167 69 825KB
English Pages 194 Year 2007
Bernd Prien, David P. Schweikard (Eds.) Robert Brandom
10. Münstersche Vorlesungen zur Philosophie 2006 10th Münster Lectures on Philosophy 2006 mit / with
Robert Brandom
Bernd Prien David P. Schweikard (Eds.)
Robert Brandom Analytic Pragmatist
ontos verlag Frankfurt I Paris I Ebikon I Lancaster I New Brunswick
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.d-nb.de.
North and South America by Transaction Books Rutgers University Piscataway, NJ 08854-8042 [email protected] United Kingdom, Ire Iceland, Turkey, Malta, Portugal by Gazelle Books Services Limited White Cross Mills Hightown LANCASTER, LA1 4XS [email protected]
Livraison pour la France et la Belgique: Librairie Philosophique J.Vrin 6, place de la Sorbonne ; F-75005 PARIS Tel. +33 (0)1 43 54 03 47 ; Fax +33 (0)1 43 54 48 18 www.vrin.fr
2008 ontos verlag P. O. Box 15 41, D-63133 Heusenstamm nr. Frankfurt Tel. ++(49) 6104 66 57 33 Fax ++(49) 6104 66 57 34 www.ontosverlag.com ISBN 978-3-938793-77-0 2008 No part of this book may be reproduced, stored in retrieval systems or transmitted in any form or by any means, electronic, mechanical, photocopying, microfilming, recording or otherwise without written permission from the Publisher, with the exception of any material supplied specifically for the purpose of being entered and executed on a computer system, for exclusive use of the purchaser of the work
Printed on acid-free paper ISO-Norm 970-6 FSC-certified (Forest Stewardship Council) Printed in Germany by buch bücher dd ag
CONTENTS
Preface
9
List of Abbreviations
11
1
LECTURE
Towards an Analytic Pragmatism
13
Robert Brandom
2
COLLOQUIUM
Brandom on Knowledge and Entitlement
47
Sebastian Schmoranzer, Ansgar Seide
Of μ-Mesons and Oranges
59
Scrutinizing Brandom’s Concept of Observability Jochen Apel, Simone Bahrenberg, Carolin Köhne, Bernd Prien, Christian Suhm
Successful Action and True Beliefs
69
Nikola Kompa, Rudolf Owen Müllan, Bernd Prien, David P. Schweikard
Are Fundamental Discursive Norms Objective?
79
Sebastian Laukötter, Bernd Prien, Till Schepelmann, Christian Thein
Realist and Idealist Interpretations of Brandom’s Account of Objectivity Michael Pohl, Raja Rosenhagen, Arne M. Weber
89
8
Contents
Comparing Brandom’s Critical Reading and Hegel’s Systematic Enterprise
101
Jana Elisa Falkenroth, Attila Karakuş, David P. Schweikard
Brandom and Frege
115
Ulrike Kleemeier, Christian Weidemann
Between Normative and Modal Vocabulary
125
A First Encounter with Brandom’s Locke Lectures Bernd Prien
How to Kripke Brandom’s Notion of Necessity
135
Benedikt Göcke, Martin Pleitz, Hanno von Wulfen
Appendix to How to Kripke Brandom’s Notion of Necessity
148
Benedikt Göcke, Martin Pleitz, Hanno von Wulfen
Replies Robert Brandom Reply to “Brandom on Knowledge and Entitlement”
163
Reply to “Of μ-Mesons and Oranges”
167
Reply to “Successful Action and True Beliefs”
171
Reply to “Are Fundamental Discursive Norms Objective?”
173
Reply to “Realist and Idealist Interpretations of Brandom’s
177
Account of Objectivity” Reply to “Comparing Brandom’s Critical Reading and
179
Hegel’s Systematic Enterprise” Reply to “Brandom and Frege”
184
Reply to “Between Normative and Modal Vocabulary”
188
Reply to “How to Kripke Brandom’s Notion of Necessity”
193
PREFACE Since the publication of Making It Explicit in 1994, Robert Brandom is among the most influential analytic philosophers and a leading figure in the NeoPragmatist movement. In that book, he presents a detailed, highly ingenious elaboration of inferentialism, i.e., the idea that the content of an assertion is determined by its place in a web of belief. An introduction to this theory appeared in 2000 under the title Articulating Reasons. It has always been a matter of great importance for Brandom to locate his views in a philosophical tradition and to show how they develop out of this tradition. He has documented this with a series of historical essays about Spinoza, Leibniz, Hegel, Frege, Wittgenstein, and Sellars, published as Tales of the Mighty Dead in 2002. His most recent major work, Between Saying and Doing, has been presented as the Locke Lectures 2006 in Oxford. In these Lectures, Brandom proposes to broaden the perspective of Classical Analytic Philosophy so as to not only consider the semantic relations between different vocabularies, but also to consider what he calls pragmatically mediated relations between vocabularies. The Philosophy Department of the University of Münster has invited Robert Brandom to the 10. Münstersche Vorlesungen zur Philosophie, which took place in December 2006. In line with what has by now become an almost venerable tradition, Brandom gave a public lecture (entitled “Towards an Analytic Pragmatism”) on the first night of the Vorlesungen and participated in a colloquium on the following two days. At this colloquium, groups of students and faculty members from Münster presented papers dealing critically with a wide range of topics from Brandom’s work. The lecture, the papers, and Brandom’s replies to them are published in this volume. We would like to express our gratitude to Robert Brandom for accepting the invitation to Münster and for the stimulating discussions about his work. We would also like to thank the students and colleagues from Münster who have invested much time and effort to prepare the papers and presentations.
