism

327

want to give is ever the kind of event peculiar to veridical perception rather than what is common to veridical perception and causally matching hallucination?”29 Martin thinks that the only satisfactory way to block the Explanation Objection is to deny CKA Redux and to insist that there is never a nontrivial kind to which both veridical perceptions and indiscriminable hallucinations belong.30 Thus he generates: (III) For certain visual experiences as of a white picket fence, namely causally matching hallucinations, there is no more to the phenomenal character of such experiences than that of being indiscriminable from corresponding visual perceptions of a white picket fence as what it is.31

29 Martin (2006), 369. 30 The argument that this is the only satisfactory response here is somewhat delicate. If the Explanation Objection is correct, then any reasonably ‘positive’ characterization of hallucinations will run foul of the objection. But clearly a veridical perception of a lemon and an indiscriminable hallucination of that lemon both belong to the kind Indiscriminable From a Veridical Perception of a Lemon. Furthermore, hallucinations sometimes have “causal (and possibly rational)” consequences, and so there are sometimes explanations we need to give which will involve an appeal to hallucinations (as, for example, in our Situation Two). Martin thinks that membership of this kind (Indiscriminable From a Veridical Perception of a Lemon) will be enough to make such explanations work (so it is in virtue of having an experience of the kind Indiscriminable From a Veridical Perception of a Lemon that Smith ordered a gin and tonic in Situation Two). But the fact that both veridical perceptions and indistinguishable hallucinations are members of this common kind does not, Martin argues, allow us to run the Explanation Objection. Essentially, this is because what makes the explanations involving hallucinations work is that these explanations are parasitic on the explanations which involve veridical perceptions (so an explanation of Smith’s behaviour in Situation One must make appeal to her veridical perceptual experience, and the explanation of her behaviour in Situation Two is only adequate in virtue of the goodness of the explanation of her behaviour in Situation One). That is, although we can explain her behaviour in Situation Two by appealing to the fact that her experience is of kind Indiscriminable From a Veridical Perception of a Lemon, this is only because we have to explain her behaviour in Situation One by appealing to the fact that her experience is of the kind Veridical Perception of a Lemon. See Martin (2006), 67 – 70. 31 Martin (2006), 369. I have termed (I) and (III) ‘(I)’ and ‘(III)’ respectively because Martin does. The progression of his argument from (I) to (III) goes via thesis (II). Both the details of this progression and of the intermediary thesis need not concern us here. In order not to needlessly exercise the reader’s curiosity, here it is: “(II) The notion of a visual experience of a white picket fence is

328

Thomas Lockhart

(III) stakes out a bold metaphysical claim about the nature of and the relationship between veridical perceptions and hallucinations. If Martin is right, all disjunctivists will need to defend the view that we can say nothing more to characterize hallucinations than to say that they are indiscriminable from veridical perceptions. This might seem like a difficult task, and disjunctivism has come under attack precisely because (III) has seemed to many not to leave us with sufficient resources for a satisfactory account of the relationship between veridical and non-veridical perceptions. But is Martin right that in order to defend So-Called Naive Realism, we will ultimately have to accept (III)? In particular, is defending (III) the only way that a disjunctivist could hold onto the idea that “there is a distinctive role for only veridical perceptual experiences to play”? I’ll argue below that McDowell’s is a disjunctivism which does allow us to insist that there is a distinctive role for only veridical perceptions to play without needing our to defend (III) or, for that matter, without our needing to commit ourselves to numerous collateral philosophical theses regarding the metaphysics of kinds (of the sort which, according to Martin, are required for the defence of disjunctivism). It is worth looking back on the territory we have just travelled. Disjunctivism, as it was first introduced in the generic characterization I offered in §2, originally looked to be a position that charges the traditional epistemologist with a philosophically overly-committal conception of what is required to make sense of the idea that veridical perceptions and non-veridical perceptions ‘have something in common’. Martin enters the scene as someone who purports to strengthen the credentials of disjunctivism. We have now seen that his way of purporting to do this places a considerable burden on the disjunctivist. Rather than merely shifting the burden onto the conjunctivist and querying whether he has left us with a satisfactory account of our perceptual capacities, according to Martin’s choreography of the dialectic between the conjunctivist and the disjunctivist, the disjunctivist must discharge two formidable philosophical obligations before he is in any position to really rebut the conjunctivist: (1) to adumbrate and defend his own metaphysics of phenomenal kinds, and, (2) to use this metaphysics of phenomenal kinds to articulate thesis (III)—his own substantive metaphysical account of that of a situation being indiscriminable through reflection from a veridical visual perception of a white picket fence as what it is” (Martin (2006), 363).

Motivating Disjunctivism

329

that which veridical and non-veridical perceptual experiences ‘have in common’. Much of the current philosophical literature on disjunctivism has now become sidetracked and bogged down in the debate about the prospects for mounting a defence of a version of (III). Prima facie, if there is an alternative sort of disjunctivism on offer which does not require of the disjunctivist that he first discharge these two prior dialectical obligations, it would seem to represent the cleaner and preferable philosophical option. Moreover, if there is such a variety of disjunctivism on offer, then Martin’s representation of the dialectic involves, in effect, a hijacking of the debate. In what follows, I will argue that McDowell’s disjunctivism is of the cleaner and preferable variety, and that it has been hijacked.

4. McDowell’s Disjunctivism We have just seen that Martin constructs an argument for disjunctivism which concludes by insisting that there can be no non-trivial common kind to which both veridical perceptions and indistinguishable hallucinations belong. McDowell—I contend—disagrees with Martin on precisely this point; indeed, according to McDowell, we will not be entitled to the idea that in a veridical perception a state of affairs makes itself manifest to the subject unless we admit that there can be a non-trivial account of that which veridical perceptions and indistinguishable hallucination ‘have in common’. We won’t see clearly why this is until §5.2. The task of the present section is to outline McDowell’s considerably less baroque argument for disjunctivism. McDowell puts the basic claim of disjunctivism this way: [P]erceptual appearances are either objective states of affairs making themselves manifest to subjects, or situations in which it is as if an objective state of affairs is making itself manifest to a subject, although that is not how things are.32

In McDowell’s terminology, the position opposed to disjunctivism is the highest common factor view, which instead maintains that what it is to be a perceptual appearance is to be whatever it is that is the highest common factor between a veridical perceptual experience and an indis-

32 McDowell (2008b), 231. Cp. OP in §2.

330

Thomas Lockhart

tinguishable hallucination.33 Recall that in §2 we noticed that it is natural to want to introduce some idea or other of an ‘apparent perception’ to account for IND. We went on, however, to distinguish three different ways to define an ‘apparent perception’: a TAP, a MAP, and an OP. What McDowell calls a ‘highest common factor’ is what I have called a TAP in §2: it is a term meant to capture in as uncomplicated a way as possible what is philosophically suspect in the traditional epistemologist’s concept of an apparent perception. Any excessive complication introduced at this point in the characterization of that concept threatens to distract us from what is hopeless about philosophical attempts to solve the TEE. McDowell himself employs the terminology of ‘perceptual appearance’ in an artfully equivocal fashion: sometimes to mean a TAP, and sometimes to mean an OP. McDowell’s point is that all a philosophically unprejudiced account of the datum of IND can entitle us to employ is ‘perceptual appearanceOP’ and that ‘perceptual appearanceTAP’ is both insufficiently motivated and epistemologically disastrous. For McDowell, disjunctivism allows us to see our way to a satisfactory conception of what it is to know something on the basis of perception. The key idea here is that the concept of perception is a concept of a way of knowing—that is, a concept of a cognitive capacity. The problem with conjunctivism according to McDowell is that it ends up rendering this truistic thought unintelligible. The aim of McDowell’s disjunctivism is to restore to us our grip on this truism. This contrasts starkly with the aim of Martin’s disjunctivism, which is to convince us that a vindication of the thesis of Naive Realism requires a sort of philosophical heavy lifting via an account of the metaphysics of kinds and a defence of claim (III). McDowell does say some things that can sound like things that Martin says. For example, McDowell suggests that a satisfactory account of what it is to come to know something on the basis of perception will require us to be able to hold onto the idea that in perception “we 33 Martin thinks that what McDowell calls the highest common factor conception of a perceptual appearance is just the claim that a veridical perception and an indistinguishable hallucination belong to the same fundamental phenomenal kind (i. e., the Common Kind Assumption (as defined in §3.1)). Martin claims this equivalence explicitly at his Martin (2006), 357. It should, however, already be clear that Martin’s CKA involves a different and more elaborate analysis of the notion than that provided above in our gloss of what is involved in an apparent perception being a TAP.

Motivating Disjunctivism

331

can have environmental facts directly available to us”.34 This can sound a lot like one of Martin’s less philosophically burdened formulations of Naive Realism: “veridical perception is a relation of awareness to mind-independent objects”. It is not part of my point to argue that McDowell would necessarily be concerned to disown such a form of words, nor to deny that his broader philosophical project is meant to allow us to maintain a conception of perception that is consistent with what that form of words might be taken to say on a sufficiently philosophically innocuous construal of it. But it is crucial for me to deny that McDowell and Martin agree on the question of how much is required of the philosopher in order to be able to restore our hold on the bare idea that in perception “we can have environmental facts directly available to us”. Although both Martin and McDowell think that the chief benefit of disjunctivism is to allow us to defend a satisfactory conception of what it is to know something on the basis of perception, they differ in their respective accounts of the relationship between disjunctivism and Cartesian scepticism. On McDowell’s account, the relationship between them is considerably more intimate than on Martin’s.35 For McDowell, disjunctivism simply amounts to the denial of what I have called conjunctivism above, where conjunctivism, on his account, is to be seen as immediately committing us to a Cartesian conception of the mind—that is, a conception of the mind according to which any exercise of our perceptual capacities necessarily falls short of achieving cognitive contact with the state of affairs which such an exercise purports to put us in touch with. For McDowell, conjunctivism therefore threatens to force us to conclude that we cannot come to know anything on the basis of perception.36 34 McDowell (2008b), 229. 35 Martin does think that disjunctivism is in tension with Cartesian scepticism, but the tension is one of some kind of intellectual aporia. See Martin (2006), §10. 36 Notice that for McDowell, it is the conjunctivist claim about what veridical perceptions are which threatens to deprive us of the idea that we can know about the world on the basis of perception; and it will only be by finding an alternative account of the nature of veridical perception that we can restore this ‘epistemological’ truism. So this epistemological truism is neither the only claim of McDowell’s disjunctivism (although vindicating the epistemological truism is part of what motivates it), nor does it separate McDowell from Martin – since, as we’ve seen, Martin too is concerned to vindicate some version of the same epistemological truism.

332

Thomas Lockhart

4.1 McDowell’s Disjunctivism and the Cartesian Sceptic McDowell’s basic strategy is to argue as follows: 1. Traditional epistemology relies on the highest common factor conception of a perceptual appearance (or, a TAP), 2. This forces on us a conjunctivist (i. e. a Cartesian) account of what is involved in veridical perception, leaving us with the task of having to solve the TEE, 3. Solving the TEE in a way that preserves the idea that veridical perception allows us to know about the world is doomed to failure, 4. The disjunctivist schema (OP) shows that we need not adopt the highest common factor conception of experience, 5. Therefore, we can reject the task of solving the TEE and thus need not be troubled by Cartesian scepticism. Notice how very different is the role which disjunctivism plays in the above argument against the traditional epistemologist from the role it plays in Martin’s argument for Naive Realism. This difference between them becomes invisible if we attempt to reduce disjunctivism to a two sentence statement of a ‘philosophical position’. At the latter level of description, it is extremely difficult to see how different Martin’s and McDowell’s approaches to their shared topic really are from each other. What we need to appreciate, in order to see this, is the very different dialectical contexts in which their respective employments of the disjunctivist schema of §2 are embedded. This difference amounts, in a significant sense, to a difference in what disjunctivism is for each of them. For McDowell, Cartesian scepticism is the inevitable result of analyzing what a veridical and a non-veridical perception ‘have in common’ in terms of a TAP—that is, Cartesian scepticism is, essentially, a way of drawing attention to the impossibility of solving TEE. We can better appreciate the role that disjunctivism plays in McDowell’s philosophy by seeing why this is the case. The Cartesian sceptical argument, according to McDowell asks us to “Consider a situation in which a subject seems to see that, say, there is a red cube in front of her. The idea is that even if we focus on the best possible case, her experience could be just as it is, in all respects, even if there were no red cube in front of her”.37 He comments that 37 McDowell (2008b), 228.

Motivating Disjunctivism

333

The familiar sceptical scenarios—Descartes’s demon, the scientist with our brains in his vat, the suggestion that all our apparent experience might be a dream—are only ways to make this supposed predicament vivid.38

The next step in the Cartesian sceptical argument—as reconstructed by McDowell—is to conclude from this observation that we never have a better reason to believe that things are as experience presents them to be than we would have in the case of a hallucination. So on McDowell’s reconstruction, the standard Cartesian sceptical argument has two steps: 1. To ‘make vivid’ the predicament posed by highest common factor conception of apparent perception; 2. To draw from this predicament the conclusion that we never have good reason to believe that the external world is as our perceptions seem to reveal it to be. Let us grant, for a moment, that McDowell is right about 1, that is, that the Cartesian sceptic must begin by presupposing the highest common factor conception of perception. What about 2? That is, how does the sceptical conclusion follow from the highest common factor conception of apparent perception? Recall that, according to the highest common factor conception, not only must we hold that a veridical and a non-veridical perceptual experience ‘have something in common’, but we must analyze that which they ‘have in common’ literally in terms of their having some thing in common, that is, in terms of the concept of a TAP. So, if that highest common factor conception is correct, then there is nothing about perceptual experience considered as such to provide me with more than inconclusive reasons for thinking that the external world is a certain way. This is because my perceptual experience would be, as such, exactly the same even if the external world were not as my experience seems to reveal it to be. The best epistemic standing which we can achieve in virtue of our perceptual experience is not yet enough to licence any beliefs about the external world. If 2 is right, what of 1? Is it really the case that the Cartesian sceptic must start by, in essence, insisting on the highest common factor conception of apparant perception? At first glance, it seems quite plausible to think that McDowell is just wrong about this—surely the familiar Cartesian sceptical arguments don’t require anything so strong as a 38 Ibid.

334

Thomas Lockhart

claim about the nature of experience? 39 We might try to defend a reading of Cartesian scepticism which does not presuppose the highest common factor conception of apparent perception this way: The sceptic urges that Smith cannot tell, by mere reflection on her experience, whether she is in the good case or a bad case. For all she knows, she might always be in a bad case. So really, she should not have any confidence that she knows any of the things she takes herself to have come to know through perceptual experience. But this reconstruction of the sceptical argument does not depend on the highest common factor conception of apparent perception. Indeed, the sceptic, as far as I can see, could concede disjunctivism and thereby acknowledge that the good cases are constitutively linked to the states of the world they represent, and the bad cases aren’t. The sceptical point is merely that Smith can never be sure which case she is in, and so she shouldn’t be confident that she knows anything on the basis of perception.

Let us give this kind of Cartesian sceptic—the sceptic who concedes disjunctivism (and denies the highest common factor conception of experience) but still attempts to run a sceptical argument—a name. Call him a weak Cartesian sceptic. And let us call the kind of sceptic McDowell is interested in—the kind who relies on the highest common factor conception of apparent perception—a strong Cartesian sceptic. McDowell’s claim is, in effect, that the Catesian sceptical argument is only interesting in so far as it is a version of strong Cartesian scepticism. We need not bother with the weak Cartesian sceptic. But why? To see why McDowell dismisses weak Cartesian scepticism, we need a further premise. McDowell thinks that if we adopt the disjunctive conception of experience, we will thereby have admitted the intelligibility of the following: that some perceptual experiences, considered as such, are cases of an objective state of affairs making itself manifest to a subject. But to admit this, thinks McDowell, is to acknowledge that the epistemic standing of an agent in the good case is different from the epistemic standing of an agent in the bad case. That is, suppose we acknowledge the conceivability of a perceptual experience in which an objective state of affairs makes itself manifest to a subject, where this is a different kind of perceptual experience from those perceptual experiences which merely seem to be cases of an objective state of affairs making itself manifest. McDowell thinks that the two cases correspond to two possible ‘standings in the space of reasons’. Although in a sufficiently vivid version of the bad case (in which she is having a non-ve39 Martin, for one, insists on this in his Martin (2006), §10.

Motivating Disjunctivism

335

ridical perceptual experience) Smith herself wouldn’t know that she were having a non-veridical perceptual experience, in the good case (where her exercise of her perceptual capacities afford her a way of knowing) she is justified in believing that there is a lemon in front of her. The bad case, as specified in the preceding sentence, is too underdescribed to determine what her degree of epistemic responsibility is in this failure to know. If she failed to exercise due care or otherwise undermined her capacity to do so (e. g. by taking hallucinogenic mushrooms) then she is not blameless for coming to believe that things are as they perceptually seem to her to be. If she is a victim of bad epistemic luck (e. g. confronted with a compelling perceptual illusion) then she may be blameless for coming to believe that the world is as her perception presents it as being. Neither of these cases, however, impugn her claim to know when she is responsibly exercising her perceptual capacities under normal conditions. We can give the underlying idea here a name: Knowledge Premise When a perception constitutively involves an objective state of affairs making itself manifest to a subject, that subject has a different epistemic standing than in the case—potentially indiscriminable to the subject—when her perception does not constitutively involve an state of affairs making itself manifest.

The Knowledge Premise has seemed contestable and has been contested by many. My point here is simply that the Knowledge Premise plays a crucial role in McDowell’s case for his disjunctive conclusion, but is not to be identified with the conclusion itself. A proper understanding of what McDowell takes to be involved in committing oneself to the Knowledge Premise should reveal, however, that it is simply a further condition on one’s being able to retain one’s grip on the idea that perception is a capacity to come to know about the world, since any adequate account of a capacity will involves a certain form of asymmetry in how successful and unsuccessful exercises of the capacity are related to one another. That is, if we are to retain the idea that perception is a capacity to come to know about the world, then we have to acknowledge that it “is a capacity—of course fallible—to get into positions in which one has indefeasible warrant for certain beliefs”.40 This is because, McDowell claims, it simply makes no sense to (a) maintain that perception is a capacity to know about the 40 McDowell (2010), 245 (in this volume, 293).

336

Thomas Lockhart

world and (b) deny that in “non-defective exercises”41 of that capacity we can have indefeasible warrants for certain beliefs. McDowell defends this claim by contrasting a view which maintains (a) and (b) about our perceptual capacities with a parallel claim about another capacity: “Some people have a capacity to throw a basketball through the hoop from the free-throw line”.42 To hold both (a) and (b) would be tantamount to thinking that “there cannot be a capacity—of course not guaranteed success on all occasions—in whose non-defective exercises one actually makes free throws”.43 I introduced the Knowledge Premise as part of an attempt to explain why McDowell thinks we can dismiss weak Cartesian scepticism. Suppose we grant McDowell the Knowledge Premise. How can he now respond to weak Cartesian scepticism? We can think of the response as itself having two steps. First, with the Knowledge Premise in hand, McDowell can point out that in the good case—since the good case is simply a case of the world presenting itself to Smith as it is—Smith is entitled to her knowledge claim. If the animal in front of me is a zebra, and conditions are suitable for exercising my ability to recognize zebras when I see them (for instance, the animal is in full view), then that ability, fallible though it is, enables me to see that it is a zebra, and to know that I do.44

To put it another way: weak Cartesian scepticism, in conceding disjunctivism, concedes the intelligibility of the idea that Smith can be justified in believing that there is a lemon in front of her in virtue of having an experience which presents that lemon to her. So the first step in McDowell’s response to weak Cartesian scepticism is this: in conceding disjunctivism, the weak Cartesian sceptic concedes that if Smith has a veridical perception of a lemon, then in virtue of the fact that that experience directly presents a lemon to her, she is justified in believing that there is a lemon there. The second step in McDowell’s way with the weak Cartesian sceptic is this: Nothing the weak Cartesian sceptic has said forces us to give up the idea that most of the time I am justified in believing that things are thus and so in virtue of the fact that I can open my eyes and see that things are thus and so. As McDowell puts it, if scepticism concedes disjunctivism, we can 41 42 43 44

Ibid., 246 (in this volume, 293). Ibid. Ibid. McDowell (2008b), 239.

Motivating Disjunctivism

337

simply “appeal to ordinary cases of perceptual knowledge in ruling out the sceptical scenarios”.45 So McDowell’s strategy here is this: we only need bother with a sceptical argument if the sceptic can put us in a position where we have to abandon something which commonsense tells us to be the case. If scepticism concedes disjunctivism, it thereby concedes the intelligibility of Smith’s being justified in believing that she sees a lemon in virtue of her seeing a lemon. And once this possibility is intelligible, then the sceptic no longer has the resources to force us to abandon our commonsense knowledge claims. Therefore, for McDowell, we are not compelled to engage with the sceptic. To put the point the other way around: if the sceptic can do no more than to draw attention to our fallibility, then there is no need to engage with him. We only need engage with the sceptic if he can force an engagement by presenting us with an argument which leaves us with no choice but to abandon the commonsense idea that I can know many things on the basis of perception. The weak Cartesian sceptic cannot do this, but the strong Cartesian sceptic does: by relying on the notion of a TAP, he introduces a conception of perception according to which Smith’s epistemic standing is identical in both good and bad cases. To summarize: the weak Cartesian sceptic admits that it is intelligible for there to be perceptual experiences which directly present the external world. In this case, McDowell urges that we need not feel compelled to engage with scepticism. Weak Cartesian scepticism, in effect, does no more than draw attention to our fallibility, and we can instead rely on our common sense and rule out the sceptical scenarios. We only need engage with the sceptic if he can rationally compel an engagement by forcing us into a position where we can no longer rely on common sense. Cartesian scepticism is only compelling, argues McDowell, if it forces us to conclude that we never have a better epistemic warrant for believing the world to be as it seems than the warrant we would have were we the victims of a hallucination. And this can only be done by foisting on us the highest common factor conception of perceptual appearances (that is, by insisting that we analyze what it is that a veridical and a non-veridical perception ‘have in common’ in terms of a TAP). So far then, McDowell’s claim is this: only strong Cartesian scepticism forces us to abandon the idea that we can ever know that the world 45 Ibid., 237.

338

Thomas Lockhart

is thus-and-so on the basis of perception. But it can do this only because it relies on the highest common factor conception of perceptual appearances (that is, by insisting that we analyze a perception in terms of a TAP). Conversely, McDowell thinks that by accepting the traditional epistemologist’s conception of an apparent perception, Cartesian scepticism is the inevitable result: for veridical experience, as such, will never give us better warrant for believing the world to be thus-and-so than would an indistinguishable hallucination. Since (strong) Cartesian scepticism is the inevitable result of accepting the analysis of an apparent perception in terms of a TAP, the project of solving TEE is doomed to failure. If, instead, we acknowledge that it is so much as intelligible that some perceptions are cases of the world making itself manifest, then we need not be troubled by Cartesian scepticism, and we need not abandon the idea that we can come to know that the world is thusand-so on the basis of perception. It is precisely the intelligibility of this conception of experience which is brought into view by the disjunctivist. 4.2 The Intelligibility of Disjunctivism For McDowell, Cartesian scepticism is an expression of what happens if we allow ourselves to be talked into analyzing what veridical and nonveridical perceptions ‘have in common’ in terms of a TAP. In order to avoid the difficulties of Cartesian scepticism and TEE, we need to remove the key prop required by the Cartesian sceptic. According to McDowell, the key prop of Cartesian scepticism is accepting the traditional epistemologist’s notion of a TAP. Disjunctivism knocks out this prop by guarantee the mere intelligibility of the idea of a perceptual experience in which the world makes itself manifest. And (still according to McDowell) the disjunctivist analysis of perceptual experience is, in turn, secured by something the Cartesian sceptic will not deny: that both veridical and non-veridical perceptions seem to present the world to be thus-and-so. The disjunctivist analysis is ‘transcendentally’ required in order to understand the very idea that experiences ‘purport’ to present the world as being a certain way.

Motivating Disjunctivism

339

McDowell claims to find an appropriate transcendental argument in Sellars.46 Sellars starts with what he calls a “simple but fundamental”47 observation: the sense of ‘red’ in which things look red is, on the fact of it, the same as that in which things are red. When one glimpses an object and decides that it looks red (to me, now, from here) and wonders whether it really is red, one is surely wondering whether the color—red—which it looks to have is the one it really does have.48

That is, “being red is logically prior, is a logically simpler notion, than looking red”49 Sellars cashes this by saying that whilst “This necktie is green” has a straightforward reporting use, “This necktie looks green” is used when the speaker is reluctant to endorse the report which she would issue if she said “This necktie is green”. On Sellars’ account, ‘is green’ is a straightforward report; but ‘looks green’ has both a ‘reporting’ and an ‘inferential’ component.50 “[W]hen I say ‘x looks green to me now’ I am reporting the fact that my experience is, so to speak, intrinsically, as an experience, indistinguishable from a veridical one of seeing that x is green”51 but I use that locution because “I may have reason to think that x may not after all be green”.52 So suppose Smith says sincerely ‘x looks green to me’. Then either Smith sees that x is green and is unwilling to endorse that claim or Smith is having an experience which is, from her point of view, exactly like the experience she would have if she were seeing that x is green, but 46 This is how McDowell glosses Sellars: “In order to find it intelligible that experience has objective purport at all, we must be able to make sense of an epistemically distinguished class of experiences, those in which (staying with the visual case) one sees how things are—those in which how things are makes itself visually available to one” (McDowell (2008b), 230). 47 Sellars (1997), §12. 48 Ibid. 49 Ibid. 50 This claim arises as Sellars imagines how looks-talk might have come to be useful to a community which lacks it. He puts the question this way: what is John the tie salesman (who has not yet learned looks-talk) to say when confronted with a tie under suboptimal lighting conditions? His (i. e. John’s) natural response is this: “I don’t know what to say. if I didn’t know that the tie is blue—and the alternative to granting this is odd indeed—I would swear that I was seeing a green tie and seeing that it is green. It is as though I were seeing the necktie to be green” (Sellars (1997), §14). 51 Sellars (1997), §16. 52 Ibid.

340

Thomas Lockhart

x isn’t green, and Smith is unwilling to endorse the claim that x is green. McDowell summarizes the point thus: In order to understand the very idea of the objective purport of visual experience (to single out one sensory modality), we need to appreciate that the concept of experiences in which, say, it looks to one as if there is a red cube in front of one divides into the concept of cases in which one sees that there is a red cube in front of one and the concept of cases in which it merely looks to one as if there is a red cube in front of one.53

Thus, for McDowell, the basic claim of disjunctivism follows from a consideration of an idea that no party to the dispute denies: that veridical and non-veridical perceptions alike seem to put us in contact with the world.