10
Preface
Furthermore, our thanks go to the many helping hands in the background which ensured that the colloquium would run smoothly. Last, but not least, we are grateful to Rafael Hüntelmann and the ontos Verlag, both for funding the Münstersche Vorlesungen for the third year now and for publishing the papers presented at the Vorlesungen.
Münster, August 2007
Bernd Prien David P. Schweikard
LIST OF ABBREVIATIONS OF ROBERT BRANDOM’S PUBLICATIONS AR
Articulating Reasons: An Introduction to Inferentialism, Cambridge, Mass.: Harvard University Press, 2000.
BSD
Between Saying and Doing: Towards an Analytic Pragmatism, Locke Lectures held at Oxford University, Trinity Term 2006, to appear 2008 from Oxford University Press. Unpublished Manuscript at http://www.pitt.edu/~brandom/locke/index.html.
MIE
Making It Explicit: Reasoning, Representing, and Discursive Commitment, Cambridge, Mass.: Harvard University Press, 1994.
NIK
“Non-Inferential Knowledge, Perceptual Experience, and Secondary Qualities. Placing McDowell’s Empiricism”, in: Reading McDowell. On Mind and World, ed. by Nicholas H. Smith, London: Routledge, 2002, pp. 92-105.
PaP
“Pragmatics and Pragmatisms”, in: Hilary Putnam: Pragmatism and Realism, ed. by James Conant and Urszula M. Zeglen, London: Routledge, 2002, pp. 40-59.
SPCRH “Sketch of a Program for a Critical Reading of Hegel—Comparing Empirical and Logical Concepts”, in: International Yearbook of German Idealism 3, ed. by K. Ameriks & J. Stolzenberg, Berlin, New York: de Gruyter, 2005, pp. 131-161. TMD
Tales of the Mighty Dead: Historical Essays in the Metaphysics of Intentionality, Cambridge, Mass.: Harvard University Press, 2002.
US
“Unsuccessful Semantics”, Analysis 54, 1994, pp. 175-178.
TOWARDS AN ANALYTIC PRAGMATISM Robert Brandom
Within the Anglophone tradition, pragmatism has often appeared as a current of thought that stands apart from, and indeed runs in opposition to, the mainstream of analytic philosophy. This is true whether one uses ‘pragmatist’ in a narrow sense tailored to the triumvirate of Peirce, James, and Dewey (here one might think of Russell’s dismissive responses to the latter two), or in a more capacious sense that includes the early Heidegger, the later Wittgenstein, and, more recently, neo-pragmatists such as Rorty and Putnam. There are good reasons on both sides for adopting somewhat adversarial stances, but I think that when we examine them more closely it becomes possible to see the outlines of a common project, in the service of which the two camps might find themselves joining forces. In my 2006 John Locke lectures, entitled Between Saying and Doing: Towards an Analytic Pragmatism (to appear in 2008 from Oxford University Press), I explore in more detail one way of pursuing such a project. In this essay I want to offer a sketch of the basic understanding of the principal aims of the two movements that motivates that more extended discussion, and to indicate in general terms the sort of pragmatic semantic analysis (not, I will be insisting, an oxymoron) that might emerge from unifying their only apparently disparate concerns. The intended spirit is irenic, synthetic, and constructive.