5 The Incompatibility of Martin and McDowell In this section, I argue that the fundamental incompatibility between McDowell’s disjunctivism and that of Martin is this: for McDowell, disjunctivism not only can, but must, leave open the possibility that there is a non-trivial account of that which veridical and non-veridical perceptions ‘have in common’. Martin cannot admit the existence of such a non-trivial account. In the first subsection, I argue that Martin’s form of disjunctivism is incompatible with any theory of perception according to which perceptions themselves have content. In the second subsection, I argue that McDowell’s disjunctivism is, and must be, compatible with such views.

5.1 Martin’s Disjunctivism and the Content of Perception McDowell, we have seen, maintains that Cartesian scepticism threatens to render unintelligible something which seems commonplace: that on the basis of perceptions which reveal the world to be thus-and-so we can come to know that the world is thus-and-so. He claims that in order to disarm the Cartesian sceptic, it will be enough to show that the mere intelligibility of this commonplace is guaranteed by something that the Cartesian sceptic takes for granted: that veridical and non-veridical perceptions alike seem to present the world to be a certain way. 53 McDowell (2008b), 230.

Motivating Disjunctivism

341

Focussing on the idea that the Cartesian sceptic already admits that veridical and non-veridical perceptions alike both purport to present the world to be a certain way is, from Martin’s point of view, dangerous. For Martin, it is critical that we do not embrace a conception of perception according to which perceptions themselves have any form of representational content. There are various ways that philosophers have maintained that perceptions themselves have content. Suppose, for example, we have a view according to which veridical perceptions draw on conceptual capacities. 54 Once we admit that a veridical perception draws on conceptual capacities, it becomes natural enough to say that an indistinguishable hallucination draws on precisely the same conceptual capacities. That is, Smith’s veridical perception of a lemon and her hallucination of a lemon both somehow draw on her lemon concept. But to say that both Smith’s veridical perception of a lemon and her hallucination of a lemon draw on her lemon concept is, from Martin’s point of view, a significant enough characterization of what they ‘have in common’ to fall prey to the Explanation Objection. That is, experience which draws on Smith’s concept lemon is a ‘positive’55 characterization of a ‘common kind’ under which both Smith’s veridical perception of a lemon and an indistinguishable hallucination she might have would fall. It is a positive enough common kind that it will be overwhelmingly natural to conclude that we must explain Smith’s gin-and-tonic-ordering behaviour in both Situation One and Situation Two (above, §3.2) by appealing to the fact that her experience is of this kind—and so make her veridical perception of a lemon explanatorily redundant. To admit that Smith’s veridical perception of a lemon and her hallucination of a lemon both draw on her concept lemon is thus to fall foul of Martin’s third tenet of disjunctivism: ‘For certain visual experiences as of a white picket fence, namely causally matching hallucinations, there is no more to the phenomenal character of such experiences than that of being indiscriminable from corresponding visual perceptions of a white picket fence as what it is.’56 Martin’s disjunctivist 54 On such a view, veridical perceptions won’t be full-blooded representations. This view is the one currently favoured by McDowell—although not much will turn on exactly how we formulate the point. 55 See fn. 17 for this terminology. 56 A Martinian disjunctivist might try to salvage the idea that veridical perceptions draw on conceptual capacities without admitting that there is a significant common kind shared by veridical perceptions and indiscriminable hallucinations. We might, for example, claim that although veridical perceptions draw on con-

342

Thomas Lockhart

must insist that veridical and non-veridical perceptions ‘have something in common’ only in the most trivial sense.57 For this reason, those who have wanted to take Martin’s form of disjunctivism seriously have insisted that there can be no sense in which either veridical perceptions or indiscriminable hallucinations draw on conceptual capacities at all: perceptions must themselves be totally non-conceptual.58 Furthermore, Martin’s disjunctivism has been criticized precisely for this reason—because it does not allow us to maintain the idea that perceptions themselves have content.59 Call this the objection from perceptual content.60 In the next section, we’ll see that McDowell’s version of disjunctivism is not subject to this objection.

5.2 McDowell’s Disjunctivism and the Content of Perception McDowell’s disjunctivism is, for all I have said so far, consistent with the claim that there might be a non-trivial way of characterizing what it is that a veridical perception and an indistinguishable hallucination ‘have in common’. In this section, I want to urge that this is a crucial feature of McDowell’s account. I’ve made the case that McDowell’s disjunctivism emerges as part of his project to defend a satisfactory account of how we could come to know about the world on the basis of perception. As a further component of that very same project, McDowell is also, notoriously, committed to the view that “capacities that belong to reason … be operative in experiencing itself”.61

57 58 59 60

61

ceptual capacities, indiscriminable hallucinations merely seem to the subject to draw on those same conceptual capacities. This strikes me as a difficult and somewhat ad hoc position to take: what is the difference between actually drawing on a conceptual capacity and merely seeming to the subject to draw on a conceptual capacity? See fn. 17. See, for example, Travis (2005). See, for example, Byrne/Logue (2008). Another objection which has been brought against Martin’s view is that it does not leave sufficient resources to give a robust account of hallucinations. See Burge (2005), Fish (2008), Brewer (2008) and Siegel (2008). If my argument below is correct, it will also follow that McDowell’s version of disjunctivism won’t fall prey to this criticism. McDowell (2008a), 258.

Motivating Disjunctivism

343

I won’t now rehearse the reasons that McDowell has for thinking that an adequate account of perception requires the idea that perceptual experience itself draws on conceptual capacities. For my purposes, it will suffice to observe that, for McDowell, a satisfactory account of how perceptual experience can put us in touch with the world must maintain both: (a) the idea that perceptual experiences involve actualizations of perceptual capacities and (b) disjunctivism. If Martin’s were the only form of disjunctivism available, then McDowell could not hold both (a) and (b) together. McDowell, from Martin’s point of view, must abandon one of his commitments. One response here is to abandon the idea that Martin and McDowell both attempt to defend the position I outlined in §2. This would be an overreaction. McDowell can consistently maintain both (a) and (b) because Martin is simply wrong to conclude that they are incompatible: in particular, Martin is wrong to think that defending disjunctivism requires insisting that the only sense in which veridical and non-veridical perceptions ‘have something in common’ is trivial. McDowell defends a conception of perception according to which there are non-trivial features shared by veridical and non-veridical perceptions62—in the case of Smith sitting in the London pub, both the veridical perception she in fact has and the indistinguishable hallucination she might have both draw on her concept lemon. This, I contended in §5.1, is sufficient to violate Martin’s condition (III) on disjunctivism. Given our work in §3.2, we are now in a position to see exactly why Martin thinks the disjunctivist should not admit that Smith’s veridical perception and her indiscriminable hallucination share this feature. To do so is to make So-Called Naive Realist veridical perceptions explanatorily redundant in an account of consciousness and mind. Rehearsing the reasoning of §3.2, appropriately modified for the case at hand: the explanation of Smith’s ordering a G&T in the situation where she merely hallucinates a lemon will appeal to the kind of experience she is having—that kind of experience which draws on the concept lemon. But since she is also having an experience of that kind when she actually perceives a lemon, it is, Martin contends, overwhelmingly natural to conclude that when we are called upon to explain her ordering of a gin and tonic in the case where she has had a veridical perception of a lemon, it is this fact about her experience (that her experience is merely one which involves the concept lemon) which does the explanatory work 62 He is insistent about this in McDowell (2010).

344

Thomas Lockhart

(rather than the fact that her experience is a veridical perception of a lemon). We are also now in a position to see how McDowell can reject this line of thought. We’ve seen that what I called the Knowledge Premise plays a critical role in McDowell’s disjunctivism.63 According to the Knowledge Premise, a subject who is having a veridical perception has a different epistemic standing than a subject who merely seems to be having a veridical perception. In Situation 1, Smith is justified in believing that she sees a lemon. In Situation 2, although Smith believes she sees a lemon, she is not justified in this belief. Let us suppose then that in both Situation 1 and Situation 2, Smith comes to believe that she sees a lemon. Suppose further that in both cases, her experience draws on her concept lemon. How can McDowell explain why she forms the belief she does? In Situation 1, Smith believes that she sees a lemon because she sees a lemon. In Situation 2, Smith believes that she sees a lemon merely because she takes herself to see a lemon. Although she has the same belief in the two cases, our explanation of her coming to form that belief is different in the two cases because her epistemic standing is different (in virtue of the fact that her experiences are different): in the one case, we explain her belief by appealing to the fact that her belief is justified by an experience which presents to her a lemon, in the second case we explain her belief by pointing to the fact that she takes her belief to be justified, in virtue of an experience which merely seems to present a lemon to her. The fact that a difference in justificatory standing leads to a difference in how we explain Smith’s patterns of belief formation will also allow us to explain the differences in Smith’s behaviour in the two cases. In Situation 1, Smith ordered a gin and tonic because she saw a lemon and this led her to crave that drink. In Situation 2, Smith ordered a gin and tonic because she took herself to see a lemon, and that led her to crave the drink. Martin issued a challenge for the disjunctivist who wishes to maintain that there can be a non-trivial characterization of what veridical and non-veridical perceptions ‘have in common’. The challenge was this: “[W]hat shows that what is relevant to the explanations we want to give is ever the kind of event peculiar to veridical perception rather than what is common to veridical perception and causally matching hal63 Although I urged that we should not identify it with McDowell’s disjunctivism.

Motivating Disjunctivism

345

lucination?”64 Although it isn’t clear that Martin takes any actual philosophers to fall into the range of his challenge, I’ve argued that McDowell is the kind of disjunctivist at whom Martin’s challenge can be directed. But McDowell can answer the challenge: in virtue of the different experiences they are, veridical perceptions have a different justificatory standing than non-veridical perceptions, and therefore they have a different explanatory status in our account of consciousness and mind.

6 Conclusion In §2, I offered a conception of disjunctivism which I take to be neutral between Martin and McDowell. In §3 and §4, I outlined the different ways that Martin and McDowell implement this basic disjunctivist project. If my reconstruction is correct, it is both wrong to think (as some have) that Martin and McDowell are simply engaged in the same project, but also wrong to think (as some have) that they are engaged in essentially different projects. In §5, I drew attention to a specific way in which the disjunctivisms of Martin and McDowell are incompatible. Martin thinks that the disjunctivist will be required to defend the view that there can be no non-trivial way of characterizing what it is that a veridical and a non-veridical perception ‘have in common’. So Martin’s form of disjunctivism rules out any account according to which perceptions themselves involve representational or conceptual content. If Martin is right, no philosopher who thinks that perceptions have representational or conceptual content can be a disjunctivist. Martin is wrong about this conclusion. McDowell is a disjunctivist who maintains that perceptions ‘draw on’ conceptual capacities. I’ve argued that McDowell can hold this view without inconsistency. McDowell avoids inconsistency because he can respond to a rhetorical question Martin poses. Martin thinks that any disjunctivist who wants to admit that veridical and non-veridical perceptions can, non-trivially, ‘have something in common’ will face the unenviable task of explaining how disjunctivist-approved veridical perceptions do not become explanatorily redundant. McDowell can respond to this challenge: even if there is a significant way of characterizing what a world-involving veridical perception and an indistinguishable hallucination ‘have in com64 Martin (2006), 369.

346

Thomas Lockhart

mon’, the veridical perception will not be explanatorily redundant because it will have very different justificatory force. Many have been persuaded by Martin’s claim that disjunctivism cannot admit a non-trivial characterization of what veridical and non-veridical perceptions ‘have in common’, and some have criticized disjunctivism in the light of this claim. I’ve argued that McDowell’s is a species of disjunctivism which is not subject to these criticisms. But the contemporary literature on disjunctivism has tended to focus on Martin’s very distinctive understanding of what veridical and non-veridical perceptions have ‘in common’, and the metaphysical presuppositions which this claim requires. Martin’s bold metaphysical claims have thus hijacked the debate about disjunctivism. I’ve shown that Martin’s is only one way to implement the generic disjunctivist project of §2, and when seen in this way, it is clear that his claims about what veridical and non-veridical perceptions ‘have in common’ are—far from representing the core of disjunctivism—only a curious feature of one of its species. The point of this paper is to arrive at a vantage point from which it is evident that McDowell’s disjunctivism is equally a way to implement the project of §2, but one which avoids the difficult and attention-grabbing metaphysical flourishes of Martin’s species. From this vantage point, McDowell’s species of disjunctivism emerges as a far more appealing instantiation of the genus.65

References Brewer (2008): Bill Brewer, “How to Account for Illusion”, in: Haddock/ Macpherson (2008a), 168 – 180. Burge (2005): Tyler Burge, “Disjunctivism and Perceptual Psychology”, in: Philosophical Topics 33, 1 – 78. Byrne/Logue (2008): Alex Byrne/Heather Logue, “Either/Or”, in: Haddock/ Macpherson (2008a), 57 – 94. Fish (2008): William Fish, “Disjunctivism, Indistinguishability, and the Nature of Hallucination”, in: Haddock/Macpherson (2008a), 144 – 167. 65 This paper grew out of a series of discussions in the Philosophy of Mind Workshop at the University of Chicago. Thanks to the participants in that workshop, especially Stina Bäckström, Rachel Goodman, and Aidan Gray. Thanks also to the participants in the Wittgenstein Workshop at the University of Chicago, who listened to a version. James Conant and David Finkelstein provided tireless support in the final stages of this paper, and I owe a great deal to both of them. Finally, enormous thanks to Jennifer Lockhart, without whose tireless efforts in support of Cartesian scepticism this paper would not exist.

Motivating Disjunctivism

347

Haddock/Macpherson (2008a): Adrian Haddock/Fiona Macpherson (eds.), Disjunctivism: Perception, Action, Knowledge, Oxford. Haddock/Macpherson (2008b): Adrian Haddock/Fiona Macpherson, “Introduction: Varieties of Disjunctivism”, in: Haddock/Macpherson (2008a), 1 – 24. Martin (2003): Mike G. F. Martin, “The Limits of Self-Awareness” in: Philosophical Studies 120, 37 – 89. Martin (2006): Mike G. F. Martin, “On Being Alienated”, in: T. Gendler/J. Hawthorne (eds.), Perceptual Experience, Oxford, 354 – 410. McDowell (2008a): John McDowell, “Avoiding the Myth of the Given”, in: McDowell (2009b), 256 – 272. McDowell (2008b): John McDowell, “The Disjunctive Conception of Experience as Material for a Transcendental Argument”, in: McDowell (2009a), 225 – 242. McDowell (2009a): John McDowell, The Engaged Intellect, Cambridge/MA. McDowell (2009b): John McDowell, Having the World in View. Essays on Kant, Hegel, and Sellars, Cambridge/MA. McDowell (2010): John McDowell, “Tyler Burge on Disjunctivism”, in: Philosophical Explorations 13, 243 – 255 (reprinted in this volume, ???-???). Sellars (1997): Wilfrid Sellars, Empiricism and the Philosophy of Mind, Cambridge/ MA. Siegel (2008): Susanna Siegel, “The Epistemic Conception of Hallucination”, in: Haddock/Macpherson (2008a), 205 – 225. Travis (2005): Charles Travis, “Frege, Father of Disjunctivism”, in: Philosophical Topics 33, 307 – 334.

Disjunctive Conceptions of Experience and Perceiving* Adrian Haddock 1. There is a man in front of me chewing on the end of his pencil. How I do know this? Why, because I see a man in front of me chewing on the end of his pencil. That is to say, I take it that there is a man in front of me doing that, because my (visual) experience supplies me with a reason for taking it that this is so, in the shape of the fact that I see a man in front of me doing that. This is a truly excellent reason, for if I see a man in front of me doing that then there is a man in front of me doing that. Consequently, because I take it that this is so for this reason, I do not merely take it that this is so, but know that this is so—I know that there is a man in front of me, chewing on the end of his pencil. These are seemingly natural thoughts; but there is an argument—a variant on the Argument from Illusion1—which maintains that they must be rejected. It is possible that, in a case in which there is no man in front of me chewing on his pencil, I have an experience ‘just like’ the experience I have when I see a man in front of me chewing on the end of his pencil. In a case like this, I could not be seeing a man in front of me doing that; I could, at best, be dreaming of a pencil-chewing man, or having a hallucination of such a man. Here we have two possible cases: a ‘good’ case, and a ‘bad’ case. In the good case, I see a man in front of me chewing on the end of his pencil. And let us say that, in the bad case, I have a hallucination of a man in front of me doing that. The experience I have in the good *

1

I would like to thank Jane Calvert, Jennifer Hornsby, John McDowell, Alan Millar, and Marcus Willaschek for their very helpful comments on earlier versions of this essay. One of these earlier versions has been published as “The disjunctive conception of perceiving”, in Philosophical Explorations 14 (2011), 23 – 42. Reprinted by permission of the publisher (Taylor & Francis Group, http:// www.informaworld.com). Here I follow John McDowell’s presentation of this variant, in McDowell (1982).

350

Adrian Haddock

case is ‘just like’ the experience I have in the bad case. According to this variant on the Argument from Illusion, it follows that the reasons which my experience supplies me with must be the same in the good case as they are in the bad case. The fact that I see a man in front of me chewing on the end of his pencil is not a reason which my experience can supply me with in the bad case, because in the bad case I do not see, but rather have a hallucination of, a man in front of me doing that. So, according to this variant on the Argument, this is not a reason which my experience can supply me with in the good case either. ( Just so, the fact that I have a hallucination of a man in front of me doing that is not a reason which my experience can supply me with in the good case, because in the good case I do not have a hallucination of, but rather see, a man in front of me doing that. So, according to this variant on the Argument, this is not a reason which my experience can supply me with in the bad case either.) The only reasons which my experience can supply me with are those which are neutral between the good case and the bad case, e. g., my experience can only supply me with a reason consisting in the fact that I seem to see a man in front of me chewing on his pencil. So, we must abandon the seemingly natural thoughts with which we began. In 1982, John McDowell gave a lecture which sought to rescue these seemingly natural thoughts from this variant on the Argument from Illusion.2 McDowell pointed out that it does not simply follow, from the fact that my experience in the good case is ‘just like’ my experience in the bad case, that the reasons which my experience supplies me with must be the same in both cases. Why not say instead that, even if my experience in the good case is ‘just like’ my experience in the bad case, it can still supply me with a reason which my experience in the bad case cannot supply me with, e. g., a reason consisting in the fact that I see a man in front of me chewing on his pencil? Expressed differently, McDowell’s suggestion is that my experience either supplies me with a reason consisting in the fact that I see such-andsuch, or supplies me with a reason consisting merely in the fact that I seem to see such-and-such.3 The fact that McDowell originally ex2 3

Cf. McDowell (1982). For an extended defence of this way of understanding McDowell, see Haddock/Macpherson (2008).

Disjunctive Conceptions of Experience and Perceiving

351

pressed his suggestion in the guise of a disjunction is one reason why it has come to be known as the disjunctive conception of experience. Since the time of McDowell’s lecture, interest in the disjunctive conception has not merely grown but spiralled—if not quite out of control, at least in several different directions. One way to summarise McDowell’s position would be to say that, even though my experience in the good case is ‘just like’ my experience in the bad case, there is a fundamental difference between my experiences, in terms of the reasons they can supply me with. The idea of a fundamental difference between experiences which are ‘just like’ one another is central to recent writings on the disjunctive conception. However, McDowell’s concern with experience’s capacity to supply me with reasons, and therefore with knowledge, is discarded by these recent writings, in favour of a quite different understanding of the idea of a fundamental difference between experiences which are ‘just like’ one another. The thoughts with which these recent writings begin are awarded the title ‘naïve realism’. They maintain that, when I see, e. g., a man in front of me chewing on the end of his pencil, what I see partly constitutes my experience itself (so a man in front of me chewing on the end of his pencil partly constitutes my experience, in this case).4 Naïve realism is threatened by a different variant on the Argument from Illusion, according to which it follows from the fact that my experience in the good case is ‘just like’ my experience in the bad case that my experience must be constituted by the same things in each case. In the bad case, my experience cannot be constituted by what I see—i. e., by a man in front of me chewing on the end of his pencil—because in the bad case I do not see a man in front of me chewing on the end of his pencil. So, according to this variant on the Argument, my experience in the good case just as much as in the bad case cannot be constituted by what I see. The version of the disjunctive conception which figures in these recent writings insists that, on the contrary, my experience in the good case and my experience in the bad case are differently constituted, even though they are ‘just like’ one another: the former is constituted by what I see, even though the latter is not. Expressed differently, and more generally, this version of the disjunctive conception insists that my experience is either constituted by what I see—e. g., by a man chewing on the end of his pencil—or con4

See especially Martin (2002) and the essays collected in the first part of Haddock/Macpherson (2008).

352

Adrian Haddock

stituted otherwise. Once again, there is a fundamental difference between the experiences; but the idea of this difference is no longer understood in terms of the different reasons which my experiences can supply me with. Because of its concern with the nature of experience, we might refer to this conception of experience as the metaphysical disjunctive conception; and because of its concern with experience as a source of reasons and knowledge we might refer to McDowell’s conception of experience as the epistemological disjunctive conception.5 It is sometimes maintained that whatever constitutes one’s experience constitutes, inter alia, the phenomenology of one’s experience. To say that my experience in the good case is ‘just like’ my experience in the bad case seems to be to say that my experience has the same phenomenology in both cases. What exactly this means is hard to pin down. But one standard way to gloss it is to say that how things seem from my point of view is the same in the good case (when I see the man chewing away) as it is in the bad case (when I have a hallucination of him doing that). So, if whatever constitutes my experience constitutes its phenomenology, then in claiming that what constitutes my experience in the good case is different from what constitutes my experience in the bad case, the metaphysical version makes it look as if my experience in the good case must differ in phenomenology from my experience in the bad—as if how things seem from my point of view when I am in the good case cannot be the same as how things seem from my point of view when I am in the bad case. This might seem to be the beginning of a criticism of the metaphysical disjunctive conception. However, many of its proponents insist on this phenomenological difference. They insist that, in this sense, my experience in the good case is not ‘just like’ my experience in the bad case.6 But they claim that my good experience is ‘just like’ my bad experience in a different sense, i. e., in the sense that it is not possible to know by introspection alone which experience I am having: whether I am having an experience in which I see a man in front of me chewing away, as opposed to an experience in which I have a hallucination of such a man. How things seem from my point of view when I am in the good case is 5

6

As far as I am aware, Alex Byrne and Heather Logue were the first clearly to distinguish between epistemological and metaphysical versions of the disjunctive conception of experience; see their “Either/Or” in Haddock/Macpherson (2008), 57 – 94. Cf. Millar (2007). This is Martin’s view. For a comprehensive discussion, see Soteriou (2009).

Disjunctive Conceptions of Experience and Perceiving

353

not the same as how things seem from my point of view when I am in the bad case. Nevertheless, introspection alone cannot tell me whether the case I am in is good rather than bad. I have rehearsed the foregoing material centrally to emphasise the differences between McDowell’s epistemologically inflected project and the project of those who endorse the metaphysical disjunctive conception. It seems to be possible to endorse the seemingly natural thoughts which McDowell’s disjunctive conception is designed to protect without taking a view as to the merits of ‘naïve realism’, whether the latter is advanced as a claim about the phenomenology of experience, or merely as a claim about what constitutes experience (assuming it is possible to keep claims about what constitutes experience apart from claims about what constitutes its phenomenology). But this essay is concerned neither with ‘naïve realism’, nor with the metaphysical disjunctive conception. Its concern is with features of McDowell’s epistemological disjunctive conception which have not yet been brought into focus, but which must be central to any assessment of its merits.7 2. It is a truth to which philosophers often pay lip service that McDowell agrees with Wilfrid Sellars “that knowledge—at least as it is enjoyed 7

By way of doing even more to distance McDowell’s disjunctive conception from the metaphysical disjunctive conception, let me note that, even though McDowell makes use of the idea of experiences which are ‘just like’ one another, it is not obvious that this idea must be understood in terms either of the idea of sameness of phenomenology, or in terms of the idea of what is knowable by introspection alone. In a recent essay, McDowell remarks that even though “the claim that from a subject’s point of view, a misleading experience can be indistinguishable from a case in which things are as they appear … might be taken as a self-standing claim about the phenomenology of misleading appearance, available to be cited in explaining the fact that subjects can be misled by appearances … the right way to take it is as registering the fact that, on that interpretation, it is supposed to explain: the undeniable fact that our capacity to get to know things through perception is fallible” (McDowell (2006a), reprinted in McDowell (2009a), 231). I will say more about the idea of a fallible capacity to know things through perception as I proceed. But for now the point to note is that to say that my experience in the good case is ‘just like’ my experience in the bad case might be merely to say that it can be reasonable for me to take it that my experience is good (e. g., I am seeing a man chewing on his pencil) when in fact my experience is bad (e. g., I am hallucinating such a man). Perhaps the notions of introspection and phenomenology need not figure in the epistemological disjunctive conception.

354

Adrian Haddock

by rational animals—is a certain sort of standing in the space of reasons”,8 “of justifying and being able to justify what one says”.9 But it ought to be stressed10 that at least part of the point of this Sellarsian doctrine is “to exclude an externalistic view of epistemic satisfactoriness, a view according to which one can be entitled to a belief without being in a position to know what entitles one to it”.11 McDowell’s central contention is that, if I enjoy perceptual knowledge that P, then not only do I have a reason, or a justification, or an entitlement, or a warrant—he tends to use the notions of reason, justification, entitlement, and warrant interchangeably12—to believe that P, but this reason (or justification, or entitlement, or warrant) does not fall short of the fact known (i. e., the fact that P). And, in most if not all of his writings which deal with perceptual knowledge, he assumes that this reason consists in the fact that I perceive that P; e. g., in the case of specifically visual perceptual knowledge, in the fact that I see that P.13 So far I have not made use of the idea of perceiving that P in my presentation of McDowell’s disjunctive conception, preferring to employ the idea of perceiving such-and-such— e. g., the idea of seeing a man in front of me chewing on his pencil, rather than that of seeing that there is a man in front of me chewing on his pencil. My reasons for this policy will emerge in the sequel. But, staying with the idea of perceiving that P, it is McDowell’s thought that if I perceive that P, then I am entitled to believe that P, precisely by the fact that I perceive that P. As he puts it, “if one undergoes an experience on the “good” side of the disjunction [i.e., if one perceives that things are a certain way] that warrants one in believing—indeed presents one with an opportunity to know—that things are [this way]”;14 i. e., if one perceives that P then one is warranted in believing that P, and it is precisely that fact—the fact that one perceives that P—which warrants 8 9 10 11 12

McDowell (1995), reprinted in McDowell (1998), 395. Sellars (1956), 288 f. Because it is often missed; for some discussion, cf. Neta/Pritchard (2007). McDowell (2008), reprinted in McDowell (2009b), 256. Here McDowell differs from those philosophers—e. g., Tyler Burge, and Alvin Plantinga—who put some of these notions (e. g., the notions of entitlement and justification, in Burge’s case, and the notions of warrant and justification, in Plantinga’s) to specific, technical uses. There is a useful remark on one instance of this technical tendency in McDowell (2006b), reprinted in McDowell (2009b), 132, fn. 4. 13 An apparent exception is McDowell (2008), to be discussed in §7 below. 14 McDowell (2006b), reprinted in McDowell (2009a), 234.

Disjunctive Conceptions of Experience and Perceiving

355

one so to believe. Putting this thought together with McDowell’s antiexternalist, Sellarsian thought that if I am entitled to believe that P then I am in a position to know what entitles me to believe that P, we arrive at the claim that, if I perceive that P, then I am in a position to know that I perceive that P.15 (McDowell sometimes16 writes as if he thinks that, if I perceive that P, then not only is there a reason for me to believe that P (a reason which consists in the fact that I perceive that P), but I have a reason to believe that P (a reason which consists in the fact that I perceive that P). It is plausible to suppose that McDowell considers himself entitled to speak in this way precisely by his endorsement of the claim at which we have just arrived, because it is plausible to think that I can have a fact as a reason to believe that P only if I am in a position to know this fact.) For reasons which will become clear as we proceed, McDowell’s claim that if I perceive that P then I am in a position to know that I perceive that P might be described as his ‘luminosity’ claim. My primary aim in this essay is to suggest that McDowell should, and to show how McDowell can, abandon this luminosity claim. After clarifying McDowell’s epistemological outlook further—(in §3), by means of a comparison between some themes in McDowell’s epistemological writings and some recent work by Sebastian Rödl—I shall suggest (in §§4&5) that there are reasons for McDowell to abandon his luminosity claim, and, as a result, reasons for him to abandon one of the two thoughts which together entail this claim. The first of these thoughts (i. e., the Sellarsian thought that if I am entitled to believe that P then I am in 15 We arrive at least at this claim. There are two ways to understand the Sellarsian thought which are relevant in the present context: as the thought that if I am entitled to believe that P by the fact that Q then I am in a position to know that Q; and, as the thought that if I am entitled to believe that P by the fact that Q then I am in a position to know that I am entitled to believe that P by the fact that Q. If we understand it in the first way, then we arrive precisely at this claim. But if we understand it in the second way, then we arrive at the stronger claim that if I am entitled to believe that P by the fact that I perceive that P then I am in a position to know that I am entitled to believe that P by the fact that I perceive that P. (That is stronger because the truth of its consequent entails but is not entailed by the fact that I am in a position to know that I perceive that P.) For my purposes, it is enough that the Sellarsian thought enables us to arrive at the weaker claim; so, I propose to understand this thought in the first rather than the second of these ways. 16 Cf., e. g., McDowell (2003), reprinted in McDowell (2009a), 152 – 159.

356

Adrian Haddock

a position to know what entitles me to believe that P) is more fundamental to McDowell’s overall epistemological project than the second thought (i. e., the thought that if I perceive that P then I am entitled to believe that P by the fact that I perceive that P). So, I shall suggest that what McDowell should do is hold on to the first of these thoughts, by rejecting the second, and replacing it with a new disjunctive conception, a disjunctive conception of perceiving. This new disjunctive conception, I shall suggest, will enable McDowell to avoid the luminosity claim, whilst retaining both the Sellarsian thought and his central contention that, if I enjoy perceptual knowledge that P, then I have a reason to believe that P which does not fall short of the fact that P. However, this new disjunctive conception will give rise to the question of whether it is possible to understand the reason integral to perceptual knowledge as consisting in the fact that I perceive such-andsuch, rather than in the fact that I perceive that P. I shall suggest that this is indeed possible. That will take us (in §6) to a new disjunctive conception of perceiving—a disjunctive conception of perceiving suchand-such, rather than a disjunctive conception of perceiving that P. I shall suggest (in §7) that McDowell has reason to endorse this disjunctive conception of perceiving such-and-such. My aim, in this essay, is simply to work through the details of McDowell’s epistemological project, with a view to showing not only that there is an aspect of his project he ought to abandon, but also how he can abandon it without unsettling his fundamental epistemological commitments. Because I share many of these commitments, the disjunctive conception of perceiving such-and-such at which we will eventually arrive is a conception which I think we all ought to endorse. But I cannot argue for this stronger claim here. In this essay, my central aim must be more limited: to argue that, in the light of the commitments which shape his broader epistemological project, there are reasons for McDowell to endorse this disjunctive conception. 3. Perceiving that P puts me in a position to know that P; or, as McDowell sometimes puts it—for he uses these notions interchangeably too— it presents me with an opportunity to know that P, or, it makes knowledge that P available to me. McDowell wants to allow for cases in which (e. g.) I see that P, but I do not know (visually) that P because I do not believe that P. He offers the following as an example of such a case:

Disjunctive Conceptions of Experience and Perceiving

357

I thought I was looking at your sweater under one of those lights that make it impossible to tell what colours things are, but I now realize I was actually seeing that it was brown.” In saying this, one registers that one had, at the relevant past time, an entitlement that one did not then realize one had. One was in a position to acquire a bit of knowledge about the world, but because of a misapprehension about the circumstances, one did not avail oneself of the opportunity.17

Perhaps a (normally) reliable friend has told me that the lights in the room we are in are of the sort which make it impossible to tell what colours things are, thereby making it reasonable for me both to refuse to believe that I am seeing that your sweater is brown, and—because vision is the only relevant source of information available to me—to refuse to believe that your sweater is brown; refusals I consequently effect, thereby ensuring that I do not know (visually) that your sweater is brown, even though, actually, I was seeing that your sweater is brown, and so had the opportunity to know this fact. And not just this fact: I also had the opportunity to know that I was seeing that your sweater is brown, but my misapprehension led me not to avail myself of this opportunity. In order to avail myself of the opportunity to know that your sweater is brown—an opportunity I have simply in virtue of seeing that your sweater is brown—it is not enough for me simply to believe that your sweater is brown; I need to believe that your sweater is brown because I see that it is brown, i. e., because of the very thing whose being the case puts me in a position to know that your sweater is brown. And this ‘because’ is the ‘because’ of the species of causal explanation which shows rationality to be in operation—as it would not be in operation if I were to believe that your sweater is brown in a case in which it is reasonable for me to refuse to believe that your sweater is brown.18 (In such a case, I might believe that your sweater is brown because I see that it is brown; but this causal explanation would not show rationality to be in operation.) According to recent work by Sebastian Rödl, I exercise my fallible capacity to know things through perception only if I believe that P because I perceive that P—where the ‘because’ is the ‘because’ of the spe17 McDowell (2003), reprinted in McDowell (2009a), 158. 18 More precisely, if one is in a position to know (perceptually) that P because one perceives that P, then one knows (perceptually) that P if, and only if, one believes that P because one perceives that P—where that ‘because’ is the ‘because’ of the appropriate species of causal explanation.

358

Adrian Haddock

cies of causal explanation which shows rationality to be in operation. In general, a fallible capacity is one whose “exercise is liable to be thwarted by unfavourable circumstances”.19 However, the exercise of my fallible capacity to know things is also liable to be thwarted if I am in what Rödl (following Andrea Kern) calls “reflectively unfavourable circumstances”20—circumstances which are not unfavourable, but which it is reasonable for me to believe are unfavourable. Circumstances in which the lights make it impossible to tell what colours things are by looking count as unfavourable; if I was in those circumstances then I could not see that your sweater is brown. By contrast, circumstances in which there is nothing wrong with the lighting, but which it is reasonable for me to take to be circumstances in which there is something wrong with the lighting, count as reflectively unfavourable; if I was in those circumstances, then even though I might see that your sweater is brown, I would—given adequate doxastic responsibility on my part, and given that vision is the only relevant information-source available—refuse to believe that your sweater is brown, and, consequently, neither believe that your sweater is brown because I see that your sweater is brown, nor exercise my capacity to get to know the colours of things through visual perception. It is no accident that Rödl’s work in epistemology is relevant at this juncture because its goal is (inter alia) “that of articulating the account of self-consciousness and spontaneous knowledge contained in McDowell’s epistemological writings”.21 The knowledge that I perceive that P which, according to McDowell, I am at least in a position to acquire whenever I perceive that P is, according to Rödl, spontaneous knowledge—in contrast to perceptual knowledge that P, which is receptive knowledge. Rödl explains the difference between these two forms of knowledge by saying that whereas the latter (e. g., perceptual knowledge that P) is “a distinct reality from its object”22 (in this case, the fact that P), the former (e. g., knowledge that I perceive that P) and its object (in this case, the fact that I perceive that P) are “the same reality”.23 And, seemingly in order to elucidate this idea of “the same reality”, Rödl says the following: “if someone bears a receptive nexus to 19 20 21 22 23

Rödl (2007). Kern (2006), 332 f.; cited in Rödl (2007), 154, fn. 19. Rödl (2007), 147, fn. 11. Rödl (2007), 140. Rödl (2007), 145.

Disjunctive Conceptions of Experience and Perceiving

359

an object by which she is in a position to gain [receptive] knowledge of it then she knows that she does, and knows it by bearing this nexus to the object”.24 Perceiving that P is one case of bearing a receptive nexus to an object (i. e., to the fact that P) by which I am in a position to gain receptive (i. e., perceptual) knowledge of it. So, it seems that, according to Rödl, if I perceive that P, then I know that I perceive that P. That claim seems to be what he has in mind when he says that my knowledge that I perceive that P is spontaneous, i. e., that it is “the same reality” as its object. But this claim is surely too strong, for a reason we have seen—a reason which Rödl not only acknowledged but helped us to formulate: I might perceive that P but not know that I perceive that P, precisely because I might be in (what Rödl, following Kern, calls) reflectively unfavourable circumstances. McDowell’s own explanation of knowledge of perceiving that P avoids this last claim, because—contra Rödl—it insists that, from the fact that “someone bears a receptive nexus to an object by which she is in a position to gain … knowledge of it”, it does not follow that “she knows that she does”, merely that she has an opportunity to know that she does, an opportunity which she may not avail herself of thanks (e. g.) to her being in reflectively unfavourable circumstances. And in order to avail herself of this opportunity—an opportunity she enjoys simply in virtue of perceiving that P—it is, once again, not enough that she acquires the associated belief, the belief that she perceives that P; she needs to acquire this belief because she perceives that P, i. e., because of the very thing whose being the case puts her in a position to know that she perceives that P—where this ‘because’ is, once again, the ‘because’ of that species of causal explanation which shows rationality to be in operation.25 This is not to deny that, unlike my perceptual knowledge that P, my knowledge that I perceive that P is, for McDowell, in some sense spontaneous rather than receptive. The sense in question can be explained by saying that whereas it is a consequence of the object of spontaneous knowledge (e. g., the fact that I perceive that P) that I am in a position to know it (e. g., to know that I perceive that P), it is not a consequence of the object of receptive knowledge 24 Rödl (2007), 143 f. 25 More precisely, if one is in a position to know that one perceives that P because one perceives that P, then one knows that one perceives that P if, and only if, one believes that one perceives that P because one perceives that P—where that ‘because’ is the ‘because’ of the appropriate species of causal explanation.

360

Adrian Haddock

(e. g., the fact that your sweater is brown) that I am in a position to know it; to know the latter, I must bear a receptive nexus to the object (e. g., I need to perceive that it is brown). That is not quite the sense Rödl had in mind when he said spontaneous knowledge and its object are “the same reality”. But it is hard to see how they can be “the same reality”, in this sense, by Rödl own lights, if I might, e. g., perceive that P, but—thanks to the fact that I am in reflectively unfavourable circumstances—fail to know that I perceive that P. 4. I have suggested that McDowell’s view is not vulnerable to the worry which Rödl’s account of spontaneous knowledge gives rise to. But McDowell’s view does give rise to a worry of its own, as we shall now see. This worry turns on what I referred to (in §2) as McDowell’s luminosity claim. Consider the following guise of this luminosity claim: if one sees that P, then one is in a position to know that one sees that P. What worries might this inspire? One worry is rather naïve. There might be circumstances in which one sees that P even though, were one to believe that P, one’s belief that P would be only accidentally true; e. g., circumstances in which were one to believe that this—what one sees—is a barn, one’s belief would be only accidentally true, because every other barn-like object in the vicinity is a façade; in these circumstances, even though one sees that P, one is not in a position to know that P, or in a position to know that one sees that P. For if one was in a position to know that P then one would know that P if one were to believe that P because one sees that P.26 But, in these circumstances, one cannot know that P. And if one was in a position to know that one sees that P, then one would know that one sees that P if one were to believe that one sees that P because one sees that P.27 But, in these circumstances, one cannot know that one sees that P, because if one could know this then one could know that P—and in these circumstances, one cannot know that P. The obvious response to this worry—one which is already contained in McDowell’s epistemological writings—is that, in these cir26 Where the ‘because’ is the ‘because’ of the appropriate species of causal explanation. 27 Where the ‘because’ is the ‘because’ of the appropriate species of causal explanation.

Disjunctive Conceptions of Experience and Perceiving

361

cumstances, one cannot see that P; e. g., one can see a barn, but one cannot see that this is a barn, precisely because, in these barn-façadeheavy circumstances, one cannot know that it is a barn.28 However, one might worry that this response is distorting our linguistic practice, so as to fit it into a mould forged by a prior philosophical theory of perceptual knowledge. And if this worry takes hold, the disjunctive conception of perceiving can come to the rescue. According to this conception, cases of perceiving that P either supply one with a reason consisting in the fact that one perceives that P, because they are essentially cases in which one is in a position to know that one perceives that P, or supply one merely with a reason consisting in the fact that one seems to perceive that P, because they are cases in which one merely perceives that P, i. e., cases of perceiving that P which are not essentially cases of being in a position to know that one perceives that P; one might be in a position to know that one perceives that P, in such a case; but equally, one might not be; and, if one is, then this will be due to something only contingently related to the case of perceiving itself. Were we to grant this conception, we would prevent the worry from arising, by allowing for the possibility of perceiving that P but not being in a position to know that one perceives that P. But we would prevent the worry from arising without insisting that one simply cannot see that something is a barn, in the barn-façade-heavy circumstances. I do not think there is very much to this worry. However, I think there is a more serious worry, which better serves to motivate the disjunctive conception of perceiving.29 McDowell’s luminosity claim is socalled because it is an instance of a general claim which has come to be known as luminosity. And there are good reasons to be suspicions of luminosity. These reasons tend to be cast in the form of a rather technicallooking argument, but they can be stated quite simply.30

28 McDowell (1982), reprinted in McDowell (1998), 390, fn. 37. 29 I mention this less serious worry because I know that it moves some people, and for this reason hope that it might do something to begin to make clear the attractions of the disjunctive conception of perceiving. 30 The argument is due to Williamson (2000). Soteriou (2009), §3.2, seems to me to go wrong in supposing that McDowell’s claim that there are some facts about one’s mind which one can know only fallibly is of a piece with Williamson’s opposition to luminosity (for this claim, cf. McDowell (1986), reprinted in McDowell (1998), 28 – 59). For one thing, things which are only fallibly knowable might still be luminous. For another, Williamson thinks that (modulo some

362

Adrian Haddock

First, let us assume something markedly stronger than the present claim. Let us assume an instance of what we might call ‘super-luminosity’, i. e., if one sees that P, then one knows that one sees that P—or, more precisely, for any time t, if at t one sees that P then at t one knows that one sees that P. And second, let us assume a certain sort of reliability principle for knowledge: for any times t and t+1, where t and t+1 are any two times spaced only fractionally—say, one millisecond—apart, if at t one knows that something is the case, then at t+1 this very thing is the case; e. g., if at t I know that I see that your sweater is brown, then at t+1 I see that your sweater is brown. Now imagine a stretch of time between two intervals, at the beginning of which I see that your sweater is brown, but at the end of which I do not (perhaps this is a stretch of time during which your sweater is slowly starting to look a different colour, because the lights which make it impossible to tell the colours of things are slowly turning on). And let ‘ti’ stand for the first interval, ‘ti+n’ stand for the second interval, and ‘ti+1’, ‘ti+2’, ‘ti+3’… and so on stand for each millisecond of time between these two intervals. At ti I see that your sweater is brown. Given the super-luminosity claim, at ti I know that I see that your sweater is brown. So, given the reliability principle, at ti+1 I see that your sweater is brown. And given the super-luminosity claim, at ti+1 I know that I see that your sweater is brown. So, given the reliability principle, at ti+2 I see that your sweater is brown. And given the super-luminosity claim, at ti+2 I know that I see that your sweater is brown. So, given the reliability principle, at ti+3 I see that your sweater is brown. And so on. The upshot is that at ti+n, I see that your sweater is brown. But, by hypothesis, at ti+n I do not see that your sweater is brown. As we might put it: the upshot is that I see that your sweater is brown during the whole of the relevant stretch of time. And that is absurd, because by hypothesis I do not see that it is brown at the end of the relevant stretch. It seems that either the super-luminosity claim or the reliability principle must be relinquished. Suspicion immediately falls onto the superluminosity claim, given that it is stronger than McDowell’s luminosity claim, which says only that if one sees that P then one is in a position to know that one sees that P—or, more precisely, that for any time t, if at t one sees that P then at t one is in a position to know that one exceptions which are irrelevant for present purposes) nothing is luminous, whereas it seems that McDowell disagrees.

Disjunctive Conceptions of Experience and Perceiving

363

sees that P. But it seems that replacing super-luminosity by luminosity will not scotch the absurd upshot. This is especially easy to see if we work with the way of understanding the idea of being in a position to know which discussions of the anti-luminosity argument tend to assume, according to which if I am in a position to know that something is the case, and I attend to the question of whether this is the case, then I know that it is the case. We need simply to imagine our stretch of time between two intervals—at the beginning of which I see that your sweater is brown, but at the end of which I do not—and to insist that throughout all of this stretch of time I am attending to the question of whether I see that your sweater is brown. For now—on the assumption that t is a time during the relevant stretch (including the two intervals)—the luminosity claim that, for any time t, if at t I see that your sweater is brown then at t I am in a position to know that I see that your sweater is brown, will entail the super-luminosity claim that, for any time t, if at t I see that your sweater is brown then at t I know that I see that your sweater is brown. So, it will still follow—given the reliability principle—that I see that your sweater is brown during the whole of the relevant stretch of time. And this is absurd, because by hypothesis I do not see that it is brown at the end of the relevant stretch. However, there is no need for us to employ this way of understanding the idea of being in a position to know, because the absurd conclusion will follow even if we work with the way of understanding this idea which McDowell favours. We need simply to imagine our stretch of time between two intervals—at the beginning of which I am seeing that your sweater is brown, but at the end of which I am not—and to insist that, for any time t, where t is a time during this stretch (including the two intervals), if at t I see that your sweater is brown, then at t I believe that I see that your sweater is brown because I see that your sweater is brown.31 For now—on the assumption that t is a time during the relevant stretch (including the two intervals)—the luminosity claim that, for any time t, if at t I see that your sweater is brown then at t I am in a position to know that I see that your sweater is brown, will, once again, entail the super-luminosity claim that, for any time t, if at t I see that your sweater is brown then at t I know that I see that your sweater is brown. So, it will still follow—given the reliability prin31 Where the ‘because’ is the ‘because’ of the appropriate species of causal explanation.

364

Adrian Haddock

ciple—that I see that your sweater is brown during the whole of the relevant stretch of time. And this is absurd, because by hypothesis I do not see that your sweater is brown at the end of the stretch. Once disabused both of the illusion that the anti-luminosity argument does not touch McDowell’s luminosity claim but merely the super-luminosity claim, and of the related illusion that it will not touch the former once the idea of being in a position to know is understood in the way McDowell recommends, suspicion naturally falls onto the reliability principle—and indeed this has been the standard target for suspicion.32 However, I think there is a ladder of thought, leading all the way up to this principle, which is sufficiently appealing that any argument for rejecting the principle, no matter how compelling, will seem to land one in the dialectically unattractive position of denying one of the steps in this ladder. The following constitutes a seemingly general requirement on knowledge: if one knows that something is the case, then one believes that it is the case because it is the case—where the ‘because’ is the ‘because’ of some species of causal explanation. This requirement appears, in the case of perceptual knowledge, in the form of the thought that if one knows (perceptually) that something is the case then one believes that it is the case because one perceives that it is the case (and so, because it is the case). And it appears, in the case of knowledge that one perceives that something is the case, in the form of the thought that if one knows that one perceives that something is the case then one believes that one perceives that it is the case because one perceives that it is the case. Now it seems that, if in one situation something is the case and one believes that it is the case, but in a very close situation it is not the case and yet one still believes that it is the case, then the claim that, in the former situation, one believes that it is the case because it is the case is in jeopardy. And if this claim is jeopardised then so is the claim that, in the former situation, one knows that it is the case—given the seemingly general requirement on knowledge. So, it seems—given this requirement—that if in one situation one knows that something is the case, but in a very close situation this thing is not the case, then in this very close situation one does not believe that this thing is the case. What is it for two situations to count as being ‘very close’? Well, it seems to be a fact about our capacity for altering confidence-levels that 32 See, e. g., Neta/Rohrbaugh (2004).

Disjunctive Conceptions of Experience and Perceiving

365

we cannot, across a fractionally small space of time (say, a single millisecond) shift from a state of believing that something is the case with the confidence required for knowing that it is the case, to a state of not believing that it is the case. And here by ‘our capacity’ I mean the capacity of human beings like us, who are at our current stage of biological development, and who maintain their confidence-levels without the benefit of special scientific instrumentation. We might cast this fact into hypothetical form by saying that, if at one time one knows (and so believes) that something is the case, then at a fractionally later time (say, one millisecond later) one still believes that it is the case—other things being equal, of course, i. e., so long as one still exists at the later time, and still exists as the ordinary human being one is. Given this, it seems that two situations temporally spaced only fractionally (say, one millisecond) apart will always count, for human beings, as being ‘very close’, no matter how much the situations may differ in other respects. We can make this point vivid—in a way which exploits the principle enunciated at the end of the previous paragraph (the principle which seems to flow from the seemingly general requirement on knowledge)—by saying that a world in which nothing is the case for any longer than one millisecond would be a world in which we cannot know anything.33 Things would zip by too quickly for us to believe that they are the case because they are the case. And with that, we arrive at the anti-luminosity argument’s reliability principle. For when we combine this understanding of the idea of a ‘very close’ situation with the principle which seems to flow from the general requirement on knowledge, we arrive at the principle that if in one situation one knows that something is the case, but in a situation temporally spaced only fractionally (say, one millisecond) apart from the former situation this thing is not the case, then in the latter situation one does not believe that it is the case. And it follows from the conjunction of this principle, and the hypothetical into which the aforesaid fact about our capacity for altering confidence-levels was cast, that if in one situation one knows that something is the case then in a situation temporally spaced only fractionally (say, one millisecond) apart from the former sit-

33 Or at least, a world in which we cannot know anything which we can know to be the case only if we can believe that it is the case because it is the case—where the ‘because’ is the ‘because’ of some species of causal explanation.

366

Adrian Haddock

uation this thing is the case. In other words, what follows is the reliability principle.34 So, it seems anyone who disputes the reliability principle of the antiluminosity argument will be in the unattractive position of disputing one of the steps in this ladder of thought, be it the seemingly general requirement on knowledge, or the principle which seems to flow from this requirement, or the apparent fact about our capacity for altering confidence-levels, or the understanding of the idea of a ‘very close’ situation which this apparent fact seems to entrain. That is hardly enough to compel acceptance of the reliability principle, even when it is understood as a principle which applies merely to the knowledge enjoyed by ordinary human beings like us. But it ought to be enough to caution against rejecting the principle—so understood—and, as a result, enough to caution against rejecting the anti-luminosity argument to which it is essential. 5. The upshot is that we have some reason to reject luminosity, and as a result some reason to reject McDowell’s luminosity claim. The fact that there is some reason to reject this luminosity claim means there is some reason to endorse the disjunctive conception of perceiving. For if we are to reject this claim, we shall need to reject one of the two thoughts which together entail it, i. e., either McDowell’s thought that if I am entitled to believe that P then I am in a position to know what entitles me to believe that P, or his thought that if I perceive that P then I am entitled to believe that P by the fact that I perceive that P. The first of these thoughts—McDowell’s anti-externalist, Sellarsian 34 It might help to spell out more explicitly why this follows. Assume that t and t+1 are two times spaced one millisecond apart, and that ‘C’ refers to some specific thing which can be the case. And grant that (1) if at t one knows that C is the case, and at t+1 C is not the case, then at t+1 one does not believe that C is the case (i. e., grant the principle which seems to follow from the seemingly general requirement on knowledge, fleshed out in line with the present understanding of the idea of being ‘very close’); and, that (2) if at t one knows that C is the case then at t+1 one believes that C is the case (i. e., the hypothetical which purports to capture the putative fact about our capacity for altering confidence-levels). Now assume that (3) at t one knows that C is the case. It follows from 2 and 3 that (4) at t+1 one believes that C is the case. And it follows from 1 and 4 that (5) either at t one does not know that C is the case, or at t+1 C is the case. And it follows from 3 and 5, that (6) at t+1 C is the case. So, it follows from 1 and 2 that (7) if at t one knows that C is the case then at t+1 C is the case. In other words, the reliability principle follows from 1 and 2.

Disjunctive Conceptions of Experience and Perceiving

367

thought—is not in dispute here, because this thought is fundamental to McDowell’s conception of “knowledge—at least as it is enjoyed by rational animals—[as] a certain sort of standing in the space of reasons”. So, we shall need to reject the second of these thoughts (given that we reject the luminosity claim). One great strength of the disjunctive conception of perceiving is that it enables us to reject this second thought without relinquishing either the first thought (the Sellarsian thought), or McDowell’s contention that one can be entitled to believe that P by the fact that one perceives that P. According to the disjunctive conception, cases of perceiving that P are, either, essentially cases of being in a position to know that one perceives that P, or, cases of merely perceiving that P. If one perceives that P, and the case of perceiving that P which one enjoys is of the former variety, then it follows that one is in a position to know that one perceives that P. Whereas, if one perceives that P, and the case of perceiving that P which one enjoys is of the latter variety, then it does not follow that one is in a position to know that one perceives that P; one might be in a position to know this, but equally, one might not be. It is precisely by allowing for the possibility of perceiving that P but not being in a position to know that one perceives that P that the disjunctive conception avoids the luminosity argument. For given the disjunctive conception, the following is possible: at t I know that I see that your sweater is brown, and so—given the reliability principle of the anti-luminosity argument—at t+1 I see that your sweater is brown; but at t+1 I do not know that I see that your sweater is brown (even though at t+1 I believe that I see that your sweater is brown because I see that your sweater is brown) precisely because when at t+1 I see that it is brown I merely see that it is brown. No general claim is being made to the effect that, for any time t, if at t I see that P then at t I am in a position to know that I see that P. So, the threat that I will never stop seeing that your sweater is brown throughout the relevant stretch of time is no longer present. (If this is not yet clear, just imagine a millisecond in the stretch at which the case of seeing that your sweater is brown which I enjoy is a case of being in a position to know that I see that your sweater is brown. And assume both McDowell’s way of understanding the idea of being in a position to know, and that whenever I see that your sweater is brown I believe that I see that your sweater is brown because I see that your sweater is brown. Given the reliability principle, it follows that at the next millisecond I still see that your sweater is brown. But because the luminosity claim has been

368

Adrian Haddock

replaced by the disjunctive conception, it is possible that, at this millisecond, the case of seeing that your sweater is brown which I enjoy is a case of merely seeing that your sweater is brown, and, consequently, possible that at this millisecond I do not know that I see that your sweater is brown—even though it is the case, at this millisecond, that I see that your sweater is brown.) But even though this threat is scotched, neither the Sellarsian thought (i. e., that if I am entitled to believe that P then I am in a position to know what entitles me to believe that P) nor the contention that one can be entitled to believe that P by the fact that one perceives that P is thereby relinquished. It is the fact that I perceive that P which entitles me to believe that P, when I enjoy perceptual knowledge that P. And it is my enjoyment of a case of perceiving that P which ensures that I am in a position to know that I perceive that P. But (and this is the crucial point) it is not my enjoyment of any old case of perceiving that P which ensures that I am in a position to know that I perceive that P; rather, what ensures that I am in this position is my enjoyment of a case of perceiving that P of the right sort, i. e., a case of the sort which figures in the left-hand disjunct— the ‘good’ disjunct—of the disjunctive conception of perceiving. 6. In §4, I mentioned the worry that there are circumstances in which one can see that P without being in a position to know (visually) that P, e. g., barn-façade scenarios. But there will be those who refuse to feel this worry. There will be those who agree with McDowell that the very idea of seeing that P is the idea of being in a position to know (visually) that P.35 And there will be those who think, contra McDowell, that the idea of seeing that P is the idea, not merely of being in a position to know (visually) that P, but of knowing (visually) that P.36 To the latter, much of the foregoing—which has followed McDowell in taking seeing that P to fall short of knowing (visually) that P—will have seemed rather odd. But even to the former, the claim that there can be cases of seeing that P which are neither cases of knowing (visually) that P, nor cases of being in a position to know (visually) that P, will have seemed odd. (To the latter this claim will have seemed odder still.) And the disjunctive conception entails this claim, given the following plausible thoughts, both of which McDowell endorses, and neither of which I wish to deny: first, that if I am in a position to know (perceptually) that P 35 Rödl seems to be an example; cf. Rödl (2007), ch. 7. 36 Williamson seems to be an example; cf. Williamson (2000), ch. 1.

Disjunctive Conceptions of Experience and Perceiving

369

then I am entitled to believe that P by the fact that I perceive that P; and, secondly, that if I am entitled to believe that P by the fact that I perceive that P, then I am in a position to know that I perceive that P. For the disjunctive conception allows that there can be cases in which I perceive that P but am not in a position to know that I perceive that P (these will be cases in which I merely perceive that P). And because I am not in a position to know that I perceive that P, in these cases, it follows—given both of the aforesaid plausible thoughts—that, in these cases, I am not in a position to know (perceptually) that P. It thereby follows that there can be (e. g.) cases of seeing that P which are not cases of being in a position to know (visually) that P. And McDowell will find that thought rather odd. This ostensible oddness might seem to put pressure on the disjunctive conception. But why not see it instead as putting pressure on McDowell’s assumption that the entitlement which I have, when I enjoy perceptual knowledge, consists in the fact that I perceive that P? I noted (in §2) that this assumption diverges from the policy I adopted in §1, of treating this entitlement as consisting in the fact that I perceive such-and-such, rather than in the fact that I perceive that P, e. g., in the fact that I perceive a man in front of me chewing on his pencil, rather than in the fact that I perceive that there is a man in front of me chewing on his pencil. Perhaps the frame in which the disjunctive conception should ideally emerge is one constituted by this policy, rather than by McDowell’s assumption. This policy does not dispute McDowell’s central contention that the reason required for perceptual knowledge must not fall short of the fact known; e. g., I cannot see a man in front of me chewing on the end of his pencil unless there is a man in front of me chewing on the end of his pencil. But replacing McDowell’s assumption by this policy will remove the ostensible oddness attaching to the disjunctive conception of perceiving. To illustrate: if I have perceptual knowledge that your sweater is brown, then I have a reason to believe that your sweater is brown, consisting in the fact that I see your brown sweater. Of course, I can see your brown sweater without knowing that your sweater is brown. But, at least from McDowell’s perspective, this is no objection to thinking that the fact that I see your brown sweater can be an entitlement, or a reason, which I have to believe that your sweater is brown. And this is because—for McDowell—I can see that your sweater is brown without knowing that your sweater is brown, precisely because I am in what Kern calls “reflectively unfavourable circumstances” (see §3). What is

370

Adrian Haddock

more, it does not follow from the fact that I can see your brown sweater without being in a position to know that your sweater is brown that the fact that I see your brown sweater cannot function as a reason for me to believe that your sweater is brown. We just need to insist that in order for it to function as such a reason I need to be in a position to know that I see your brown sweater. For if I am in a position to know that I see your brown sweater, how can I fail to be in a position to know that your sweater is brown? From the mere fact that I see your brown sweater it may not follow that I am in a position to know that your sweater is brown. But if this fact functions as a reason which I have to believe that your sweater is brown, then it does follow that I am in this position. The general idea is that, if I enjoy perceptual knowledge that P, then I believe that P for a reason consisting in the fact that I perceive such-and-such—where what replaces ‘such-and-such’ suitably matches what replaces ‘P’. We can shed some light on this idea of a suitable match by more precisely specifying both the form of the content of perceptual knowledge, and the form of what is perceived in an episode of perceiving such-and-such as—respectively—“there is an object with such-and-such features at such-and-such a location”, and “an object with such-and-such features at such-and-such a location”.37 And now the disjunctive conception of perceiving can enter the scene—where this is understood as the disjunctive conception, not of perceiving that P, but of perceiving such-and-such. The disjunctive conception of perceiving such-and-such maintains that cases of perceiving such-and-such either supply one with a reason consisting in the fact that one perceives such-and-such, because they are essentially cases of being in a position to know that one perceives such-and-such, or supply one merely with a reason consisting in the fact that one seems to perceive such-and-such, because they are cases of merely perceiving such-and-such, i. e., cases of perceiving suchand-such which are not essentially cases of being in a position to know that one perceives such-and-such. It should be clear that, unlike the disjunctive conception of perceiving that P, which did entail the ostensibly odd thought that one can (e. g.) see that P without being in a position to know (visually) that P, the new disjunctive conception of perceiving such-and-such does not carry this entailment. For consider the following two plausible thoughts, cousins of the two plausible thoughts mentioned in the first paragraph of 37 Compare McDowell (2008), reprinted in McDowell (2009b), 266.

Disjunctive Conceptions of Experience and Perceiving

371

this section: first, if one is in a position to know perceptually that P then one is entitled to believe that P by the fact that one perceives such-andsuch; secondly, if one is entitled to believe that P by the fact that one perceives such-and-such then one is in a position to know that one perceives such-and-such (where, in each case, what replaces ‘such-andsuch’ suitably matches what replaces ‘P’). Given these plausible thoughts, the new disjunctive conception entails that there can be cases in which one perceives such-and-such but is not in a position to know (perceptually) that P. But there is nothing even ostensibly odd about this thought. (The thought that I can (e. g.) see a man in front of me, but not be in a position to know (visually) that there is a man in front of me is routine.) What is more, there is no obvious route from this thought to the ostensibly odd thought that one can perceive that P but not be in a position to know (perceptually) that P. Indeed, it seems that, now that we have replaced the disjunctive conception of perceiving that P with the disjunctive conception of perceiving such-and-such, we are free to use the idea of perceiving that P either in such a way that if one perceives that P then one actually knows (perceptually) that P, or in such a way that if one perceives that P then one is in a position to know (perceptually) that P. Nothing in the new disjunctive conception of perceiving such-and-such, even when it is conjoined with some or even all of the aforesaid plausible thoughts, will prevent one from using the idea of perceiving that P in either of these ways. 7. In a recent essay, McDowell insists that seeing such-and-such—as it is enjoyed by rational animals—requires the possession of suitable conceptual capacities on the part of the perceiving subject; specifically, conceptual capacities associated with concepts of “proper sensibles of sight and common sensibles accessible to sight”38—concepts of (e. g.) colours, and shapes. This is required to ensure that cases of seeing such-and-such are intrinsically opportunities to know that what one sees exhibits certain sensibles. Which sensibles these are depends not just on which sensibles are exhibited by what one sees but on which conceptual capacities are drawn into operation in the case of seeing which one enjoys. The idea is that a case of seeing something which exhibits a certain sensible, in which the conceptual capacity associated with the concept of this sensible is suitably drawn into operation, is intrinsically an opportunity to know that what one sees exhibits this sensible. It is taken for granted 38 McDowell (2008), reprinted in McDowell (2009b), 260.

372

Adrian Haddock

that, in every case of seeing, some conceptual capacities associated with concepts of the relevant sensibles are suitably drawn into operation. And it is taken for granted that the fact that such-and-such conceptual capacities are suitably operative, in a particular case of seeing, is itself an intrinsic feature of the case of seeing in question. Insisting on these points is consistent with denying (as McDowell does deny) that cases of seeing are intrinsically opportunities to know more sophisticated things. E.g., no case of seeing your brown sweater is intrinsically an opportunity to know that I see your brown sweater; my being in a position to know that always requires the extrinsic presence of a suitable re-cognitional capacity, i. e., the capacity to tell that what I see is your sweater. However, McDowell thinks that my being in a position to know that I see something brown need not require the extrinsic presence of the capacity to tell that what I see is something brown; I could be in that position simply by seeing (e. g.) your brown sweater, so long as when I see it the conceptual capacity associated with the concept brown is suitably drawn into operation. What we have here is a glimpse of a disjunctive conception of perceiving such-and-such (or at least, a disjunctive conception of seeing such-and-such). Some cases of seeing something brown are intrinsically cases in which the conceptual capacity associated with the concept brown is suitably drawn into operation, whereas other cases of seeing something brown are not intrinsically cases in which the conceptual capacity associated with this concept is suitably operative. And it seems that McDowell ought to insist that the latter—or at least, included amongst the latter—are cases of merely seeing something brown, so as to avoid another luminosity claim, i. e., the claim that if one sees something brown then one is in a position to know that one sees something brown. And what goes for brownness goes for every other sensible one can see. So, if McDowell’s recent thought is on the right lines, then, on pain of the anti-luminosity argument, he has reason to endorse the disjunctive conception of perceiving such-and-such, for cases of seeing something which exhibits a certain sensible. That is, he has reason to think that such cases come in two sorts: those which are essentially cases of being in a position to know that one sees something which exhibits the relevant sensible; and those which are cases of merely seeing something which exhibits the relevant sensible. (Whether McDowell’s recent thought is on the right lines is a topic for a different essay.)

Disjunctive Conceptions of Experience and Perceiving

373

8. We are now in a position to express the disjunctive conception of perceiving as follows: cases of perceiving such-and-such either supply one with a reason consisting in the fact that one perceives such-andsuch, because they are—essentially—cases of being in a position to know that one perceives such-and-such, or supply one with a reason consisting merely in the fact that one seems to perceive such-andsuch, because they are cases of merely perceiving such-and-such, i. e., cases in which, even though one perceives such-and-such, it does not follow that one is in a position to know that one does so. This way of putting things enables us to press the question of the relation between the disjunctive conception of perceiving and McDowell’s original disjunctive conception of experience; specifically, whether the former should be thought to supplement, or to supplant the latter, where the latter says that experience either supplies one with a reason consisting in the fact that one perceives such-and-such, or supplies one with a reason consisting merely in the fact that one seems to do so. I suggest we think of the disjunctive conception of perceiving as supplementing the latter, by issuing the cautionary reminder that for one’s experience actually to supply one with the former reason it is not enough for one’s experience to be a case of perceiving such-andsuch; it must be a case of perceiving such-and-such of a certain sort, a case of perceiving such-and-such which is—essentially—a case of being in a position to know that one perceives such-and-such, as opposed to (what I call) a case of merely perceiving such-and-such. That is the nub of the disjunctive conception of perceiving, which it has been my aim to recommend in this essay.

References Byrne/Logue (2008): Alex Byrne/Heather Logue, “Either/Or”, in: Haddock/ Macpherson (2008), 57 – 94. Haddock/Macpherson (2008): Adrian Haddock/Fiona Macpherson (eds.), Disjunctivism: Perception, Action, Knowledge, Oxford. Kern (2006): Andrea Kern, Quellen des Wissens. Zum Begriff vernnftiger Erkenntnis-Fhigkeiten, Frankfurt/Main. Martin (2002): M.G. F. Martin, “The Transparency of Experience”, in: Mind and Language 17, no. 4, 376 – 425. McDowell (1982): John McDowell, “Criteria, Defeasibility, and Knowledge”, in: Proceedings of the British Academy 68, 459 – 479 (reprinted in McDowell (1998), 369 – 394).

374

Adrian Haddock

McDowell (1986): John McDowell, “Singular Thought and the Extent of Inner Space”, in: P. Petit/J. McDowell (eds.), Subject, Thought, and Context, Oxford, 137 – 168 (reprinted in McDowell (1998), 228 – 259). McDowell (1995): John McDowell, “Knowledge and the Internal”, in: Philosophy and Phenomenological Research 55 (4), 877 – 893 (reprinted in McDowell (1998), 395 – 413). McDowell (1998): John McDowell, Meaning, Knowledge, and Reality, Cambridge/MA. McDowell (2003): John McDowell, “Subjective, Intersubjective, Objective”, in: Philosophical and Phenomenological Research 67 (3), 675 – 681 (reprinted in McDowell (2009a), 152 – 159). McDowell (2006a): John McDowell, “The Disjunctive Conception of Experience as Material for a Transcendental Argument”, in: Teorema, 25 (1), 19 – 33 (reprinted in McDowell (2009a), 225 – 242). McDowell (2006b): John McDowell, “Conceptual Capacities in Perception”, in: G. Abel (ed.), Kreativitt, Hamburg, 1065 – 1079 (reprinted in McDowell (2009b), 127 – 144). McDowell (2008): John McDowell, “Avoiding the Myth of the Given”, in: J. Lindgaard (ed.), Experience, Norm, and Nature, Malden/MA, 1 – 14 (reprinted in McDowell (2009b), 256 – 272). McDowell (2009a): John McDowell, The Engaged Intellect. Philosophical Essays, Cambridge/MA. McDowell (2009b): John McDowell, Having the World in View. Essays on Kant, Hegel, and Sellars, Cambridge/MA. Millar (2007): Alan Millar, “What the Disjunctivist is Right About”, in: Philosophy and Phenomenological Research 74 (1), 176 – 198. Neta/Pritchard (2007): Ram Neta/Duncan Pritchard, “McDowell and the New Evil Genius”, in: Philosophy and Phenomenological Research 74 (2), 381 – 396. Neta/Rohrbaugh (2004): Ram Neta/Guy Rohrbaugh, “Luminosity and the Safety of Knowledge”, in: Pacific Philosophical Quarterly 85 (4), 396 – 406. Rödl (2007): Sebastian Rödl, Self-Consciousness, Cambridge/MA. Sellars (1956): Wilfrid Sellars, “Empiricism and the Philosophy of Mind”, in: H. Feigl/M. Scriven (eds.), Minnesota Studies in the Philosophy of Science 1, Minneapolis, 253 – 329. Soteriou (2009): Matthew Soteriou, “The Disjunctive Theory of Perception”, Stanford Encyclopedia of Philosophy, http://plato.stanford.edu/entries/perception-disjunctive/ (2011/04/11). Travis (2004): Charles Travis, “The Silence of the Senses”, in: Mind 113 (449), 59 – 94. Williamson (2000): Timothy Williamson, Knowledge and its Limits, Oxford.

Trusting One’s Senses: McDowell on Experience, Belief and Justification Martin Gustafsson Must I not begin to trust somewhere? That is to say: somewhere I must begin with not-doubting; and that is not, so to speak, rash but forgivable. Rather, it belongs to judging.1

1. Introduction In this paper, I will ascribe to the author of Mind and World the following view. The basic, default attitude we have toward sense experience cannot be a matter of critical assessment. It cannot typically be the case that we step back from and reflect on the veridicality of our experiences before we come to believe what they purport to reveal about the world. Rather, in the default case, what happens is that we trust our senses without further ado – we trust that things are as our experience represents them as being. It is only against the background of such widespread trust that critical scrutiny of particular experiences is possible. Specific doubts make sense only because doubt is not generally present. Readers of Mind and World might find this interpretation surprising, questionable, or even downright mistaken. For in that book, John McDowell may seem to be downplaying or even ignoring the importance of perceptual belief-formation that is unreflective in the just described sense. Indeed, some of the wordings he uses to spell out the contrast between the active nature of judgment and the passive character of experience may be taken to suggest a view in direct opposition to the one I have just sketched – a view according to which the testimony of the senses is in the typical case a matter of decision-making based on reflection on the credentials of the relevant experience. 1

Wittgenstein (1969), §150: “Muß ich nicht irgendwo anfangen zu trauen? D. h. ich muß irgendwo mit dem Nicht-zweifeln anfangen; und das ist nicht, so zu sagen, vorschnell aber verzeihlich, sondern es gehört zum Urteilen.” Translation amended.

376

Martin Gustafsson

For example, he writes that “[h]ow one’s experience represents things to be is not under one’s control, but it is up to one whether one accepts the appearance or rejects it.”2 In a similar vein, he claims that the content of an experience “becomes the content of a judgment if the subject decides to take the experience at face value.”3 Such formulations may seem to imply that unreflective trust is at most of secondary importance to perceptual belief-formation, and have been taken as evidence that McDowell’s conception of the relation between belief and experience is implausibly intellectualist and voluntarist. A more comprehensive study of his writings, however, shows that this intellectualist-voluntarist interpretation is mistaken. In fact, McDowell is quite explicit that he never wanted to deny that our default attitude toward sense experience is a matter of unreflective trust. Thus he writes that “[u]nless there is ground for suspicion, such as odd lightning conditions, having it look to one as if things are a certain way […] becomes accepting that things are that way by a sort of default.”4 At another place, he notes that “coming to believe something on the basis of experience is not in general happily conceived as deciding what to think. […] Normally we arrive at our perceptually based beliefs without reflection.”5 It should be noted that such passages involve no concession to the idea that sense experience is non-conceptual. ‘Without reflection’ does not mean ‘without the use of concepts’. Quite the contrary: McDowell’s point presupposes that experience is conceptual. For what he is claiming is that in the standard case, perceptual beliefs are formed without critical assessment of the particular experience whose content is being endorsed as true; and for there to be any experiential content to be endorsed, reflectively or unreflectively – for the endorsement to be a matter of accepting that things are thus and so (namely, as one’s experience represents them as being) – that content must be conceptual. Or, so McDowell believes; and I do not want to question that idea, at least not here. Now, McDowell also claims that for such unreflective trust to be a matter of accepting that things are as one’s experience represents them as being, she who trusts must have the capacity to sometimes engage in the 2 3 4 5

McDowell McDowell McDowell McDowell

(1996), 11. (1996), 26. (2009a), 11. (2009a), 139 ff.

Trusting One’s Senses: McDowell on Experience, Belief and Justification

377

sort of reflection in which one steps back from and reflects on the credibility of some particular experience. According to McDowell, a perceiver’s acceptance of what her senses tell her constitutes an epistemologically significant move – a move in the logical space of reasons – only insofar as the conceptual resources at work in that process are resources that the perceiver is also able to employ in assessing the credentials of particular experiences whenever there are concrete grounds for suspicion. Trust of the sort we are considering here can figure only in subjects who can also now and then refuse to believe their senses.6 So, for McDowell, there is an important sense in which the ability to employ conceptual capacities in critical reasoning is essential to the possibility of forming perceptual beliefs. His claim is that a creature totally incapable of such free, critical reasoning has no concepts, and hence cannot form any perceptual beliefs at all, reflectively or unreflectively. But this claim does not entail that the unreflective formation of perceptual beliefs is any less basic than the reflective formation of such beliefs. Nor does it entail that such unreflective belief-formation is somehow illegitimate, or “rash but forgivable” (to use Wittgenstein’s phrase in the epigraph). Rather, the idea is that this sort of trust inevitably goes together with a capacity to engage in critical assessment whenever there is a genuine, positive reason to do so. What I am going to argue is that he thinks the converse is also true. Just as he takes unreflective perceptual belief-formation to require a capacity for critical assessment, he takes the capacity for critical assessment to require that one’s default attitude toward experience is a matter of unreflective trust. According to McDowell, the idea of someone whose typical stance is to step back from and scrutinize her experiences before she accepts what they reveal about the world is no more intelligible than the idea of someone who can form beliefs only by unreflectively accepting that things are as her experience represents them as being.

2. Two Worries In everyday life, if I am asked for a justification of my belief that p, an answer of the form, ‘I see that p’ is often considered to do the job. Suppose Julia asks me over the phone what reason I have to think that Paul 6

Cf. McDowell (2009a), 141.

378

Martin Gustafsson

is wearing a red shirt at Peter’s birthday party, and I reply: ‘Well, I’m at the party and Paul is right in front of me; I see that he is wearing a red shirt.’ Unless Julia has some specific positive reason to question my honesty or my ability to visually discern such things – unless she knows I am color-blind, for example – it would be outrageous of her not to find my answer satisfactory. It is a central feature of McDowell’s conception of experience that he thinks this form of justification is all right as it stands. A fruitful approach to his view is therefore to look at two philosophical worries that may arise with respect to this form of justification, and consider McDowell’s responses to them. The first worry is this: Seeing that p is a belief-dependent, or doxastic notion. I see that p only if I believe that p. However, the mere fact that I believe that p in no way justifies my belief that p. So, if we want to identify what it is about my seeing that p that justifies my belief that p, we have to purge it of its doxastic element; the justificatory power must reside in the non-doxastic part of my seeing that p. In other words, a philosophically proper characterization of the everyday justificatory pattern described above would have to start from an analysis of ‘I see that p’ into ‘I believe that p’, and a residue V – where V signifies that which really does the justificatory work. The philosophical task will then be to find the adequate characterization of V. What we can be sure of is that V will not be identical with, but only a part of, my seeing that p. Hence, the everyday way of specifying the ground of my perceptual belief that p in terms of my seeing that p is inexact.

As we shall see, McDowell’s answer to this worry is simply to deny that we have to think of ‘seeing that p’ in doxastic terms. According to McDowell, a philosophically more significant notion of ‘seeing that p’ – the notion that figures in the sort of everyday justificatory procedure whose legitimacy he is concerned to defend – is such that one can see that p without believing that p. However, even if one accepts that the relevant notion of ‘seeing that p’ is non-doxastic, one might still be worried that the envisaged procedure of justification cannot be a matter of genuine justification. This second worry can be spelled out as follows: Someone’s seeing that p entails that p is actually the case: seeing that p is a factive notion. It follows that I cannot take myself (or anyone else) to be seeing that p, unless I already believe that p. After all, if I didn’t believe that p, I would not take myself (or anyone else) to be seeing that p, but only to be ‘having the impression that p’, or something similar. In other words, I would then characterize the relevant experience in non-factive terms. Con-

Trusting One’s Senses: McDowell on Experience, Belief and Justification

379

sequently, making use of my factively characterized experience to justify my belief means putting the cart before the horse. I can take myself to be entitled to such a factive characterization only if I take myself to be entitled to believe that p; but this latter entitlement was precisely what the experience was supposed to give me.

As we shall see, McDowell’s answer to this second worry is that it involves thinking of the justificatory procedure in terms of making an inference – say: I see that Paul is wearing a red shirt. Consequently, Paul is wearing a red shirt.

The worry is not about the validity of this inference. It is clearly valid. Nor is it about the truth of the premise or of the conclusion. It can be assumed that they are both true. Rather, the worry is that my performing this inference cannot be a matter of my providing genuine support for the belief that Paul is wearing a red shirt, since I (or anyone else) can take myself to be entitled to accept the premise only if I already take myself to be entitled to accept the conclusion. McDowell’s way of rejecting this worry is to argue that the presumed inferential model does not fit what is going on in the sort of everyday case where I justify my belief that Paul is wearing a red shirt by reference to my seeing that Paul is wearing a red shirt. This procedure, McDowell argues, is justificatory even if it involves no inference at all. By exploring in greater detail the two worries that I have just described and McDowell’s responses to them, I will find the material needed to support my earlier claim about the centrality of unreflective beliefformation for McDowell’s outlook. The structure of my discussion will be as follows. I begin, in section 3, by clarifying McDowell’s response to the first worry, concerning the allegedly doxastic character of ‘seeing that p’. In section 4, I provide a detailed discussion of various aspects of McDowell’s alternative, non-doxastic conception – a discussion which lays bare an important connection between the first worry and the second one (concerning facitivity and justification). In section 5, I spell out McDowell’s response to that second worry. I then explain how this response leads up to the view I have ascribed to him in section 1. Very roughly, my claim will be that McDowell’s way of navigating between an inferential and a brutely non-conceptual view of perceptual belief-formation is precisely to think of such belief-formation in terms of unreflective yet fully rational trust.

380

Martin Gustafsson

3. Is Seeing Believing? I begin by considering McDowell’s response to the worry about the alleged belief-dependence of ‘seeing that p’. Sentences such as the following do seem somehow absurd: (1) I see that the tie is green, but I don’t believe that it is green. Whence the absurdity? One possible explanation is that seeing that p entails believing that p. According to Barry Stroud, it is a “truism” that “[a] person who sees that it is raining judges or believes or accepts or otherwise puts forward as true that it is raining. […] To see that p is to judge that p.”7 If this is right, (1) is absurd since it entails the contradiction, ‘I believe that the tie is green, but I don’t believe that it is green’. This would explain not only the absurdity of (1). It would also mean that the following sentences are just as bizarre: (2) I saw that the tie was green, but I didn’t believe that it was green. (3) She sees that the tie is green, but she doesn’t believe that it is green. And aren’t they? I think it should be admitted that Stroud’s “truism” captures one intelligible notion of ‘seeing that p’ – a notion which is belief-dependent or ‘doxastic’ in the same sense as ‘recognizing that p’, ‘realizing that p’, and so on. Still, there might well be some other notion of ‘seeing that p’ which is equally legitimate and of deeper philosophical interest. According to McDowell, there is indeed such a philosophically more fundamental notion. In his discussions of experience, knowledge and thought, McDowell is working with a notion of ‘seeing that p’ such that (2) and (3) are not absurd. According to this belief-independent or nondoxastic conception, seeing that p does not entail believing that p. Thus, suppose that you are looking at what you mistakenly believe is an instance of the Müller-Lyer illusion – a drawing of the familiar sort, but where one of the lines is in fact a little longer than the other. Knowing about the Müller-Lyer, you refuse to believe what your eyes tell you. You are convinced that the lines are equally long, despite the fact that your experience correctly represents one line as longer than the other. In such a case, McDowell would say: You see that one line is longer than the other, but you don’t believe it. 7

Stroud (2002), 84.

Trusting One’s Senses: McDowell on Experience, Belief and Justification

381

Or, to use the example that figures in the sentences (1)-(3): Suppose you are looking at a green tie in what you wrongly take to be illumination of a sort that makes it impossible to tell what color things are. In fact, the lightning conditions are normal. Again, McDowell claims, you refuse to accept what you do in fact see. You see that the tie is green, but you do not believe it.8 McDowell is not denying the absurdity of (1). Rather, the point is that his notion of ‘seeing that p’ is such that the absurdity of (1) stems only from the fact that ‘seeing that p’ entails ‘p’, and, hence, that ‘I see that the tie is green, but I don’t believe that it is green’ entails the Moore-paradoxical construction: (4) The tie is green, but I don’t believe that it is green. So according to McDowell, even if saying (1) is clearly absurd, this in no way forces us to conclude that seeing that p entails believing that p. As he construes the notion of ‘seeing that p’, the absurdity has to do instead with the fact that one cannot take oneself to be seeing that p without believing that p. Hence the adding of the second conjunct, ‘I don’t believe that it is green’, makes my utterance of the whole sentence Moore-paradoxically absurd. One may feel that McDowell’s notion of ‘seeing that p’ is strained or artificial. Again, I suspect the spontaneous reaction among most speakers to sentences such as (2) and (3) is that they are inconsistent, or at least quite awkward. As Hannah Ginsborg notices, in such cases it is probably more natural to say of the person in question that even if the green color she perceives is indeed the color of the tie, she fails to see (realize, appreciate) that the tie is green.9 On the other hand, there are natural occurrences of ‘seeing that p’ that are in line with McDowell’s analysis rather than with Stroud’s. Consider a situation in which a tie appears green to me and I haven’t yet decided whether to trust the appearance or not. So, I ask myself: Do I really see that the tie is green? Given Stroud’s notion of ‘seeing that p’, this question can be readily answered in the negative simply in virtue of the fact that I have not yet formed the 8

9

McDowell (2002), 277 f., McDowell (2004), 214, McDowell (2009a), 9, fn. 10. Sellars uses a similarly non-doxastic notion of ‘seeing that p’ in ‘Empiricism and the Philosophy of Mind’. Cf. Sellars (1963), 145. In fact I think the view I am ascribing to McDowell in this paper is quite close to Sellars’s view in his most famous essay. Ginsborg (2006), 298.

382

Martin Gustafsson

belief that the tie is green. But that is obviously to miss the whole point of the question.10 In any case, it seems clear that Stroud’s and McDowell’s different notions are both intelligible. So, again, the central issue is perhaps not which one is the more natural, but which one is philosophically the most consequential. McDowell thinks his non-doxastic notion of ‘seeing that p’ is the one we need to clarify the nature of perception, knowledge and thought. “[T]his notion,” he says, “is the right one for my purposes” – the suggestion being that someone like Stroud, who is blind to the very possibility of this notion, will also be blind to the insights that it can help us achieve.11 In order to understand what these insights might be, we have to look at McDowell’s response to the second worry described in section 2: that about the factivity of ‘seeing that p’. Interestingly, this worry and McDowell’s response to it are best approached via some further reflections on what he has to say about the belief-independence of experience.

4. Two Aspects of Belief-Independence The non-doxastic character of ‘seeing that p’ is explicitly discussed in several of McDowell’s writings from the beginning of the 2000’s and onward. In Lecture 3 of Mind and World, there is also a discussion of the belief-independence of experience, but there a somewhat different point is being made. More precisely, what McDowell is talking about there is a sort of belief-independence that becomes discernible when experience is characterized in non-factive terms. He is discussing and arguing for the non-doxastic character of ‘It appears to S as if p’, ‘S has the impression that p’, and so on. This is also how Gareth Evans talks about belief-independence in The Varieties of Reference, and McDowell thinks Evans is “very perceptive” on this point.12 Evans notes that “[i]t is a well-known fact about perceptual illusions that it will continue to appear to us as though, say, one line is a little longer than the other (in the Müller-Lyer illusion) even when we are quite sure that it is not”.13 McDowell makes the same point: “a familiar 10 11 12 13

Ginsborg (2006), 298 f. McDowell (2002), 277. McDowell (1996), 61. Evans (1982), 123.

Trusting One’s Senses: McDowell on Experience, Belief and Justification

383

visual illusion continues to present its illusory appearance even though the subject does not believe that things are as they look”.14 This seems to be the easiest way of illustrating belief-independence with respect to non-factive characterizations of experience: In confronting what I rightly take to be an instance of the Müller-Lyer, it appears to me as if one line is longer than the other even if I do not believe it. Contrast this with the sort of case most easily used to illustrate belief-independence with respect to factive constructions such as ‘seeing that p’: In confronting what I wrongly take to be an instance of the Müller-Lyer, I see that one line is longer than the other even if I do not believe it. What is the relation between these two notions of belief-independence? At first sight, they may seem utterly different – so different, indeed, that only one of them (the ‘non-factive, Mind and World-Evans variety) has to do with experience as such. Thus, compare: (1) I see that the tie is green, but I don’t believe that it is green. (5) I have the visual impression that the tie is green, but I don’t believe that it is green. The difference between the two notions of belief-independence must surely be tied to the fact that an utterance of (1) has the sort of Moore-paradoxical absurdity described above, whereas an utterance of (5) is not absurd at all. Now if one reflects further on the absurd character of (1), it may start to look as if we can explain the special beliefindependence of ‘I see that p’ exclusively in terms of the fact that p – a fact seemingly external to the experience itself. After all, as has already been pointed out, (1) is absurd precisely because it entails, (4) The tie is green, but I don’t believe that it is green. So, it would seem that if we subtract the fact that the tie is green from my seeing that the tie is green, we subtract the very thing that makes the belief-independence of ‘seeing that p’ have the special, Moore-paradoxical character described earlier. After the subtraction (the argument continues), all that remains is the experience itself – my having the visual impression that the tie is green – and this experience, purged of any reference to the external fact that the tie is green, is belief-independent only in the straightforward, non-Moorean way described by Evans and in Mind and World. The upshot seems to be that the other, Moorean sort of belief-independence is, so to speak, an artificial product of con14 McDowell (1996), 60.

384

Martin Gustafsson

joining the experience itself with the external fact due to which the experience is veridical. And the conclusion might seem obvious: If we want to understand the nature of experience as such, the factive yet non-doxastic notion of experience that McDowell is working with cannot be of fundamental philosophical significance. For that notion now appears like a mere conglomerate of the experience itself – something that can be fully captured in non-factive terms – and a fact external to that experience. And what we want to understand when we ask philosophical questions about experience is surely not how such a conglomerate works. We want to know about how the experience itself works – how it is related to beliefs, to reality, and so forth. Hence (the argument continues), being blind to the mere conglomerate involved in McDowell’s factive yet non-doxastic notion of experience does not seem so disastrous after all. Indeed, ignoring this notion of experience may even appear like sound philosophical instinct. It is crucial to McDowell’s conception that he rejects this argument, and the compositional view of veridical experience that underlies it. According to McDowell, I cannot perform the sort of subtraction suggested above without thereby distorting the nature of the very experience I am undergoing when I see that the tie is green. In such a veridical case, impression and fact are not distinguishable in the sense presupposed by the subtraction maneuver: “For a subject in the best [i.e., veridical] case, the appearance that there is a candle in front of her is the presence of the candle making itself apparent to her. This is not a mere seeming, which would be compatible with there being no candle there”.15 So, we must not think that the belief-independence of experience factively characterized is a feature of a complex entity consisting of the experience itself plus some (or several) external fact(s). Nor should we think of the beliefindependence of experience non-factively characterized as a feature of what would be a separable part of such a complex – the experience itself considered in isolation. Rather, these are points about two different aspects of one integrated phenomenon: the belief-independence of veridical experience. In a non-veridical case, only one of the aspects is there. But this is not to say that the second aspect – the aspect consisting of the belief-independence of, say, my seeing that p – is an aspect of something else, or something more, than the very experience I am having in a case where that experience reveals how things are. I can describe the same 15 McDowell (2009b), 281.

Trusting One’s Senses: McDowell on Experience, Belief and Justification

385

veridical experience in factive or in non-factive terms, and the different modes of belief-independence tied to these two different forms of description are both aspects of that experience – nothing more, nothing less, and nothing else.

5. The Non-Inferential Character of Perceptual Belief-Formation It is exactly at this point that we can start to really sense how important it is for McDowell to respond to the second worry described in section 2: the worry that a seeing that p cannot be what justifies my belief that p, since I will take myself (or someone else) to be seeing that p only insofar as I already believe that p. If that worry is sound, McDowell’s noncompositional conception of experience is at best epistemologically irrelevant. For what the worry seems to show is that in order to identify what it is in experience that plays a justificatory role, we have precisely to decompose my seeing that p into (i) the fact that p, and (ii) a non-factively characterized component, such as my having the visual impression that p; and it seems that it can only be the second, non-factive component that matters when it comes to justification. Now it is certainly agreed by everyone that if such a decompositional conception is allowed to frame one’s approach to epistemology, one will soon encounter well-known and terribly hard problems. The general ban on factive characterizations makes it hard to avoid a traditional conception of perceptual justification as a matter of bridging a gap between the inner realm of non-factive impressions and an outer realm of external facts. And no one has yet managed to explain what is supposed to make an inference such as the following sufficiently trustworthy to yield knowledge: I have the visual impression that Paul is wearing a red shirt. Consequently, Paul is wearing a red shirt.

In particular, the adding of more non-factive premises does not seem to help very much. Indeed, it is not clear that the envisaged distinction between inner and outer can even be made clear sense of. Many thinkers – including, as we shall see, McDowell – have come to the conclusion that the inner realm’s alleged insulation from external reality makes it impossible to understand how the elements of that inner realm can have any content at all.

386

Martin Gustafsson

Struck by the inadequacies of such a picture, modern epistemology has looked for other options. One such influential suggestion is coherentism. The coherentist straightforwardly denies that there is a justificatory relation between experiential input and beliefs. According to the coherentist, what happens when I form the perceptual belief that Paul is wearing a red shirt is that I am caused, via my senses, to believe that Paul is wearing a red shirt. Thus, the worry about my use of an inference to justify my belief is supposed not to arise, simply because I do not engage in any justification. If there is justification going on at all, it can only be a matter of internal coherence: How does the belief I find myself saddled with after my encounter with Paul – namely, that he is wearing a red shirt – cohere with my system of other beliefs? McDowell is dissatisfied with both the traditional and the coherentist responses. His most fundamental objection against the coherentist line of thought is not that it makes it difficult to understand how internal coherence among beliefs can give reason to think that those beliefs are true. Rather, his basic objection is that if there is no role other than a merely causal one for experiential input to play vis-à-vis our system of beliefs, it becomes inexplicable how those beliefs can have content (and thus be beliefs) at all – how they can be as much as true-or-false. According to McDowell, to understand how beliefs can have content, we need to conceive of experiential input as providing rational and not merely casual constraint. However – and this is crucial to understanding McDowell’s conception – he agrees with the coherentist that what happens when I form a perceptual belief about, say, the color of Paul’s shirt, is not that I perform an inference. Like in coherentism, McDowell’s way out of the traditional quandary involves denying that I infer that Paul is wearing a red shirt from some antecedently given, factive or non-factive, premise about what my experience is like. However, unlike the coherentist, McDowell nonetheless insists that I engage in a rational procedure of justification. According to McDowell, there is a rational connection between my seeing that Paul is wearing a red shirt and my belief that he is wearing a red shirt – a rational connection I exploit in finding reason to believe that Paul is wearing a red shirt. McDowell’s point is that this rational connection cannot be understood in inferential terms. How is this rational connection to be conceived, if not inferentially? This question goes to the very heart of McDowell’s philosophy. It is a difficult question – which is not to say that, in the end, the answer might not be simple and straightforward. In what remains of this

Trusting One’s Senses: McDowell on Experience, Belief and Justification

387

paper, I will try to explain how I think McDowell conceives of the rational yet non-inferential relation between experience and belief.

6. The Rationality of Non-Inferential Belief-Formation McDowell writes: [M]y aim was to spell out how the idea of rationality is in play when we explain perceptual beliefs in terms of experience. And here the notion of inference gets no grip. When one acquires a belief in this way, one comes to believe that things are as one’s experience reveals, or at least seems to reveal, that things are. The content that the explanation attributes to the experience is the same as the content of the belief explained, not a premise from which it would make sense to think of the subject as having reached the belief by an inferential step.16

So, what happens when I form the perceptual belief that Paul is wearing a red shirt is simply this: I come to believe what I see, namely, that Paul is wearing a red shirt. No inference takes place: I do not conclude that Paul is wearing a red shirt on the basis of a premise to the effect that I see that he is wearing a red shirt. I just come to believe what I do in fact see – that Paul is wearing a red shirt. According to McDowell, this is a rational operation, even if it is not an inference. Indeed, when we explain someone’s believing that things are thus and so in terms of her perceiving that things are thus and so, we are displaying the belief as a result of this kind of operation of rationality in its ideal form.17

Consider a variation of the example. Paul is in fact wearing a yellow shirt. Unbeknownst to me, the lightning conditions are such that yellow fabrics appear red. In this case, I do not see that Paul is wearing a red shirt. It only appears to me as if he is wearing a red shirt. So, what I come to believe is only what things appear to me to be like. In this case, I am still engaged in the sort of rational operation McDowell is talking about. I am still moving within the space of reasons, the space of justification. But I go wrong, misled by the appearance. In contrast to the previous case, where I come to believe what I do in fact see, in this second variation I fail to achieve a satisfactory standing in the space of reasons: 16 McDowell (2009a), 131. 17 McDowell (2009a), 132.

388

Martin Gustafsson

When we explain someone’s believing that things are thus and so in terms of the fact that her experience merely seems to reveal to her that things are thus and so, the explanation depicts the belief as a result of rationality leading its possessor astray[.] 18

At this point, the following objection is likely to arise: How can McDowell say that my standing in the space of reasons is less satisfactory in the second case than in the first? Surely, from my point of view, there is no difference as far as my right to believe that Paul is wearing a red shirt is concerned. In both cases, it appears to me that Paul is wearing a red shirt, and this appearance is what I go on in both cases. So, why say that my rational proceedings are somehow flawed in the second case, if they were flawless in the first? Isn’t the difference simply that in the first case, the world did me a favor? – it just so happened that things were as I experienced them to be. McDowell’s response to this objection is that it takes for granted a mistakenly restricted conception of rationality – a conception according to which “we ought to be able to achieve flawless standings in the space of reasons by our own unaided resources, without needing the world to do us any favors.”19 According to McDowell, my success in being fully rational may indeed depend on the world doing me a favor. He is rejecting what he describes as an “interiorization of the justifications available to us for claims about the external world”, noticing that such interiorization “threatens to deprive us of the justificatory power of, for instance, the form ‘I see that …’”.20 It is a consequence of McDowell’s view that I may in a sense be doxastically blameless – I may have done all that my peers can reasonably require of me by way of justification – and yet fail to be rationally entitled to my belief that p. It is this notion of rational entitlement that McDowell thinks we need to understand. McDowell says, I “come to believe” that Paul is wearing a red shirt by way of having a certain experience. The crucial question is: How are we to spell out this process, this “coming to believe”? On the one hand, my coming to believe that Paul is wearing a red shirt is supposed to be rational. So, it is not (as the coherentist thinks) a matter of the experience just causing the formation of a belief. On the other hand, the process is not to be conceived of in inferential terms. For, as we saw in the 18 McDowell (2009a), 132. 19 McDowell (1998), 395 f. 20 McDowell (2002), 98.

Trusting One’s Senses: McDowell on Experience, Belief and Justification

389

previous section, this would make inevitable the traditional and hopeless idea that the process has as its starting-point a non-factive characterization of the experience I am having. So, we are looking for a rational yet non-inferential process of justification. What on earth could that be? At this point, the crucial thing is to realize that, by McDowell’s lights, the question that we are asking has in an important sense already been answered. More precisely, the question is misplaced to the extent that it presumes that the account McDowell has already given leaves it open how the ‘coming to believe’ is to be spelled out – as if what he has said so far allows for a wide spectrum of construals, from the coherentist’s non-rational, causal one, to the traditionalist’s conception of the process as inferential. In fact, McDowell takes it that what he has said in the passages quoted above – that, in the ideal case, acquiring a perceptual belief involves coming to believe that things are as one’s experience reveals that things are – is determinate enough to be incompatible with both the coherentist and the traditionalist accounts of the beliefforming process. The apparently underspecified term ‘coming to believe’, as it occurs there, is in fact already specific enough to exclude both the idea that the process is merely causal and the idea that the process is inferential. Why? Well, what we are talking about is the process of coming to believe, of accepting, that things are thus and so – namely, as one’s experience reveals that they are. According to McDowell, this description is not a description of a dumb, non-conceptual, merely causal process. As was emphasized already in section 1 above, he thinks it presupposes that the perceiver responds to her experience as something that has conceptual content, and that requires that she is a rational animal who can employ the relevant conceptual resources also in free, critical reasoning.21 Nor is the description a description of an inferential process. The perceiver comes to believe precisely what she sees, namely that p. Hence, due to the Moorean dependence of ‘seeing that p’ on ‘believing that p’, the process cannot be one of going from an initially accepted premise (I see that p) to an only subsequently accepted conclusion (that p). Now one may perhaps continue to feel that even if such unreflective acceptance presupposes that one is a rational animal who is able to employ the relevant conceptual resources also in free, critical argumentation, the particular process of acceptance itself still cannot be rational. 21 McDowell (2009a), 140 f.

390

Martin Gustafsson

For this process is, after all, unreflective – and isn’t lack of reflection precisely a mark of non-rationality? This objection does not seem very strong, however. To be sure, in cases where reflection and critical scrutiny is called for, forming a belief without such reflection and scrutiny is not rational. But in a case where there is no specific positive reason to distrust one’s senses, why not say that trusting them without further ado is rational? In fact, it seems quite strained to deny that it is rational to accept that things are as one’s experience reveals that they are. If I see that the tie is green, and I have no specific reasons not to believe that the tie is green, then my believing that the tie is green seems like a clear case of rational belief-formation. Of course, one may stipulate that only inferential processes are to be counted as ‘rational’. By McDowell’s lights, however, such a stipulation only serves to hide the deep connection that he is trying to expose, between critical assessment and unreflective trust. At this point, McDowell’s imagined opponent is likely to feel that his question is being begged. For what he wants to say is that reflection and critical scrutiny are called for in the relevant cases, and that we spuriously hide this requirement from view by employing a factive characterization of the experience in question: ‘I see that the tie is green’. Of course it seems rational to accept that things are as one’s experience reveals that they are. But, the objector continues, what the unreflective acceptance is really based on is what is captured by a non-factive characterization of the experience in question. And it is not very strained to deny that it is rational to accept without further ado that the tie is green if it only looks green. On the contrary, given such a non-factive characterization of the experiential basis for my perceptual belief, reflection on the experience’s credibility might indeed seem required if the acceptance is going to count as rational. But is there a good question that is being begged here, really? After all, what our imagined objector is doing is to slide back into precisely that “interiorization of the justifications available to us for claims about the external world” that McDowell is trying to show us how to avoid. According to McDowell’s diagnosis, it is just this interiorization of the space of reasons that stops us from acknowledging the rationality of accepting without further ado what one’s senses reveal about how things are. Given that the space of reasons is seen as purely interior, it will appear as if non-factive characterizations of experience is what we need if we want to capture the real structure of the perceptual beliefforming process, and that this process will be rational only if it involves

Trusting One’s Senses: McDowell on Experience, Belief and Justification

391

reflection on the credibility of one’s experience. The problem, McDowell argues, is that this makes it impossible to understand not only how we can gain empirical knowledge, but, more fundamentally, how the very concepts we use when we purport to describe external reality can have any content at all. And if we do not understand how the concepts we use in purporting to describe external reality can have any content, we do not really understand how internal experience can be described either. If it is incomprehensible how ‘The tie is green’ can have content, it is no less incomprehensible how ‘It looks as if the tie is green’, or ‘I have the impression that there is a green tie there’, can have content.22 In trying to think of the space of reasons as interior, we ultimately lose our grip on how that space can be a space of reasons, of concepts, at all. If this is right, the question that McDowell may seem to be begging is not a coherent question at all. Returning now to the issue of inferential versus non-inferential perceptual belief-formation: McDowell would of course not deny that some perceptual beliefs are formed only after certain inferential steps have been taken. In cases where I initially suspect that the lightning conditions are not normal, I may reason as follows: Yesterday a special bulb made the lightning conditions here such that yellow fabrics looked red. I’m now having the visual impression that Paul is wearing a red shirt. In order to decide whether to trust that impression, I better check if the lightning conditions are now normal. (After inspecting the bulb:) I see that the bulb is now a normal one. Consequently, Paul is wearing a red shirt.

Here, I do step back from and scrutinize the veridicality of my impression that Paul is wearing a red shirt. According to McDowell, the ability to engage in this sort of critical scrutiny is essential to having conceptual capacities, and, hence, to the reflective and unreflective formation of perceptual beliefs. At the same time, I hope that what I have said in this paper makes it clear that, according to McDowell, such critical assessment – involving, as it inevitably does, inferential steps – cannot generally or typically be involved in perceptual belief-formation. For if it were, then there would be no genuinely significant material to start reasoning with: the alleged inferences would no longer contain premises with a determinate content. 22 McDowell (1996), 29 ff.

392

Martin Gustafsson

I have reached the conclusion I said I was going to substantiate in this paper. Despite some passages in Mind and World that might seem to point in another direction, it is in fact a cornerstone of McDowell’s view that in the typical case of perceptual belief-formation, one does not engage in critical assessment but simply trusts one’s senses without further ado. Indeed, McDowell’s conception entails that it cannot be otherwise – that the very idea of someone who forms perceptual beliefs only after having reflected on the credibility of his experiences makes no sense at all. Importantly, however, McDowell’s view should not be conflated with the idea that our unreflective acceptance of what our senses reveal about the world is a matter of non-rational, merely causal coercion. The relevant sort of acceptance or trust is a process that can occur only in rational, concept-using animals – animals that, in particular cases, can step back from a given experience and consider whether that experience is really trustworthy. The possibility of trust and the possibility of critical assessment are in that sense interdependent, and equally basic. In another sense, however, trust is more fundamental. For it must constitute the default attitude we have toward our senses, whereas distrust requires some special, concrete motivation. According to McDowell, in those typical cases where there is no special reason to doubt one’s senses, unreflective trust is what full rationality amounts to.

7. Concluding Remark: Recent Changes in McDowell’s View In this paper, I have described what I take to be McDowell’s view in Mind and World. I have also argued that this is a view which can be found in many of his later writings. It is, however, clear that in recent years – most explicitly in ‘Avoiding the Myth of the Given’, a paper originally published in 200823 – McDowell has changed his conception of experience. In particular, even if he still holds that experiences have conceptual content, he no longer thinks of that content as propositional. Hence, some of the passages I have quoted in this paper are passages to which he would no longer subscribe, at least not without substantial qualification. For example, consider the earlier quoted claim that “the content that [an explanation of a perceptual belief in terms of experience] attributes to the experience is the same as the content of the belief explained, not a premise from which it would make sense to think of 23 Lindgaard (2008).

Trusting One’s Senses: McDowell on Experience, Belief and Justification

393

the subject as having reached the belief by an inferential step.” McDowell would no longer say that the content of the experience is the same as the content of the belief. Does this mean that he would also have to take back what he says in the ensuing clause – that the content of the experience is not a premise from which it makes sense to think of the subject as having reached the belief by an inferential step? If so, much of what I have said about McDowell may no longer be true of him. To decide this issue, it is of course necessary to describe the details of his new conception. I cannot do that here, not just due to a lack of space, but also due to a lack of understanding – I’m still struggling with getting clear about McDowell’s present views. I suspect and would like to think that the interpretation I have given is largely in line also with his current ideas. But that will have to be the topic of another paper.24

References Evans (1982): Gareth Evans, The Varieties of Reference, Oxford. Ginsborg (2006): Hannah Ginsborg, “Reasons for Belief”, in: Philosophy and Phenomenological Research 72, 286 – 318. Lindgaard (2008): Jakob Lindgaard (ed.), John McDowell: Experience, Norm, and Nature, Oxford. McDowell (1996): John McDowell, Mind and World. With a New Introduction, Cambridge/MA. McDowell (1998): John McDowell, Meaning, Knowledge, and Reality, Cambridge/MA. McDowell (2002): John McDowell, “Responses”, in: Smith (2002), 269 – 305. McDowell (2004): John McDowell, “Reply to Kathrin Glüer”, in: Theoria 70, 213 – 215. McDowell (2009a): John McDowell, Having the World in View. Essays on Kant, Hegel, and Sellars, Cambridge/MA. McDowell (2009b): John McDowell, The Engaged Intellect. Philosophical Essays, Cambridge/MA. Sellars (1963): Wilfrid Sellars, Science, Perception and Reality, London. Smith (2002): Nicholas Smith (ed.), Reading McDowell: On Mind and World, London. Stroud (2002): Barry Stroud, “Sense-Experience and the Grounding of Thought”, in: Smith (2002), 79 – 91. Wittgenstein (1969): Ludwig Wittgenstein, ber Gewissheit/On Certainty, Oxford. 24 Thanks to James Conant for very useful comments on an earlier version of this paper.

Essentially Rational Animals Matthew Boyle One may call this whole disposition of the human being’s powers whatever one likes: understanding, reason, awareness, etc. It is indifferent to me, so long as one does not assume these terms to name discrete powers or mere increased levels of the animal powers. It is the whole organization of all human powers; the whole domestic economy of his sensing and cognizing, his cognizing and willing nature… The difference is not in levels or the addition of powers, but in a quite different sort of orientation and unfolding of all powers. 1

1. Introduction 1.1 According to a tradition reaching back at least as far as Aristotle, human beings are set apart from other terrestrial creatures by their rationality. Other animals, according to this tradition, are capable of sensation and appetite, but they are not capable of thought, the kind of activity characteristic of the rational part of the soul. Human beings, by contrast, are rational animals, and an understanding of our minds must begin from a recognition of this distinctiveness. For, the tradition holds, the presence of rationality does not just add one more power to the human mind, or increase the scope and efficacy of mental powers already present in nonrational creatures. Rather, rationality transforms all of our principal mental powers, making our minds different in kind from the minds of nonrational animals.2 Although the historical roots of this tradition run deep, I think it is fair to say that many contemporary philosophers regard it with suspicion. No one doubts, of course, that there are all sorts of differences between human beings and other animals, but we do not have much use these days for the idea of a single, all-pervading difference. Our philos1 2

Herder (2002), I, §2, 82 f. Aristotle himself would of course speak, not of kinds of mind, but of kinds of soul (psuche¯). But our word “psychology” descends from this Aristotelian word, and allowing ourselves the modern term “mind” permits us to describe the Aristotelian position in terms that bring out its bearing on topics of contemporary concern.

396

Matthew Boyle

ophy of mind seeks not primarily to characterize the human mind’s distinctiveness but to show how our minds fit into the natural world, and the demand that human mentality be conceived as fundamentally continuous with the mentality of other animals looks to many like just a piece of naturalistic common sense. For whatever we mean by calling our minds “rational,” surely this must be compatible with a recognition that the human mind is a species of animal mind, which has arisen through the same sorts of evolutionary processes that also produced the minds we call “nonrational.” And the more we learn about the cognitive, behavioral, and neurophysiological similarities between ourselves and other animals, and about the extent to which we “rational” creatures frequently think and choose in ways that systematically deviate from what rational principles would dictate, the more we seem compelled to regard the specialness of our minds as merely a matter of degree, not a difference in kind.3 Jerry Fodor expresses this thought with characteristic directness: [T]he whole idea that there are two (or more?) fundamentally different kinds of minds might strike one as unparsimonious… Surely it’s reasonable, absent contrary evidence, to suppose the differences between our minds and theirs are largely quantitative. The latter, after all, are widely supposed to have evolved from the former; and, indisputably, our babies turn into us. The gap can’t be impassable in either case.4

1.2 The difficulty facing the Aristotelian position, however, is not merely one of justifying the distinction it draws, but of explaining what this sort of distinction could even amount to. For what could it mean to posit a difference “in kind” between our minds and those of other animals? On a loose understanding of the idea of a difference in kind, we could say that we have two different kinds of thing wherever we have two things which differ in respect of some nonrelational property. No one will deny that a typical human mind differs in kind from (e. g.) a typical chimpanzee mind in this sense, but by the same token, no one will be inclined to make a big deal of it. Of course there are differences between human minds and the minds of other species of animals, but why should these differences be of any more interest than the differences 3

4

For a review of continuities between human cognition and the cognition of other primates, cf. for instance Tomasello/Call (1997). Standard works on nonrational cognitive biases in human judgment and choice include Nisbett/Ross (1980) and Kahneman et.al. (1982). Fodor (2003), 16.

Essentially Rational Animals

397

between the minds of chimpanzees and those of orangutans, or dolphins? Why, for that matter, should they be of any more interest than the differences between one person’s mind and another’s? Each of these differences may be of interest to some particular inquiry – to comparative psychology, to “cognitive ethology,” to the study of variations in individual cognitive ability, etc. – but none of them seems to merit the sort of absolute interest that the rational/nonrational distinction was traditionally supposed to command. There are familiar images used to indicate the nature of this special interest. Jonathan Bennett, for instance, begins his book Rationality by offering the following gloss on the idea that human minds differ in kind from those of other creatures: It is commonly believed… that between a genius and a stupid man there is a smooth slide while between a stupid man and an ape there is a sharp drop, not just in the sense that there are no creatures intellectually half-way between apes and stupid men, but in the sense that there could not be such creatures. Any possible creature whose intellectual level was higher than that of normal apes and lower than that of normal men—so the common belief runs—either would or would not have that special something which puts humans importantly above other animals.5

This characterization of the idea of a difference in kind is evocative, but in the end, I do not think it clarifies what sort of the difference is at issue. The suggestion that there could be no creatures whose intellects stand half-way between apes and men just amounts to the insistence that the rational-nonrational opposition is exclusive: for any creature, we will say either that it is rational or that it is not. But this might be true although the opposition in question was simply stipulative, an arbitrary line drawn at a certain point on what is in fact a continuum. Thus we might draw an exclusive distinction between persons over six feet tall and persons of six feet or less, calling the former group “tall” and the latter “non-tall”; but although this distinction might be useful for certain purposes, it clearly would not mark a difference in kind in the intended sense. Nor does it seem sufficient to require that the difference be discontinuous. The difference between a steam engine and an internal combustion engine, for instance, presumably involves a sharp break rather than a continuous transition: for what would the relevant continuum be? But this difference, although undoubtedly significant, does not 5

Bennett (1964), 4.

398

Matthew Boyle

seem to possess whatever sort of necessity and inevitability the rational/ nonrational contrast is traditionally supposed to have. 1.3 I have been emphasizing these difficulties in order to bring out that the idea of a difference in kind between rational and nonrational minds needs clarification before it can be assessed, and that supplying the needed clarification is not a straightforward task. I think this task is often overlooked by both fans and detractors of the distinction. People hostile to the idea that there is a deep distinction here tend to marshal a familiar battery of arguments against it, as Fodor does in the passage quoted earlier, without asking exactly what sort of claim the “difference in kind” thesis is supposed to be. But equally, people sympathetic to the idea often rush to specify what the distinction is, without explaining why the specification they give should count as a fundamental distinction between kinds of mind. Any attempt to evaluate the thesis that rational minds differ in kind from nonrational ones must, I think, begin by asking what the significance of this thesis is supposed to be. What sort of difference is a difference “in kind” meant to be, and how is the rational-nonrational contrast supposed to amount to that sort of difference? The present essay is a contribution to this preliminary but essential task. I want to understand what sort of distinction writers in the Aristotelian tradition meant to be drawing when they distinguished rational from nonrational minds, and what sort of depth they were claiming for this distinction. I will begin by offering a sketch of the outlook to which the rational/nonrational distinction traditionally belonged, and will suggest that we can only understand the nature and importance of this distinction if we recognize how it is bound up with an attempt to characterize the form of a certain type of substance, one possessing powers of a certain distinctive kind (§2). I will go on to argue that a variety of standard objections to the thesis that rational minds differ in kind from nonrational ones rest on a misunderstanding of the character of this thesis through failing to see it against this background (§3). The present essay will thus be a contribution to the defense of the rational/nonrational distinction in its classical form; but it will fall far short of being a full account of what rationality amounts to. It should be seen, rather, as a kind of prolegomenon: a specification of the framework into which a satisfying account of rationality would have to fit.

Essentially Rational Animals

399

2. The Classical View 2.1 It is often said that Aristotle defined man as a rational animal.6 What is less frequently discussed is the outlook to which the project of “defining man” belonged: What is a definition supposed to be, and how should we understand the import of the terms that appear in one? My aim in this section is to sketch what I will call the “Classical View” of these matters, in order to bring out its bearing on our question about what it means to claim that a rational creature has a different kind of mind than a nonrational one. By the Classical View, I mean the view to be found in Aristotle and in that strain of medieval Aristotelianism of which Thomas Aquinas is the greatest expositor. Of course the interpretation of Aristotle is highly contested, as is the question of the relation of his ideas to the views of later thinkers inspired by him, so it hardly needs emphasizing that what I am presenting is only a reading of certain well-known Aristotelian and post-Aristotelian texts. I will present this reading more or less dogmatically, without addressing alternative interpretations or scrutinizing texts in detail. My excuse for this procedure is that my interest is not finally historical anyway: I am interested in bringing out a point of view that has a plausibility in its own right and a bearing on contemporary debates. I want to explore three key ideas belonging to the Classical View: first, the idea that “rational animal” belongs to the specification of the essence of humankind; secondly, the idea that, more specifically, this phrase characterizes our form; and finally, the idea that “rational” designates a characteristic that differentiates the genus “animal.” These ideas are often treated as elements of an alien and antiquated metaphysical outlook, one that modern philosophy has proved to be unjustifiable and that modern science has shown to be superfluous. I will suggest, however, that they can be understood in a way that makes them neither antiquated nor alien: as characterizing the distinctive categorial framework in which we must understand claims about the central powers and activities of a certain sort of living thing – a framework we constantly presuppose when we think about persons and their activities, and one whose soundness few philosophers seriously question. A crucial implication of the Classical View, I will argue, is that rationality is not a particular power rational animals are equipped with, but their 6

I discuss the attribution of this definition to Aristotle more carefully below.

400

Matthew Boyle

distinctive manner of having powers. Appreciating this idea will, I think, allow us to avoid various dead ends and false dilemmas at which discussions in contemporary philosophy of mind recurrently arrive. In particular, I will argue that it enables us to avoid the following choice, which many authors take to be mandatory: either offer an account of cognition and action that applies uniformly to both rational and nonrational animals, or else deny that nonrational animals can literally be said to cognize and act. Furthermore, I will suggest that appreciating this idea puts us in a position to answer the common charge that conceiving of our minds as essentially rational involves a hyper-intellectualized or hyperidealized view of how our minds operate. I will draw out these consequences in §3. First, though, I need to describe the Classical View. 2.2 On the Classical View, although the concept rational undoubtedly has other applications, it appears primarily as part of a definition of a certain kind of living creature. Such claims as that a certain individual has judged or acted rationally or irrationally, or that certain sorts of activities (e. g., choosing, inferring, or deliberating) are exercises of rational capacities, involve a concept whose significance must be explained by relating it to the more basic idea of a rational animal.7 The kind whose definition is of primary concern to us is of course our own kind: human beings, or in an older idiom, “man.” To say that “rational animal” belongs to the definition of man is to say that it belongs to the specification of what it is to be a human being. Aristotle is in the habit of nominalizing this phrase, so that he frequently speaks of “the what-it-is-to-be” (to ti e¯n einai) of a thing, and this is what comes, by way of Latin translation, to be known as its “essence.” A definition explicates what it is to be a certain kind of thing, and Aristotle famously suggests that this explication should take the form of a specification of a genus under which that thing falls, qualified by some “difference” or “differences” that distinguishes its particular species within the genus.8

7 8

The occurrence of “rational” in “rational animal” thus gives what Aristotle scholars commonly call the “focal meaning” of this term: the primary meaning in relation to which various other senses of the term are to be understood. On definition by genus and difference, cf. Topics I. 5 and VI. 4, and Parts of Animals, I. 2 – 3. Cf. also Metaphysics VII. 12. All quotations from Aristotle in the text are from the translations in Aristotle (1984).

Essentially Rational Animals

401

It is sometimes suggested that “rational animal” just is Aristotle’s definition of man9, but I can find no place where he says this, and a number of places where by implication he seems to deny it. It is true that he often suggests that reason (or speech, thought, or some other capacity he regards as characteristic of rational creatures) is unique to human beings; but that is not necessarily to claim that “rational animal” gives a sufficient account of what it is to be a human being, which is what an Aristotelian definition is supposed to do. A kind’s essence, its what-itis-to-be, is not merely supposed to be some property that uniquely characterizes it: thus “featherless biped” may be a predicate that characterizes human beings uniquely, but it does not ipso facto describe our essence. And there are various indications that, although Aristotle does think that human beings are essentially rational animals, he does not think “rational animal” exhausts our essence. One is that, when he discusses the task of defining “man,” he frequently mentions properties (e. g., “twofooted”) that would seem to belong to a specification of what particular sort of rational animal a man is.10 Another is that, in the De Anima, rational is introduced as a kind of soul on a par with vegetative and animal; but obviously the latter two are not differentiae of some particular species of life, but of whole categories of living things, of which there can be many particular species. Similarly, the concept rational animal seems to be such that other species at least could fall under it. But if that is right, then what it is to be a man must be distinguishable from what it would be to be one of these other possible species, and since by hypothesis they would also be rational animals, “rational animal” cannot be a complete characterization of what it is to be a man.11 2.3 To understand the sense in which being a rational animal nevertheless partially characterizes the essence of man, we first need to consider what it is to characterize the essence of something. As a first approximation, we might say that to characterize the essence of an individual thing 9 See for instance Cohen (2009), §9. 10 See. e. g. Aristotle (1984), Categories 5, 3a21 and Aristotle (1984), Metaphysics VII. 12, 1037b9 – 12. 11 This certainly seems to have been how Aristotle was read by many medieval commentators. Thus Aquinas says that “to man as man belong rational, animal, and whatever else his definition includes” (Aquinas (1949), Part III, Para. 3, emphasis added), and Porphyry writes that “[r]ational animal is a species of animal and a genus of man. Man is a species of rational animal” (Porphyry (2003), §2, 4.30 – 31).

402

Matthew Boyle

is to specify what its “being” comes to – i. e., to specify the concepts that would have to be involved in an adequate explanation of the nature of its existence. Aristotle’s thought is that not everything which can be said truly about a particular individual belongs to such a specification. Only a certain sort of predicate, one that qualifies what he calls the “substance” of a thing, belongs to the specification of its essence. A thing’s substance is what it must be if it is to be at all. It is designated by some fundamental sortal predicate which the thing must bear at all times when it exists. For you and for me, the relevant predicate is: “human being.” I may at a certain point in my life become a father, or be for a period an officer in the Navy, but I never became a human being (for I did not exist before I was one), and I could not cease to be one without ceasing to be, period. Furthermore, although, having been born and raised in the United States, I always was and always will be an American, still American is not a concept that must be invoked in explaining what it is for me to exist as an individual subject of predications. Thus I could informatively explain what an American is by saying: it is a human being who was born and raised within the territory of the United States. But I could not in a similar way explain what it is to be a human being by saying that it is an individual instantiating some more fundamental sortal concept, which qua individual is subject to certain further determinations. For what would the sortal concept be? It would have to be a kind such that there could be one of those which as a matter of fact possessed certain further properties. But sortal predicates more abstract than “human being” do not seem to characterize kinds of which there could be one just as such. “Mammal,” for instance, is not a predicate that can be instanced by something that is as a matter of fact “configured human-being-ly” but might conceivably have been configured in another way so that it – that very same individual – would have instead have counted as a horse. The concept mammal does not by itself suffice to sustain the idea of an individual which might be configured this way or that: there can of course be three mammals in the room, but only because there are three individuals belonging to concrete mammalian species. The possibility of there being three human beings in the room does not rest in a similar way on the possibility of there being three humans of determinate nationalities.12 12 See especially Categories 5 and Metaphysics VII. 4. For a recent defense of such a standpoint on individuation, cf. Wiggins (2001).

Essentially Rational Animals

403

These claims might of course be disputed, but they are not obviously indefensible, and I take them to be core commitments of the Classical View. On the Classical View, the concept human being is the basic concept of the kind of thing I am: for it is by being this kind of thing that I exist at all, and so the applicability of any other description to me rests on the applicability of this description. Being a human being is therefore an irreducible property of the individuals that bear it: we cannot say what it is to be a human being by specifying other more basic features individuals might have that would make them human beings. Nevertheless, Aristotle holds that there is another sense in which we can (at least in principle) explicate what it is to be a human being: we can give a definition of human being. To say what it is to be a human being in this sense is not to specify a more fundamental kind that individuals might fall under plus some features which would make such individuals count as human beings: that, as we have seen, is something the Classical View holds to be impossible. To say what it is to be a human being is not to describe properties of individuals that make them count as human beings, but rather to characterize the nature of the kind human being itself, the kind of thing in virtue of being which you and I are particular individuals at all. The predicates that appear in a definition of man thus attach, not primarily to individuals of that kind, but to the kind itself, and their applicability to individuals is always in an important way mediated by this more primary application.13 This is a difficult idea, but I think we can come to understand it better by reflecting on a point that Michael Thompson makes in his important paper “The Representation of Life” (1998). Thompson points out that we are all familiar with a certain mode of description of living things which is not a description of them as individuals. It is a mode of description that is familiar, for instance, from nature documentaries: “The grizzly bear digs a den under rocks or in the hollow of a tree, or in a cave or crevice. It goes into its den between October and December and stays there until the early spring. It has a protective layer of fat that allows it to stay in its den while the weather is cold. It does not really 13 For these doctrines, see especially the discussion of things said of a substance “in respect of itself” (or “per se”) in Aristotle (1984), Metaphysics VII. 4: “The essence of each thing is that which is said of it in respect of itself. For being you is not being musical, since you are not by your very nature musical. What, then, you are by your very nature is your essence” (1029b13 – 1029b15). Compare also Aristotle (1984), Categories 3 – 5 and Aristotle (1984), Metaphysics X. 9.

404

Matthew Boyle

hibernate and can easily be woken up in the winter…” These sentences describe, not what this or that grizzly bear does (indeed, this one may fail to make a den, and that one may fail to go into it at the standard time) but what is done by “the grizzly bear,” or by grizzly bears in general – where “in general” is heard in a special register. These sentences do not necessarily describe what holds of most grizzly bears: it may be, for instance, that, given human encroachment on their habitat, most actual grizzlies are not in a position to build up the layer of fat that allows them to survive the winter. Even so, it would be a true description of how “the grizzly bear” lives to say it goes into hibernation with a protective layer of fat. This truth seems to belong to a story about how things are supposed to go for grizzlies: a system of judgments constituting a teleologically-structured story about how they get by in the world. Recognizing this, we might try saying that the sentences describe how things “normally” or “properly” go for grizzly bears. But, as Thompson persuasively argues, this is true only if “properly” means something like “if things go right with respect to being a grizzly bear” – and then determining when this condition is met simply returns us to the sentences whose truth-conditions we were trying to understand. Thompson concludes – after showing the futility of various other proposals about how to cash out the truth-conditions of such sentences in terms of truths about individuals of the kind in question – that these truths are exactly what they appear to be, namely truths about “the grizzly bear” (or other concrete kind of living thing). They are truths whose natural expression takes the form of sentences linguists call “generics,” sentences of the form: “Ss are/have/do F” or “The S is/has/does F.” In general, it is notoriously difficult to give an account of the truth-conditions of such generic propositions in terms of the truth-conditions of sentences about propositions of kind S. I think Thompson argues convincingly that, at least in the case of generics that characterize the natures of living things, there could not be such an account, but I will not try to present his argument here. Instead I will simply observe that, even if some account were discovered that proved to be extensionally correct, it would be implausible to suggest that our understanding of the truthconditions of such sentences depends on our understanding specific principles connecting them with truths about individuals of the relevant kinds. Our grasp of the relevant claims seems to be a grasp of predicates applying directly to the kind, not mediately to the kind in virtue of their application to individuals of that kind. And that, in effect, is the Aristo-

Essentially Rational Animals

405

telian thought about the predicates that appear in a definition of a certain substantial kind: they state, not features that individuals must have if they are to belong to that kind, but rather attributes that directly characterize the nature of the substantial kind itself.14 Thompson’s paper also brings out the force in the idea that the application of such predicates to individuals of a given kind is mediated by their application to the kind itself. For, he argues, when I take a structure in a particular organism to be a wing, or a tooth, or when I take a certain chemical process to be a part of its digestion, or indeed when I make any judgment which implies that the subject is alive, I implicitly commit myself to various assumptions about the function of such features or occurrences in the life of that kind of creature. To identify this excrescence as a wing rather than some sort of deformity is already to see it as a case of this organism’s being as it is in the nature of this kind of thing to be: it is to see this aspect of its shape as a realization of a way of being shaped that has a function in the life of this kind of creature, a way of being that came to exist here – to the extent that it did – because this is the way they are. The description of the features and activities of individual living organisms thus goes with a way of seeing those individuals in which what they are like is characterized in terms of its (more or less perfect) realization of potentialities, and in which happenings in which they are involved are characterized as (more or less successful) acts of powers, where such potentialities and powers are attributable to individuals only in virtue of their being instances of a certain kind of living thing. And this mode of description is no mere superficial addition to our understanding of living things. As Thompson observes in another paper: Even such apparently purely physical judgments as that the organism starts here and ends here, or weighs this much, must involve a covert reference to something that goes beyond the individual, namely its life form. It is only in the light of a conception of this form, however dim that conception might be, that you could intelligibly suppose, for example, that the[se] tentacles are not parasites or cancerous excrescences or undetached bits of waste.15

But to say that judgments about individuals refer in this way to “life forms” is to say that they refer to the teleologically-organized system of generic judgments that characterize the life cycle of that kind of 14 The connection between generic propositions and the Aristotelian notion of essence is also noted in Moravcsik (1994), another paper to which I am indebted. 15 Thompson (2004), 52.

406

Matthew Boyle

thing (the grizzly bear, the horseshoe crab, etc.).16 Thus, if Thompson is right, predications that ascribe vital characteristics to particular organisms are mediated by predications holding of the substantial kind per se, just as the Classical View would suggest. 2.4 It is worth emphasizing how different this way of understanding the idea of an essential property is from standard contemporary approaches to the topic.17 The dominant contemporary understanding of the notion of an essential property explicates this notion in modal terms. On this view, x is essentially F just in case it is a necessary truth that if x exists, it is F, i. e.: (N)

(9y(y=x) ! Fx)

&

The Aristotelian conception of essence presented above gives a more complex account of the relation between an individual and the properties that characterize its essence. The account involves a distinction between two questions: (1) What is the substantial kind to which an individual belongs? (2) What is the nature of that kind? Question (1) is answered by a term S that designates the substantial kind to which the individual belongs, as “human being” designates our substantial kind; and it is indeed the case that, on the Aristotelian view, individual Ss can only exist at all in virtue of being Ss.18 Question (2), however, is answered by a definition that explicates what it is to be an S, and the traits mentioned in this definition, although they characterize 16 This obviously need not imply that a given judger’s conception of that system is complete or even correct. But in judging that a certain vital predicate applies to a certain organism, I commit myself to assumptions about the shape that system takes. 17 I am grateful to Dorit Bar-On for pressing me to address this issue. 18 It is not clear, however, that this point is well expressed by saying, as (N) does, that such individuals are human beings in any possible world in which they exist. It is open to question whether a possible worlds framework adequately captures the relation between existing and being-a-such-and-such that holds here. Indeed, if being human beings is our way of being actual existents at all, then it is open to question whether our being human beings is well represented as a case of our having a certain property at all. I cannot pursue these questions here, however. Even if being a human being can be treated as an essential property in the manner of (N), there are further features of the Aristotelian standpoint that decisively differentiate it from this approach.

Essentially Rational Animals

407

what it is to be an S, will not necessarily be possessed by every individual that is an S. The account will thus yield propositions of two importantly different types: (E1) The essence of x is to be an S. (E2) Ss are essentially F. Only propositions of form (E1) directly concern individuals; propositions of form (E2) are self-standing generic propositions that characterize a substantial kind as such – though of course they are connected with propositions characterizing individuals of that kind in the complex ways we have been considering. It is not immediately clear how to render propositions of form (E2) in modal terms. A simple proposal would be: (N2)

&

(x)(Sx ! Fx)

That is, in all possible worlds, if something is an S, then it is F. (Other readings of the claim can be constructed by placing the necessity operator in other positions.) But the immediate difficulty for (N2) is that it plainly fails to capture what Aristotelians mean by saying, e. g., that human beings are essentially two-footed. For this is consistent, as we have noted, with the existence of individual human beings who do not have two feet. Having two feet might belong to the characterization of what it is to be a human being, yet I, who am a human being, might lose a foot without ceasing to exist (and without ceasing to be a human being). So the claim that human beings essentially have two feet does not imply a modal proposition of form (N2). And it is hard to see how any variant of this proposal can escape the objection, so long as it retains the ambition of reducing claims of form (E2) to claims about what is necessarily true of individuals.19 19 Similar points apply to the interesting nonstandard treatment of essence proposed by Kit Fine, cf. Fine (1994), (1995a), (1995b). Fine does not attempt to reduce claims about essence to modal propositions: he treats “Essentially” as a primitive operator on propositions, and seeks to explain modality in terms of essence. Nevertheless, for Fine, “[a] property of an object is essential if it must have the property to be what it is” (Fine (1995a), 53); and this leads him to adopt an axiom to the effect that a proposition which truly characterizes the essence of a certain kind of thing must also be true simpliciter (his axiom &F A ! A: cf. Fine (1995b), 247 – but note that the interpretation of this axiom is complicated by the fact that Fine’s “&F” is not a function applying to predicates but an operator on propositions, which Fine introduces by stipu-

408

Matthew Boyle

This fact is a reflection of the distinction between descriptions which apply primarily to a substantial kind per se and descriptions which apply primarily to particular individuals of that kind, a distinction which is crucial to the Classical View but which contemporary treatments of essence generally do not draw. On the Classical View, propositions about the essential features of human beings are propositions about the kind human being itself, and there is no immediate inference to be drawn from such truths to freestanding propositions about what particular individuals of this kind are like. This is not, of course, to suggest that truths about the kind and truths about individuals of that kind are simply unconnected: they are connected inasmuch as the truths about the kind describe how things go for individuals of that kind if nothing interferes. But to allow for the possibility of interference is to allow for the possibility of exceptions which do not disprove the rule. 2.5 Having said this much about living things and their essences in general, we can return to rational animals in particular. We noted earlier that “rational animal” seems on the Classical View to be only a partial characterization of the human essence. It has the form of a specification by genus and differentia, but the kind of thing it specifies seems to be still generic with respect to human beings. What then does it characterize? Does it, like “mammalian,” merely pick out a set of traits that certain animal species exhibit? Aristotle seems to regard the rational/nonrational contrast as marking a deeper sort of distinction than that. One indication of this is the fact, mentioned earlier, that he treats rational as on a par with vegetative and animal – as designating one of the three fundamental kinds of soul. But what is a kind of soul?

lating that the proposition “&F A” means roughly “The proposition A is true in virtue of the nature of objects which are F”). Hence, in Fine’s system, the claim that it belongs to me essentially as a human being that I have two hands will presumably be schematized as &human being I have two feet. And this will imply I have two feet. But on the understanding of essence proposed here, it might be true that human beings essentially have two feet, and thus true to say of me that, as a human being, it belongs to my essence to have two feet; and yet this does not rule my actually failing to have two feet, whether through defect of birth or misadventure.

Essentially Rational Animals

409

To answer this question, we must recall some points about how Aristotle explains the notion of soul in general. Aristotle famously thinks of living things in hylomorphic terms, as cases of matter of a certain sort bearing a certain form. A soul, for Aristotle, is the form of a living thing: it is that structuring principle in virtue of which matter of a certain sort constitutes a living thing.20 To be “ensouled” is to partake of the mode of organization characteristic of life. What sort of organization is that? Aristotle holds that this question cannot be answered in a completely general way. There are three different ways of being a living thing, ways which are not just unrelated, but which are not definable by reference to a single abstract schema. Rather, these three modes of life are in a certain way successive – not in their order of appearance in the world, but in the way they are defined. To understand what an animal soul is requires understanding how animality transforms the mode of organization characteristic of a nutritive soul, the type of soul that appears primitively in plants; and to understand what a rational soul is requires understanding how rationality transforms the mode of organization characteristic of an animal soul.21 This implies, on the one hand, that the idea of being rational has content only in virtue of building upon the idea of being an animal, a living thing capable of negotiating the imperatives of its life by exercising the powers of perception and desire-governed action. But it also implies, on the other hand, that what it is to be an animal is fundamentally transformed where rationality is present. “Rational” counts as a differentiating predicate of “animal”, rather than merely as the name of a trait that certain animals exhibit, in virtue of the fact that what is rational differs in its way of being an animal from what is not. Thus Aristotle explains the notions of genus and differentia as follows: By genus I mean that one identical thing which is predicated of both and is differentiated in no merely accidental way… For not only must the common nature attach to the different things, e. g. not only must both be animals, but this very animality must also be different for each… For I give the name of ‘difference in the genus’ to an otherness which makes the genus itself other.22 20 Cf. Aristotle (1984), De Anima II. 1, 412a16 – 21. 21 Cf. Aristotle (1984), De Anima II. 2 – 3, which argues that “life is spoken of in many ways” (413a22) and therefore that “we must inquire in each case what is the soul of each thing, what is that of a plant, and what is that of a man, or a beast” (414b31). 22 Aristotle (1984), Metaphysics X. 8, 1057b38 – 1058a7.

410

Matthew Boyle

If being rational did not transform what it is to be an animal, a discussion of rationality would not need to appear in a general account of what it is to live: “rationality” would just be a characteristic of certain animal species, which are living things in whatever sense any animal species is a living thing. This, presumably, would be Aristotle’s attitude toward “mammality”. But it is not his view of “rationality”: he holds that being rational transforms the nature of being an animal, and thus constitutes a new way of being a living thing. And since “for living things, to be is to live,”23 this implies that a rational animal has a distinctive form of essence, a distinctive type of “what-it-is-to-be.” This is the thought at which I have been aiming to arrive. Let me clarify how I understand it. We were asking what kind of distinction the rational-nonrational distinction is supposed to be, and in what sense it is supposed to be a deep distinction, one that makes for a difference between kinds of minds. We have arrived at the following conclusions. First, on the Classical View, “rational” does not differentiate us merely because it names a trait that happens to be unique to humans. It belongs, rather, to a characterization of our essence, which is to say, to an account of what our existing as particular individuals comes to. As such, it is predicable of us as individuals only in virtue of being predicable of the substantial kind to which we belong (namely: human being). Furthermore, it characterizes this kind not in the way that concrete descriptive predicates like “mammalian” or “two-footed” do: it does not specify the specific content of our essence, but the form of essence that we have. We could put it this way: “rational” specifies the sort of frame that undergirds any concrete description of what it is to be a human being. For it does not specify a particular characteristic that we exhibit but our distinctive manner of having characteristics. This, I believe, is the significance of saying that “rational” characterizes the form of human being. A substantial kind is, as we have seen, the subject of which essential traits are predicated; and where we have a different form of kind, the predicates that characterize that kind per se, and in consequence the predicates that apply to individuals only insofar as those individuals belong to that the kind, will admit of a different sort of significance than they would have in application to kinds with other forms. The predicate that differentiates one form of life from another thus does not name a concrete characteristic (or set of characteristics) that certain species of living 23 Aristotle (1984), De Anima II. 4, 415b13.

Essentially Rational Animals

411

things possess; it marks the possibility of a different form of predication of vital characteristics in general. 2.6 I have stated this point very abstractly. Some examples will help to make its significance clearer. Let us start with a simpler case, namely the difference between a plant and an animal – for this too is a difference of form, on the Classical View. Consider what it means to talk about activity in the case of a plant and in the case of an animal. Many people will be tempted to say that plants do not act at all, and there is of course a sense in which that is right: they do not act in the sense in which animals act. Nevertheless, there are clearly some episodes in the lives of plants in which they figure as agents rather than as mere patients. When a tree is chopped down, it is a patient; but when it grows a new branch, or flowers in the spring, it is an agent: the latter are things the tree in some sense does, not things that are done to it. Now, the opposition between agency and patiency depends in general on where the primary explanation of the relevant event or process lies. Thus growing a new branch counts as something the tree is doing because, although various environing circumstances may facilitate this happening, the primary explanation of it is to be sought simply in the nature of the tree itself (as characterized in a system of generic propositions of the sort discussed earlier). The tree’s growing a new branch is thus in a broad logico-grammatical sense its own act, one expressible in an active-voice progressive judgment that has the tree as its subject. And the act is even goal-directed in a clear enough sense: trees grow branches precisely because having branches permits them to have an extensive canopy of leaves that absorb sunlight. Nevertheless, when we speak of the goal-directed acts of an animal, we are clearly speaking of agency and goal-directedness in an altogether different register. It is not merely that an animal can do more than a plant; it is that talk of “doing” can apply in a wholly new way to an animal.24 I think this is intuitively clear, but one way to see the underlying basis of our intuition here is to reflect on the fact that descriptions of the here and now can never enter into the characterization of the “acts” of plants except in the form of triggerings, helps or hindrances. The root of a tree can be growing round a stone, but it would be at 24 Thus Aristotle says that plants have “movement in respect of growth and decay” but not “movement in respect of place,” which is a different form of movement: see especially Aristotle (1984), De Anima II. 4 and III. 9.

412

Matthew Boyle

best sentimental to suggest that the root is growing in a certain way in order to get round this stone. The presence of this stone here and now does not inform the content of the tree’s act of root-growing, that toward which it is goal-directedly tending. The tree’s roots simply grow, as far as possible, according to a certain pattern: the stone enters as a hindrance to this growth, something that interferes, and hence qualifies the sense in which the shape of the resultant growth can be understood as the tree’s own doing, rather than as a reflection of something done to it. An animal, by contrast, can act with respect to the here and now: descriptions of present circumstances can enter into the content of what it is doing. Its capacities for perception and desire transform its mode of being alive precisely because they make this possible: they open animal life, not merely to the causal influence of present circumstances in the form of triggering, hindrance, or facilitation, but to the kind of influence that enters into the constitution of what the subject is doing. Thus an animal can try to get that object, or do something in order to avoid this obstacle. The thing that can fill the “A”-slot in “S is doing A” will thus be an “A” of a fundamentally different kind where the subject in question is an animal. It will be, not merely a type of content that adverts to some generic form of activity (growing a branch, flowering, etc.), but a type that embraces particularity within itself: the general sort of thing the animal is doing (hunting something, fleeing something, playing with something, etc.) has – to borrow a Fregean phrase – an unsaturated position, one that waits for perception (or, in more sophisticated animals, memory, imagination, etc.) to fill it. The sorts of things that plants can do contain no such gaps. And it is just another aspect of the same point to say that animals act as individuals in a way that plants do not. A particular oak can be growing a new branch, but the explanation of its doing so does not really look to it in particular: it is just doing the kind of thing that all healthy mature oaks do at this time of year, given enough sunlight, enough water, etc. It has not determined that this should occur, although various things about it and its circumstances certainly affect whether the general disposition of its kind will be realized here and now. Nevertheless, the fundamental explanation of its act is just a certain generic fact about its kind: this is what they do. The here and now enters only as trigger, help, or hindrance. To understand why an animal is doing what it is doing, by contrast, we have to look to this particular one, for what it pursues is not simply fixed by the laws of

Essentially Rational Animals

413

its kind, but involves facts about the experience of this one in particular.25 I have been describing the differences between plant and animal activity in some detail in order to bring out what it might mean to say that, in contrast to a predicate like “mammalian” or “land-dwelling,” the predicate “animal” does not name a concrete characteristic exhibited by certain kinds of living things, but rather marks the possibility of a distinctive form of predication of characteristics to such things. My aim has been to suggest that, although both plants and animals can be said to do things, can be characterized as agents of some of what happens to them, can perfectly correctly and literally be said to pursue goals, etc., nevertheless the manner in which such predications apply is different in the two cases. It is not merely that animals can do things that plants cannot; it is that the whole language of “doing” takes on a new significance, a new logical character, when we turn from plants to animals. I think similar points could be made about predications in each of the various Aristotelian categories: an animal admits, in a manner fundamentally different from a plant, of, e. g., having something, being acted on, being qualified in a certain way, and even of being in a certain place and of being one thing.26 25 If the depth of this difference does not seem evident, it may help to reflect on the distinctive kind of failure that animal agency makes possible. If a cat is chasing a certain mouse and does not catch it, but manages as it happens to pounce on a different mouse, then although the cat has in one sense got what it was after (namely, a mouse), there is clearly another sense in which it has failed to get what it was after (namely, this mouse, the one it was chasing). By contrast, if a tree’s roots are taking up water, and we somehow contrive to replace the water molecules that would ceteris paribus have been taken up with other water molecules, then although the circumstances of the tree have changed, there is surely no purpose belonging to the tree which it has failed to achieve. Plants, as plants, simply do not engage with the particular things present here and now in this sort of way. (This is true even of the Venus Flytrap, which reacts to stimuli in a way that bears an uncanny resemblance to animal percipience. The Flytrap is triggered to activity by some particular fly’s touching its trigger-hairs, but then it simply does what all such plants do when those hairs are touched: it snaps shut. The snapping is triggered by a particular fly, but the act itself instances a completely generic mode of activity which is not aimed at, but merely occasioned by, this fly in particular.) 26 This across-the-board transformation is a reflection of the primacy of the category of substance (cf. esp. Aristotle (1984), Metaphysics VII. 1): where the type of substance is transformed, the significance of predications in other categories undergoes a correlative transformation. Thus where two types of substance are generically alike but specifically different (e. g., both living but only one ani-

414

Matthew Boyle

This is not to suggest that every predicate in each of these categories applies differently to animals than to plants: I take it that there is no fundamental difference between, say, talking about the weight of a plant and the weight of an animal, or between talking about the color of a plant’s leaves and the color of an animal’s fur. My claim is not that animals do not admit of predications of the same character as those that apply to plants, but that they do in addition admit of predications of a distinctively different character, and that this is the kind of difference that the term “animal” marks. In a fuller discussion, I would also want to argue that these distinctive predications characterize the core of what it is to be an animal; for it is only in virtue of what an animal has, does, and is in this distinctive register of having, doing, and being that it exists as a particular individual at all, and hence its bearing these predicates is the principle of its bearing whatever other predicates it bears (weight, color, etc.). But to develop this idea would take us too far afield; the crucial point for present purposes is that the animal/nonanimal contrast differentiates two forms of life: it differentiates living kinds, not merely in respect of certain particular characteristics they possess, but in their whole manner of having characteristics, the form that predications of being, having and doing can take for them. Likewise, on the Classical View, the rational/nonrational contrast marks this sort of difference. A rational animal is capable, not just of being, having, and doing more than a nonrational creature, but of being the subject of ascriptions of being, having, and doing in a distinctive sense. Consider action once again as an illustration. It is perhaps even more obvious in this case that the generic notions of being an agent, doing something, and pursuing a goal apply to both rational and nonrational animals. What is perhaps less obvious is that agency, doing and goal-directedness take a different form in the rational case. Nevertheless, it is widely recognized that there is a sense of “doing something” that applies only to rational creatures: we are the only creatures that act intentionally. Furthermore, it is widely conceded that a condition of the applicability of ascriptions of doing in this distinctive sense is that the creature should be doing what it is doing knowingly, in virtue of exercising its power to determine what ends are worth pursuing and how to pursue them. On the Classical View, the power to act mal), this introduces the possibility of predications in the other categories that are also generically similar but specifically different (e. g., both of vital activity but only one of animal action).

Essentially Rational Animals

415

in this distinctive sense – to engage in doings whose ascription implies that the subject knows what he is doing and what for – is the special prerogative of rational creatures. Thus Aquinas holds that although nonrational animals can be said to intend an end and act voluntarily in pursuit of it in “an imperfect sense,” they are not capable of intention or voluntary action in “the perfect sense,” since they do not ordain their movement to an end in virtue of knowledge of that end “under the aspect of an end”. Rather, they merely apprehend an object they desire and act from instinct or acquired habit in pursuit of it.27 Some will want to argue that various species of nonhuman animals are capable of more than this. I take no position on this issue. My aim here is not to argue for a certain classification of this or that species of living thing, but to point out a kind of distinction that seems at least intelligible, whatever one thinks of its application in particular cases. If the distinction is intelligible, then talk of animal “doing” (and relatedly, of animal agency, responsibility, pursuit of an end, etc.) admits of two different registers, one nonrational and the other rational.28 I think it would be possible to show that this distinction corresponds to a difference in the form of the “A” that can be the content of rational doing, and in the manner in which the predicate “is doing A” attaches to an individual rational subject. But to develop these differences would involve beginning to give a substantive theory of rationality, and that is not my purpose here. My aim is just to point out the possibility of a certain sort of conception of the difference that rationality makes, one that would give sense to the idea of a different kind of mind. The quotation from Herder at the head of this paper captures the crux of this conception: the idea is that our rationality does not merely increase the extent of our ability to do things in the sense that nonrational animals can already be said to do things. Nor again is “reason” the name of a particular power – like sight or hearing or the power to 27 Cf. Aquinas (1948), IaIIae, Q. 6, A. 2 and Q. 12, A. 5. 28 This is how I understand G. E. M. Anscombe’s cryptic claim that the term “intentional” does not name “an extra feature” that accompanies certain actions but rather “has reference to a form of description of events” (cf. Anscombe (1957), §§19, 47). Her aim is to argue that “intentional action” does not simply pick out a certain class of events but a certain distinctive type of event-predication, one that can only apply to creatures who can be the subject of the special sort of “why?”-explanation that she identifies, one whose application presupposes that the subject in question itself knows what it is doing and why, and acts precisely in virtue of that knowledge.

416

Matthew Boyle

walk on two legs – which enables us to do some specific sort of thing, but to do it in the same general sense of doing that applies to the powers from which it is differentiated. Rather, as Herder puts it, “reason” names “the whole organization of all human powers,” an organization which determines “a quite different sort of orientation and unfolding of all powers.” What he means, I think, is that our rationality transforms the sense in which powers and their corresponding acts are ascribable to us. It marks a new form of power- and act-predication, in the sense I have been trying to explain. And as a characterization of our essence, it implies that reference to powers of this distinctive form belongs to an account of the sense in which we exist, as individual subjects of predication, at all.29

3. Applications 3.1 This reconstruction of the Classical View will have been worthwhile if it helps us with difficulties that face us here and now. I believe it does. Let me conclude by mentioning some common objections to the idea of a difference in kind between rational and nonrational minds which the foregoing reflections help us to answer. The idea that reason brings with it a new kind of mind, one that admits a distinctive form of predication, is not without recent defenders. Something like this view has been defended, for instance, by Donald Davidson, who famously claimed, first, that to understand our kind of mind, we must focus on a certain class of predicates, namely those that ascribe so-called “propositional attitudes”; and second, that the application of such predicates is governed by a “constitutive ideal of rationality”30. To claim that rationality makes a constitutive difference to the kinds of predicates we are capable of bearing amounts, I take it, to claiming that we are not merely capable of representing and doing more than nonrational animals, but that we can figure as the subject of predications of representation and action of a distinctive form, a form that applies to us only in virtue of our rationality. Now, Davidson’s claim has been influential, but it has also faced various recurrent 29 A distinction whose most immediate effect is, as Herder suggests, on our sensing, cognizing, and willing nature. These are the types of predication most immediately affected because they are the types specific to animality, which is the genus of substantial being that is differentiated by rationality. 30 Davidson (1980), 223.

Essentially Rational Animals

417

objections. We might think of these as falling under two heads: (i) objections to the idea that rationality can make a constitutive difference to the kind of representing and acting of which we’re capable, and (ii) objections to the idea that reference to a mere ideal of rationality can play a crucial role in determining what is actually true about us. I want to say something about each of these sorts of objections, and how the foregoing discussion bears on it. 3.2 One common objection to views like Davidson’s begins from the observation that, except when we are defending a philosophical view that requires us to say the contrary, we all take it for granted that many kinds of nonhuman animals can believe things about their environment, can learn from past experience, and can act intelligently in pursuit of things they desire. All of these descriptions seem to apply perfectly literally to nonhuman animals, and they seem to figure in genuine explanations of how they behave. And this impression is only reinforced by rigorous studies of animal behavior. It would thus be perverse – so the objection goes – to deny that nonhuman animals can believe, desire, and so on; but this is exactly the sort of perversity involved in the claim that only rational animals can have propositional attitudes. If the literal application of propositional attitude ascriptions presupposes rationality, then rationality must be present quite generally in the animal kingdom; while if rationality is something special to human beings, it cannot be presupposed in the application of propositional attitude talk. At any rate, the objectors conclude, if there is a difference between our representing the world and that of “lower” animals, it must be merely a difference in the sophistication of the representational contents we can entertain and the complexity of the operations we can perform on them, not a distinction between altogether different kinds of representational states. I think this objection rests on an assumption that the foregoing discussion has given us the resources to question. The assumption is that a psychological or epistemic concept which applies both to rational and to nonrational animals must be susceptible of a single, undifferentiating account that covers both sorts of application. We might call this the Univocality Assumption, for it amounts to the claim that such concepts must be treated as univocal in their application to rational and nonrational animals. This assumption manifests itself in the frequently heard insistence that an account of belief, warrant, knowledge, etc. must not make any demands that a nonrational animal could not meet, since nonrational an-

418

Matthew Boyle

imals plainly hold beliefs, possess warrant for their beliefs, have knowledge, and so on. It should be clear that this inference is valid only given the Univocality Assumption, for only if an account of these concepts must not differentiate between rational and nonrational animals does the fact that we speak of nonrational animals as holding beliefs, being warranted, etc., show that our account of the application of such concepts to humans cannot make demands that a nonrational animal could not meet. Must we make the Univocality Assumption? In his recent “Perceptual Entitlement,” Tyler Burge writes: Children and higher nonhuman animals do not have reasons for their perceptual beliefs. They lack concepts like reliable, normal condition, perceptual state, individuation, defeating condition, that are necessary for having such reasons. Yet they have perceptual beliefs. There is no sound basis for denying that epistemology can evaluate these beliefs with respect to norms governing their formation, given the perspectival limitations and environmental conditions of the believer. There is no sound basis for denying that epistemology can evaluate their perceptual beliefs for epistemic warrant.31

I think there is a reading of what Burge says here on which it is undeniable: nonhuman animals patently respond to the world on the basis of representations of what is the case, representations that we have every right to call “beliefs,” representations concerning which we can certainly raise questions of warrant. But Burge makes these points in the context of attacking the thesis – which he associates with authors such as Sellars, Davidson, and McDowell – that the kind of warrant that a rational creature has for its beliefs must be a warrant that “lies within ‘the space of reasons’.” Indeed, he writes as though these points themselves constituted a refutation of the thesis: if nonhuman animals can have beliefs, and be warranted in having them, then, Burge reasons, being warranted in a belief cannot depend on capacities that nonhuman animals do not possess.32 But surely this inference reflects a blinkered view of the options. Whatever exactly it means to claim that a rational creature’s warrant must “lie within the space of reasons,” a sensible defender of this 31 Burge (2003), 528. 32 Burge’s views on these topics deserve a much fuller discussion than I can give them here. I quote him simply as exemplifying a widespread readiness to assume the univocality of various important cognitive concepts across the rational-nonrational boundary. For fuller presentation of Burge’s position, cf. Burge (2010). I hope to address Burge’s views in detail in future work.

Essentially Rational Animals

419

claim should not hold this to entail that nonhuman animals cannot have perceptual beliefs, or be warranted in having them. He should hold, instead, that the concepts of belief and warrant have a different application here from the one they have in connection with nonrational creatures – that with the advent of reason comes a new form of belief and new standards of warrant associated with it. To hold that concepts such as belief and warrant apply in one way to rational creatures and in a different way to nonrational animals need not be to suggest that we are merely being ambiguous when we speak of “belief” and “warrant” in connection with creatures of both kinds. It might be rather to claim that merely animal belief and rational belief, merely animal warrant and rational warrant, are different species of the same genus. And as we have seen, the Classical View holds that the rational-nonrational distinction allows for precisely this combination of generic similarity and specific difference in the way basic types of predicates apply. Thus, just as both rational and nonrational animals can be said to act, although the idea of action is applied in a distinctive register in the rational case, so too it might be that both rational and nonrational animals can be said to represent what is the case, although again, the idea of such representation is applied in a distinctive register in the rational case. We might choose to reserve the word “belief” for the distinctively rational case or we might not: either choice would amount to a bit of terminological legislation, and either would be acceptable so long as we did not lose sight of the specific combination of likeness and difference that obtains here. My present aim is not to defend these views about rational belief but merely to note their possibility. This is at least an intelligible sort of position to take, whether or not it is defensible in the case at hand. The Univocality Assumption in effect rules out such a position with regard to concepts such as belief, knowledge, inference, and warrant. But the claim that these concepts have a different and more stringent meaning in application to rational beings than they have in application to nonrational animals at least deserves a hearing. If such a position were vindicated, then, despite what Burge says, we would have a basis for denying that epistemology can evaluate the perceptual beliefs of children and nonhuman animals for warrant in the sense of warrant that is proper to rational beings. There is a tendency, in studying the differences between human beings and nonhuman animals, to look for some crucial experiment that will either vindicate or disprove the idea that there is some basic cognitive difference between us and them. Thus people study whether other

420

Matthew Boyle

primates can use tools, can recognize themselves in a mirror, can learn to use a symbolic system that looks like a human language, etc. I think these studies are fascinating, but insofar as they are supposed to test the proposition that human beings are rational in a sense that other primates are not, I think they rest on a distorted conception of what this fact would have to amount to. If the Classical View is right, we should not expect the rational-nonrational contrast to manifest itself primarily in the fact that rational creatures can do some specific thing which nonrational creatures cannot. Rather, we should expect that the cognition and action of rational creatures is pervasively, essentially different from the cognition and action of nonrational creatures. That there should be analogies between human tool use and things done by other primates, between human language and the communicative activities of other primates, etc. – this is only to be expected, for it is granted on all sides that their powers fall under a common genus. The crucial question, though, is whether we are speaking in the same register when we say that we and they “use tools,” or “communicate,” or whatever. And the way to answer this question is not to fixate on some particular pattern of behavior taken in isolation – comparing human mirror-behavior with chimpanzee mirror-behavior, or the human readiness to use a hammer to drive a nail with the chimpanzee readiness to use a stick to get ants out of a hole in the ground – but rather to consider the general shape of the life-form of the kind of creature in question, the system of generic propositions that characterizes their way of living and the forms of explanation that relate individuals to those generic truths. Do these presuppose the capacity for reflective thought or do they not? To make such a determination will involve a holistic consideration of the form of life in question, but it is not a determination made without an empirical basis, nor is it even a particularly difficult one to make. If the difference between a rational and a nonrational creature seems unclear or insignificant, this may be because we are looking for it in the wrong place.33

33 For a summary of empirical work on the distinguishing features of human cognition that is consistent with the standpoint developed here, cf. Tomasello (1999). Tomasello is expert at bringing out how – as I would want to put it – our capacity for discursive thought transforms the sense in which we are capable of various forms of intelligent activity.

Essentially Rational Animals

421

3.3 A second common objection to views that posit an essential difference between rational and nonrational cognition is that they tie propositional attitude language to a framework that is too idealized to be plausible. Davidson says that we understand the significance of propositional attitude ascriptions by reference to a constitutive ideal of rationality, and that, in determining what attitudes people hold, we must apply a “principle of charity” which requires us to find in their thought and action as much rationality as possible. But what justifies us in supposing that actual people will live up to this ideal? Aren’t tendencies to inconsistency, to weakness of will, to rash judgment, and so on, as real a part of the human constitution as any tendency to get things right? If this is not already obvious to untutored observation, the objectors note, it is amply confirmed by rigorous studies of biases in human choice and judgment. As Stephen Stich puts it in his essay “Could Man Be an Irrational Animal?”: Aristotle thought man was a rational animal. From his time to ours, however, there has been a steady stream of writers who have dissented from this sanguine assessment… During the last decade or so, [the] impressionistic chroniclers of man’s cognitive foibles have been joined by a growing group of experimental psychologists who are subjecting human reasoning to careful empirical scrutiny. Much of what they have found would appall Aristotle. Human subjects, it would appear, regularly and systematically invoke inferential and judgmental strategies ranging from the merely invalid to the genuinely bizarre.34

The existence of such biases is intelligible enough: given that we have finite time to think about the choices we make, and given that our ancestors faced certain kinds of situations where a rapid judgment or choice was called for, it might very well have been adaptive for us to possess certain “nonideal” tendencies in our judging and acting. But then presumably the way to interpret the attitudes people hold is not necessarily to assume that their system of attitudes is as rational as possible.35 Again, I think this objection rests on an assumption that our reflections have put us in a position to question. The assumption is on display in Stich’s remark, which evidently presupposes that the idea that man is a rational animal must be taken as a claim about how most men think most of the time. Otherwise, how could it be a threat to this idea 34 Stich (1985), 115. 35 For a lucid statement of this sort of objection, cf. Cherniak (1981).

422

Matthew Boyle

that human subjects regularly and even systematically make invalid inferences or judge questions on unsound bases? But as we have seen, the claim that man is a rational animal is not meant as some sort of statistical generalization. It is a claim about our essential nature, about what it is to be a human being, and to say that it is in our nature to be rational is not necessarily to say that most members of our species draw rational inferences most of the time. This is connected with a point that came up in our discussion of the grizzly bear: the powers and activities that belong to the essence of a certain kind of creature are not necessarily powers and activities that most such creatures exhibit. They are powers and activities that belong to an account about how creatures of that kind exist – an account whose exemplification in any given case is subject to all the sorts of obstacles and interferences that the world can produce, but which nevertheless supplies the explanatory principle in relation to which what does occur is intelligible. For, as Aristotle observes, the account of a power figures not only in the explanation of cases in which the power is successfully actualized, but also “by negation and subtraction” in the explanation of cases in which it is not successfully actualized. That is to say: we understand the shape that things have taken in such cases precisely by understanding how the normal course of things has gone awry, either because it has been interfered with (negation) or because some precondition was missing (subtraction).36 We can call the assumption that statements about the nature of a certain kind of living thing must be read as involving an implicit quantifi36 Cf. Aristotle (1984), Metaphysics IX. 2, 1046b13. Compare also the way Herder responds to an objection to the claim that the power of speech is part of our essence as human beings: “‘But those savage human children among the bears, did they have language? And were they not human beings?’ Certainly! Only, first of all, human beings in an unnatural condition! Human beings in degeneration! Put the stone on this plant; will it not grow crooked? And is it not nevertheless in its nature an upwards-growing plant? And did this power of straight growth not express itself even in the case where the plant entwined itself crookedly around the stone?” (Herder (2002), 93). The connection between rationality and language is not my topic here; what interests me is the kind of response Herder is offering. The response is: Pointing to cases in which Ss are not F does not necessarily falsify the claim that it is essential to being an S to be F, for it may be that cases of an S’s not being F are intelligible precisely as cases of a power to be F operating under interference. For helpful discussion of the idea of powers and their fallibility, and of the ways in which they can figure even in the explanation of their failed acts, cf. also Kern (2006), chs. 6 – 8 and Rödl (2007), ch. 5.

Essentially Rational Animals

423

cation over (all or most) individuals of that kind the Quantificationalist Assumption. We have seen that this assumption embodies a basic misunderstanding of the logic of essentialist claims. The claim that it is an essential property of horses to have four legs (or more simply: that the horse has four legs) is not falsified by the existence of three-legged horses. It would not necessarily be falsified even if actual horses for the most part had three legs. Likewise, if the proposition that human beings are rational animals, and the more specific claims of essential connection that articulate the content of this proposition, are claims about what it is to be a human being, then the way to evaluate these claims is not to ask whether human beings for the most part give cogent accounts of their reasons for belief, draw inferences in accordance with the laws of logic and probability, or choose in accordance the principles of decision theory. The way to evaluate such claims is rather to ask what kinds of powers are exercised in human thinking, what should count as the normal operation of such powers, and what should count as malfunctions calling for special explanation. The claim that these powers belong to human nature is entirely consistent with the observation that we very often fail in their exercise. Indeed, it is consistent with the observation that there are human beings who never attain to these powers at all. For even human beings who lack these powers belong to a species whose form of life involves the development of these powers: they are individuals whose potentiality to develop such powers has not been realized, and their minds are thus defective in an important way. A nonrational animal, by contrast, does not count as defective for want of the capacity to deliberate, to reflect on its own beliefs, to produce reasons for what it believes, etc. The idea that ascriptions of propositional attitudes to human beings make reference to an ideal of rationality must be understood against the background I have just been sketching. To say that the application of concepts of belief and desire to a human subject presupposes a constitutive ideal of rationality is not to claim that it is a necessary condition for the application of these concepts that we find that subject for the most part rational in his beliefs and choices. The point is rather that the fundamental employment of these concepts is one in which they figure in representations of a subject as believing and acting for adequate reasons, grasped as such – as exercising powers to get things right in the distinctive way in which rational creatures can get things right. I do not claim that this is consistent with everything Davidson says about idealization, charity, etc.; but I think it captures what is insightful in his view. Of course a

424

Matthew Boyle

person can believe irrationally and act irrationally, but, if Davidson is right, what underwrites our recognition of such cases as involving irrational beliefs and irrationally-efficacious desires is our grasp of the role of these concepts in a framework of rationalizing explanation. We could put it this way: cases of believing and acting rationally are the ones we must consider in understanding the “what-it-is-to-be” of belief and desire. A common contemporary opinion – often expressed in newspapers and magazines, but also, I think, held by some philosophers – is that someone who asserts a difference in kind between rational and nonrational animals must be intent on exalting human beings above all other living creatures. No doubt there have been defenders of the rational-nonrational distinction who have had such aims, but the idea that this is the basic motive for philosophical interest in the concept of a rational animal seems to me mistaken. The interest of this concept does not depend on a concern with drawing comparisons. The claim that we human beings possess a distinctively rational kind of animal mind embodies a thesis about the framework in which to understand our own minds, a thesis we can accept without adopting any view about the minds of other animal species. Roughly stated, the thesis is this: an account of our minds must not treat rationality as an isolable capacity belonging to a kind of animal mind whose other capacities could be realized, essentially unchanged, in a mind lacking this special further power. The claim that rational animals have a distinctive kind of animal mind thus implies that rational capacities for perception and desire cannot be explained as: the kinds of capacities for perception and desire to be found in nonrational animals, supplemented with a further, independent power to regulate these capacities in the light of reflective reasoning. Rather, an account of our sort of perceiving and desiring must itself refer to the role of these capacities in supporting a specifically rational form of life.37 If this is right, we are not merely animals who are in fact rational; we are essentially rational animals. 3.4 To query the Univocality Assumption and the Quantificationalist Assumption is not yet to demonstrate the need for a basic distinction between rational and nonrational minds, nor is it to give a substantive ac37 For further development of this idea, see my “Tack-on Theories of Rationality: A Critique” (forthcoming).

Essentially Rational Animals

425

count of what rationality amounts to.38 But it is, I hope, to remove some important obstacles to such an account. If I have shown that these obstacles depend on questionable assumptions about the logic of essentialist claims, and that the Classical View of the rational-nonrational distinction offers an alternative to these assumptions, I have achieved my purpose here.39 38 An adequate treatment of these issues would need to begin, I believe, by contesting the idea that it is simply an empirical observation about human beings that they are rational animals. I think the claim that human beings are rational animals is grounded, not fundamentally in empirical observation, but in selfconscious reflection. One way to see this is to consider that each of us can exhibit for himself the grounds for this definition simply by reflecting on the question “What sort of creature am I?” For the power exhibited in even considering this question – the capacity to reflect on a question and form a view on the basis of grounds recognized as such – is the power of rationality itself. So the idea that we are rational beings is an idea that each of us can verify for himself simply by considering the question. And if to be an animal is to be a living creature capable of perception and desire-governed action, then the fact that we are animals is also not something we need discover about ourselves by observation. The fact that we are subject to perceptual appearances of the world around us, feel desires for things, and make choices about what objects to pursue, are also facts which belong immediately to our self-conscious, nonobservational knowledge of ourselves. The characterizations of human beings as rational, and as animals, thus seem to articulate facts we are in a position to know, not primarily by looking at ourselves, but, so to speak, by looking into ourselves – facts we are in a position to know about our own minds in virtue of self-consciously having minds of the relevant sort. We could therefore say – switching philosophical idioms – that the concept rational animal is a “concept of reflection” in Kant’s sense: it is a concept whose source lies in our reflective consideration of our own cognitive activity, rather than in our empirical observation of particular objects with which our cognition is concerned, cf. Kant (1998), A260/B316. The fact that we have this sort of access to our rational nature is, I believe, what underwrites Kant’s confidence that this nature can be investigated systematically and completely by philosophy. Thus he remarks at the beginning of the first Critique: “I have to do with nothing save reason itself and its pure thinking; and to obtain knowledge of these, there is no need to go far afield, since I come upon them in my own self” (Kant (1998), Axiv). This is obviously only sketch of a program for grounding the concept rational animal. I hope to pursue these matters further in future work. 39 For responses to earlier drafts of this paper, I am indebted to audiences at Auburn University, the University of Chicago, and the Universität Leipzig. I am especially grateful to Dorit Bar-On, Jim Conant, Matthias Haase, Sean Kelsey, Eric Marcus, Sebastian Rödl, and Pirimin Stekeler-Weithofer for comments and advice.

426

Matthew Boyle

References Anscombe (1957): Gertrude Elizabeth Margaret Anscombe, Intention, Oxford. Aquinas (1948): Thomas Aquinas, Summa Theologica, New York. Aquinas (1949): Thomas Aquinas, On Being and Essence, Toronto. Aristotle (1984): Aristotle, Complete Works of Aristotle: The Revised Oxford Translation, Jonathan Barnes (ed.), Princeton. Bennett (1964): Jonathan Bennett, Rationality, London. Boyle (forthcoming): Matthew Boyle, “Tack-on Theories of Rationality: A Critique”, to appear in: The European Journal of Philosophy. Burge (2003): Tyler Burge, “Perceptual Entitlement”, in: Philosophy and Phenomenological Research 67 (3), 507 – 548. Burge (2010): Tyler Burge, Origins of Objectivity, Oxford. Cherniak (1981): Christopher Cherniak, “Minimal Rationality”, in: Mind 90 (358), 161 – 183. Cohen (2009): Marc S. Cohen, “Aristotle’s Metaphysics”, in: E. N. Zalta (ed.), The Stanford Encyclopedia of Philosophy, http://plato.stanford.edu/archives/spr2009/entries/aristotle-metaphysics (2011/03/31). Davidson (1980): Donald Davidson, “Mental Events”, in: D. Davidson (ed.), Essays on Actions and Events, Oxford, 207 – 224. Fine (1994): Kit Fine, “Essence and Modality”, in: Philosophical Perspectives 8, 1 – 16. Fine (1995a): Kit Fine, “Senses of Essence”, in: W. Sinnott-Armstrong/D. Raffmann/N. Asher (eds.), Modality, Morality, and Belief: Essays in Honor of Ruth Barcan Marcus, Cambridge/MA. Fine (1995b): Kit Fine, “The Logic of Essence”, in: Journal of Philosophical Logic 24 (3), 241 – 273. Fodor (2003): Jerry Fodor, “Review of José Luis Bermúdez, Thinking without Words”, in: London Review of Books 25 (19), 16 – 17. Herder (2002): Johann Gottfried Herder, “Treatise on the Origin of Language”, in: J. G. Herder, Philosophical Writings, Cambridge/MA. Kahneman et. al. (1982): Daniel Kahneman/Paul Slovic/Amos Tversky (eds.), Judgment under Uncertainty: Heuristics and Biases, Cambridge/MA. Kant (1998): Immanuel Kant, Critique of Pure Reason, Cambridge/MA. Kern (2006): Andrea Kern, Quellen des Wissens, Frankfurt/Main. Moravcsik (1994): Julius M. Moravcsik, “Essences, Powers, and Generic Propositions”, in: T. Scaltas/D. Charles/M. L. Gill (eds.), Unity, Identity, and Explanation in Aristotle’s Metaphysics, Oxford, 229 – 244. Nisbett/Ross (1980): Richard E. Nisbett/Lee Ross (eds.), Human Inference: Strategies and Shortcomings of Social Judgment, Englewood Cliffs/NJ. Porphyry (2003): Porphyry, Introduction to Aristotle’s Categories, Oxford. Rödl (2007): Sebastian Rödl, Self-Consciousness, Cambridge/MA. Stich (1985): Stephen Stich, “Could Man Be an Irrational Animal?”, in: Synthese 64 (1), 115 – 135. Thompson (1998): Michael Thompson, “The Representation of Life”, in: R. Hursthouse/G. Lawrence/W. Quinn (eds.), Virtues and Reasons. Philippa

Essentially Rational Animals

427

Foot and Moral Theory, Oxford, 247 – 296 (reprinted in Thompson (2008), 25 – 82). Thompson (2004): Michael Thompson, “Apprehending Human Form”, in: A. O’Hear (ed.), Modern Moral Philosophy, Cambridge/MA, 47 – 74. Thompson (2008): Michael Thompson, Life and Action, Cambridge/MA. Tomasello/Call (1997): Michael Tomasello/Josep Call (eds.), Primate Cognition, Oxford. Tomasello (1999): Michael Tomasello, The Cultural Origins of Human Cognition, Cambridge/MA. Wiggins (2001): David Wiggins, Porphyry, Cambridge/MA.

Notes on Contributors Matthew Boyle is Associate Professor of Philosophy at Harvard University. His main research interests are in the philosophy of mind, on the one hand, and Kant and German Idealism, on the other. He is the author of The Significance of Self-Consciousness (forthcoming). Quassim Cassam is Professor of Philosophy at the University of Warwick. He was previously Professor of Philosophy at University College London and Knightbridge Professor of Philosophy at University of Cambridge. He works on epistemology, philosophy of mind and Kant. He is the author of two books, Self and World (1997) and The Possibility of Knowledge (2007). James Conant is Chester D. Tripp Professor of Humanities, Professor of Philosophy, and Professor in the College at the University of Chicago. He works broadly in philosophy and has published articles in philosophy of language, philosophy of mind, aesthetics, German Idealism, and history of analytic philosophy, among other areas, and on a wide range of philosophers, including Kant, Emerson, Nietzsche, Kierkegaard, Josiah Royce, William James, Frege, Carnap, Wittgenstein, Putnam, Cavell, Rorty, and McDowell, among others. James Doyle is Visiting Scholar at the Institute for Advanced Study in Princeton. He is interested in the history of philosophy, especially ancient philosophy, and the philosophy of mind and action. Amongst his publications are “Socrates and Gorgias” (2010) and “Moral Rationalism and Moral Commitment” (2000). David Finkelstein is Associate Professor of Philosophy at the University of Chicago. His research interests include: first-person authority, consciousness, expression, animal minds, perception, and Wittgenstein. Amongst his publications are: Expression and the Inner (2003) and “Replies to My Commentators” (2011). Martin Gustafsson is Professor of Philosophy at Åbo Akademi University, Finland. His research interests are in the philosophy of language,

430

Notes on Contributors

the philosophy of action, philosophical methodology, and the history of analytic philosophy. He is the author of “Perfect Pitch and Austinian Examples: Cavell, McDowell, Wittgenstein and the Philosophical Significance of Ordinary Language” (2005), and the editor of The Philosophy of J. L. Austin (2011). Matthias Haase is Assistant for Philosophy at the University of Basel, Switzerland. He completed his PhD at the University of Potsdam, Germany. His research interests include ethics and the metaphysics of life, Wittgenstein, the nature of judgment and action, with special focus on self-consciousness, the nature of concepts and capacities. He is the author of “The Laws of Thought and the Power of Thinking” (2011). Adrian Haddock is Senior Lecturer in Philosophy at the University of Stirling. He works mainly on action, perception, and objectivity. His recent publications include “‘The knowledge that a man has of his intentional actions’” (2011), and “Meaning, Justification, and ‘Primitive Normativity’” (forthcoming). Thomas Lockhart is a PhD student at the University of Chicago. His dissertation explores Frege’s conception of the relationship between logic and ontology. He is the author of “Prolegomena to a Proper Treatment of Mathematics in The Critique of Pure Reason” (2006). John McDowell is Professor of Philosophy at the University of Pittsburgh. His main research interests are philosophy of mind, metaphysics, epistemology, Greek Philosophy, philosophy of language, and ethics. Amongst his publications are Mind and World (1994), Mind, Value, and Reality (1998), and Meaning, Knowledge, and Reality (1998). Benjamin McMyler is Assistant Professor of Philosophy at Texas A&M University. He works in epistemology, philosophy of mind, and social and political philosophy, and he is the author of Testimony, Trust, and Authority (2011). Will Small is a graduate student in the Department of Philosophy at the University of Chicago. He is completing a doctoral dissertation on action and practical knowledge. His areas of interest include philosophy of action, ethics, self-knowledge, and the relation between theoretical and practical reason.

Index of Persons Agrippa 2 Albritton, Rogers 69 Anscombe, Gertrude E.M. 121, 130-224, 244, 415 Aristotle 145, 209, 214, 254, 395413, 421-425 Augustine 129 Austin, John L. 26, 38, 70, 183, 191 Avramides, Anita 101 Ayer, Alfred 36, 38 Bäckström, Stina 101, 346 Baier, Annette 153, 167, 172 Baker, Gordon 69 Bar-On, Dorit 88, 112, 406, 425 Bennett, Jonathan 397 Berkeley, George 44 Boghossian, Paul 77, 83 Bollas, Christopher 81 Boyle, Matthew 101, 102, 104107, 11, 176, 190, 229 Brandom, Robert 61, 234f., 246, 271 Bratman, Michael 135, 140-144, 165, 173ff., 190 Brewer, Bill 342 Bridges, Jason 97, 101, 224 Broad, Charlie D. 36 Burge, Tyler 83f., 258, 289-307, 310, 342, 354, 418f. Burnyeat, Myles F. 290 Byrne, Alex 79, 85, 310f., 342, 352 Calvert, Jane 349 Carruthers, Peter 85 Cavell, Stanley 3, 5, 23f., 26, 28, 41, 60-72, 229-243, 247f., 250255 Cherniak, Christopher 421

Chisholm, Roderick 36 Cohen, Marc S. 401 Comrie, Bernard 176ff., 186 Conant, James 97, 101, 119, 171, 174, 193, 224, 229, 239, 346, 393, 425 Crane, Tim 92 Darwall, Stephen 260f., 277ff., 284 Davidson, Donald 52, 61, 77, 130, 135, 140-143, 150-154, 161, 165, 175ff., 197, 200, 202f., 416ff., 421, 423f. Dennett, Daniel 34 Descartes, René 1-15, 18-38, 4048, 52-66, 69ff., 80, 290f., 306, 331-334, 336ff., 340f., 346 Dewey, John 53-61 Donnellan, Keith 198 Duchamp, Marcel 14 Egan, Frances 302 Enç, Berent 216 Engstrom, Stephen 145, 168 Evans, Gareth 102f., 111-117, 233, 289, 305f., 382f. Falvey, Kevin 147, 176, 181ff., 190-193, 199f. Feinberg, Joel 152 Fine, Kit 407 Finkelstein, David 97, 119-132, 160, 346 Fish, Wiliam 342 Fodor, Jerry 232, 394 Ford, Anton 158, 195, 224 Frankfurt, Harry 80f.

432

Index of Persons

Frege, Gottlob 5, 52, 58, 230, 304, 306, 412 Fricker, Elisabeth 263ff. Galton, Antony 147, 178, 181, 186, 189 Gibbons, John 142, 203 Ginsborg, Hannah 381f. Goldman, Alvin I. 150 Green, Thomas Hill 52, 122 Grice, H. Paul 225 Haddock, Adrian 97, 101, 311 Hamawaki, Arata 101 Hampshire, Stuart 82, 86 92, 186 Heal, Jane 101 Hegel, Georg W.F. 2, 234, 298 Heidegger, Martin 36ff. Herder, Johann G. 395, 415f., 422 Hornsby, Jennifer 207-210, 217, 349 Hume, David 2, 47f., 52, 169, 194 James, William 54, 58, 60 Jones, O. R. 133 Kafka, Franz 34 Kant, Immanuel 2ff., 6f., 20, 24f., 27, 30, 38, 42-49, 51f., 60, 62, 64, 69, 71f., 145, 187, 219, 221, 425 Kelly, Thomas 93 Kern, Andrea 229, 241, 358f., 369, 422 Kierkegaard, Søren 2 Kimhi, Irad 101 Koch, Felix 224 Kripke, Saul 23f., 61ff., 65, 71, 243, 248 Lackey, Jennifer 262f, 265, 273 Ladyman, James 119 Lavin, Douglas 176, 190 Lear, Jonathan 97, 123 Lewis, Clarence I. 36-61 Lichtenberg, Georg Christoph 80 Lindgaard, Jakob 392 Lockhart, Thomas 9, 101 Logue, Heather 310f., 342

Malcolm, Norman 69 Martin, Michael 92f. Martin, Mike G. F. 309ff., 316-336, 340-346, 352 McDowell, John 9, 48-73, 218-229, 242, 309-393, 418 McGinn, Colin 120ff., 128, 133f., 143 Michaelson, Eliot 224 Millar, Alan 349, 352 Moore, George E. 36, 381, 383, 389 Moran, Richard 77f., 81f., 87f., 93-97, 102-113, 117, 144, 148, 160, 176, 264f., 269, 273 Moravcsik, Julius M. 405 Müller, Anselm Winfried 153 Nichols, Shaun

95

O’Brien, Lucy

208, 210

Paul, Sarah K. 172 Peacocke, Christopher 90f. Pippin, Robert 101 Plantinga, Alvin 354 Porphyry 401 Price, Henry H. 36-39, 42, 55, 65 Pryor, James 79 Rödl, Sebastian 136, 163, 170f., 174, 176, 179, 183, 185, 189, 211ff., 224, 229, 241, 249, 254, 307, 355, 357-360, 368, 422, 425 Rorty, Richard 38 Russell, Bertrand 36 Ryle, Gilbert 79, 84f., 160, 215f. Sartre, Jean-Paul 36 Searle, John R. 160, 209, 212, 276 Sellars, Wilfrid VII, 36ff., 60f., 80, 84, 239, 295, 353f., 356, 366ff., 381, 418 Setiya, Kieran 139, 140-144, 162, 170ff., 191, 203ff., 214-217 Shaddock, Justin 224 Shah, Nishi 91f. Shakespeare, William 28 Shoemaker, Sidney 90, 103, 117

Index of Persons

Siegel, Susanna 42 Small, Will 342 Smith, Nicholas 393 Snowdon, Paul 191, 216 Stanley, Jason 216 Steward, Helen 217 Stich, Stephen 95, 421 Stone, Martin J. 148, 162, 176 Strawson, Peter F. 37 Stroud, Barry 101, 380ff. Thomas Aquinas 134, 203, 399, 401, 426 Thompson, Michael 136, 147, 176f., 185, 199, 210-214, 229, 241, 254, 403-406 Tomasello, Michael 396, 420 Travis, Charles 310, 342

433

Velleman, David 91f., 134, 139, 141, 143ff., 149, 162 Vogler, Candace 166, 174, 224 Warhol, Andy 14 Wiggins, David 402 Willaschek, Marcus 349 Williams, Michael 2 Williamson, Timothy 92f., 96, 216, 303, 305, 361, 368 Wittgenstein, Ludwig 20, 29, 52, 55ff., 61-72, 116, 119, 121f., 224, 237f., 241, 243, 248, 252, 254f., 283, 346, 375, 377 Wright, Crispin 60, 69, 89, 97

Index of Topics Accordion-effect 151f., 155 Action – Basic action 136f., 142, 165, 173, 198, 206-214, 216-224 – Calculative structure of action 137, 142, 154, 196f., 207, 213 – Communal action 249ff. – Intentional action 12f., 16, 131, 135-139, 141f., 144ff., 149f., 153ff., 158-166, 169, 171, 173, 175, 177ff., 181f., 187f., 190ff., 195ff., 199, 201ff., 205-208, 210214, 216, 222-224, 415 – Temporal structure of action 135f., 142, 146, 149, 165, 175ff., 187, 189f., 196f., 203, 210f. – Temporal unity of action 187ff., 192, 195, 198, 203, 205, 209, 213f., 216 Address 257-284 Aesthetics 13, 18 Anti-luminosity argument 363367, 372 Assertion 90, 114f., 126, 181, 229, 231, 233, 235ff., 245, 263f. Awareness 85f., 94, 104-107, 110, 114, 168, 249f., 258, 301, 307, 317, 331, 395 Carbon-copier-example 130, 140ff., 175, 197-203 Certainty 26, 40, 196 Cognition 48f., 134-137, 139, 141-145, 149f., 153ff., 162-166, 174, 177ff., 193, 197, 201, 203ff., 208, 216, 223f., 298f., 396, 400, 420f., 425 Common factor 9ff., 13, 313, 315, 329f., 332ff., 337f.

Common Kind Assumption 319324, 329 Community 231, 235ff., 25ff., 258, 339 Conceptual capacities 219-223, 241, 254, 312, 341ff., 345, 371f., 377, 391 Conceptual content 219, 239f., 243, 253, 311f., 345, 389, 392 Conjunctivism 313, 315, 330f. Consciousness 125, 159, 229, 232, 234, 325, 343, 345, 358 Direction of fit 134, 143f., 149 Disjunctivism 111, 289-307, 309346 Disposition 92, 96f., 214-218, 223, 241f., 247-251, 395, 412 Doubt 19f., 25ff., 29ff., 34f., 38ff., 44f., 47, 52, 64, 85, 130, 140, 144, 238, 244, 249, 375, 392, 395, 400 Epistemic – Epistemic buck-passing 260, 266, 270, 272 – Epistemic immediacy 79f. , 88ff., 94ff. – Epistemic responsibility 260, 266ff., 270 – Epistemic standing 31, 333ff., 337, 344 Epistemology 133, 204, 210, 223, 229, 262, 271f., 299, 309, 313, 332, 358, 385f., 419f. Facts 77, 133, 148f., 153, 187, 331, 385, 425 Fallibility 200, 292-299, 305, 307, 337, 422

436

Index of Topics

First person 229-255 – First-person authority 102-117 – First-/third-person asymmetry 119f. Idealism 3, 7, 46-50, 52 Illusion 47, 289f., 298, 312, 335, 349ff., 364, 382f. Imagination 46, 53, 144, 168, 305, 307, 412 Individuation 150, 402, 418 Inference / inferential 5, 19, 38, 45, 77ff., 86, 89f., 128, 133, 137, 178, 205, 215f., 257ff., 261, 271f., 294, 296, 339, 379, 385393, 408, 418f., 421ff. Intentionality 66, 250 Interpretation 11, 19, 29, 35, 67, 152, 230, 237, 248, 351, 375f., 393, 399, 407 Intuition 134, 201, 219, 221f., 411 Knowledge – Form of knowledge 133f., 189, 206, 210, 215, 223 – Knowledge How 164, 170, 173, 191f., 204-210, 213-216, 223f. – Immediate knowledge 77ff., 86f., 89 – Knowledge in intention 137, 154, 164ff., 170, 175, 178f., 187ff., 197, 203ff., 207, 216, 223f. – Non-observational knowledge 78, 105, 138, 139, 198, 245 – Procedural knowledge 136, 173, 209, 212, 216f., 222, 224 – Self-knowledge 77-97, 101f., 104, 110f., 113, 117, 139, 291, 294 Language 11, 17, 57, 61, 66, 84f., 116, 130, 176, 217, 229-233, 236, 238-242, 246, 250-254, 259, 300, 302, 315, 413, 420ff. Luminosity 355f., 360-367, 372

Meaning 11f., 16, 21, 33, 35, 5558, 62f., 176f., 186, 229, 231, 238ff., 254, 259, 400, 419 Mental state 15, 79, 92, 96f., 101, 119-132, 139, 144, 146, 160, 289, 298, 302 Metaphysics 54, 205, 321, 323, 328, 330, 346, 399 Müller-Lyer-Illusion 380-383 Myth of the Given 218, 220, 392 Naturalism 320 Normative VI, 162, 172, 235, 240243, 246f., 253, 265, 296 Passing thoughts 78-92 Perception / perceptual – Apparent perception 8-10, 313316, 321, 324f., 330, 333f., 338 – Veridical / non-veridical perceptions 8f., 13, 15, 289, 298, 309-346 – Perceptual capacities 292-301, 307, 312, 318, 328, 331, 335f., 342f., 371, 377, 391 – Perceptual content 272, 296, 300, 309, 312, 342, 376, 386f., 392 – Perceptual experience 9, 48, 218, 290, 301, 309-314, 321f., 327, 333-335, 338, 343 – Perceptual knowledge 86, 139, 201, 295, 337, 354, 356, 358, 364, 368-370 – Perceptual systems 300f. Perspective 3, 149, 229, 260, 273, 418 Phenomenon / phenomenal 3, 21, 26, 45, 68, 106, 112, 219, 322330, 341 Phenomenology 85, 87, 159, 352f. Philosophy of Action 133, 145, 208, 224 Philosophy of Language 10, 16, 18, 19, 24, 58, 229, 233, 240 Philosophy of Mind 18, 59, 246, 258, 400

Index of Topics

Powers 145, 183, 191, 203, 378, 388, 395, 399, 414f., 422, 424f. Practice 13, 23, 25, 34f., 64, 133, 170, 201, 208f., 235, 237, 239, 243, 246f., 251-255, 282, 361 Pragmatism 246, 268f. Premise 90, 144, 180, 205, 215, 282, 334ff., 344, 379, 385-393 Proposition 78f., 81, 83, 86, 89, 91, 106, 120, 215f., 221ff., 230, 296, 312, 392, 404f., 408, 411, 416, 420f., 423 Psychoanalysis 123, 125 Rational capacities 222ff., 295, 424 Realism 3, 52, 54, 57, 60 – Naive Realism 317-325, 328332, 343, 351, 353 Representation / representational 4, 37, 46f., 57f., 95, 134, 143, 162f., 174, 178, 184, 196, 222, 231, 258, 272, 292, 300, 311, 341, 345, 416-419, 423 Rule-following 20, 62, 66, 240, 242f., 252ff. Self-ascription 120-131

102f., 109-116,

437

Self-conscious 159, 163, 174, 229, 232, 234, 241, 248, 252, 294f., 304, 307, 358, 425 Self-knowledge 77-97, 101f., 104, 110f., 113, 117, 139, 291, 294 Sensibility 15, 31, 40f., 51 Skepticism 175, 179, 193, 215, 231, 238f., 252, 254, 290, 306, 309, 331-340 Skill 137, 173, 192f., 202f., 206210, 214-224 Social 231, 235, 242, 252, 257ff., 284 Social practice 235, 237, 241, 246, 252f. Standing attitudes 78, 87, 97 Substance 15, 38, 45, 398, 402f., 405-408, 410 Teleology / teleological 163, 174, 188f., 208, 404f. Testimony 9, 14, 27, 110, 120, 257-285, 375, 413 Transcendental 30, 46, 51f., 108, 338f. Transparency 77-79, 83, 87-97, 101-118 Trust 8, 28, 257, 263, 375-393 Warrant 292-296, 301, 305, 307, 335, 337f., 354, 417ff.