1
THE CLASSICAL PROJECT OF ANALYSIS
I think of analytic philosophy as having at its center a concern with semantic relations between what I will call ‘vocabularies’. Its characteristic form of
14
Robert Brandom
question is whether and in what way one can make sense of the meanings expressed by one kind of locution in terms of the meanings expressed by another kind of locution. So, for instance, two early paradigmatic projects were to show that everything expressible in the vocabulary of number-theory, and again, everything expressible using definite descriptions, is expressible already in the vocabulary of first-order quantificational logic with identity. The nature of the key kind of semantic relation between vocabularies has been variously characterized during the history of analytic philosophy: as analysis, definition, paraphrase, translation, reduction of different sorts, truthmaking, and various kinds of supervenience—to name just a few contenders. In each case, however, it is characteristic of classical analytic philosophy that logical vocabulary is accorded a privileged role in specifying these semantic relations. It has always been taken at least to be licit to appeal to logical vocabulary in elaborating the relation between analysandum and analysans—target vocabulary and base vocabulary. I will refer to this aspect of the analytic project as its commitment to ‘semantic logicism’.1 If we ask which were the vocabulary-kinds whose semantic relations it was thought to be important to investigate during this period, at least two core programs of classical analytic philosophy show up: empiricism and naturalism. These venerable modern philosophical traditions in epistemology and ontology respectively were transformed in the twentieth century, first by being transposed into a semantic key, and second by the application of the newly available logical vocabulary to the self-consciously semantic programs they then became. As base vocabularies, different species of empiricism appealed to phenomenal vocabulary, expressing how things appear, or to secondary-quality vocabulary, or, less demandingly, to observational vocabulary. Typical target vocabularies include objective vocabulary formulating claims about how things actually are (as opposed to how they merely appear), primary-quality vocabulary, theoretical vocabulary, and modal, normative, and semantic vocabularies. The generic challenge is to show how what is expressed by the use 1
In this usage, the logicism about mathematics characteristic of Frege’s Grundgesetze and Russell and Whitehead’s Principia is a semantic logicism about the relations between mathematical and logical vocabularies.
Towards an Analytic Pragmatism
15
of such target vocabularies can be reconstructed from what is expressed by the base vocabulary, when it is elaborated by the use of logical vocabulary. As base vocabularies, different species of naturalism appealed to the vocabulary of fundamental physics, or to the vocabulary of the natural sciences (including the special sciences) more generally, or just to objective descriptive vocabulary, even when not regimented by incorporation into explicit scientific theories. Typical targets include normative, semantic, and intentional vocabularies.
2
THE PRAGMATIST CHALLENGE
What I want to call the “classical project of analysis”, then, aims to exhibit the meanings expressed by various target vocabularies as intelligible by means of the logical elaboration of the meanings expressed by base vocabularies thought to be privileged in some important respects—epistemological, ontological, or semantic—relative to those others. This enterprise is visible in its purest form in what I have called the “core programs” of empiricism and naturalism, in their various forms. In my view the most significant conceptual development in this tradition—the biggest thing that ever happened to it—is the pragmatist challenge to it that was mounted during the middle years of the twentieth century. Generically, this movement of thought amounts to a displacement from the center of philosophical attention of the notion of meaning in favor of that of use: in suitably broad senses of those terms, replacing concern with semantics by concern with pragmatics. The towering figure behind this conceptual sea-change is, of course, Wittgenstein. In characterizing it, however, it will be useful to approach his radical and comprehensive critique by means of some more local, semantically corrosive argumentative appeals to the practices of deploying various vocabularies rather than the meanings they express. Wilfrid Sellars (one of my particular heroes) criticizes the empiricist core program of the classical project of analysis on the basis of what one must do in order to use various vocabularies. He argues that none of the various candidates for empiricist base vocabularies are practically autonomous, that is, could
16
Robert Brandom
be deployed in a language-game one played though one played no other. For instance, no discursive practice can consist entirely of making non-inferential observation reports. For such reliably differentially elicited responses qualify as conceptually contentful or cognitively significant only insofar as they can serve as premises from which it is appropriate to draw conclusions, that is, as reasons for other judgments. Drawing such conclusions is applying concepts inferentially— that is, precisely not making non-inferential observational use of them.2 Quine offers an even broader pragmatist objection, not only to the empiricist program, but to essential aspects of the whole analytic semantic project. For he attacks the very notion of meaning it presupposes. Quine is what I have elsewhere called a “methodological” pragmatist.3 That is, he takes it that the whole point of a theory of meaning is to explain, codify, or illuminate features of the use of linguistic expressions. He, like Dummett, endorses the analogy: meaning is to use as theory is to observation. And he argues that postulating meanings associated with bits of vocabulary yields a bad theory of discursive practice. If there were such things as meanings that determine how it would be correct to use our expressions, then those meanings would at least have to determine the inferential roles of those expressions: what follows from applying them, what applying them rules out, what is good evidence for or against doing so. But what follows from what depends on what else is true—on laws of nature and obscure contingent facts—that is, on what claims can serve as auxiliary hypotheses or collateral premises in those inferences. If we look at what practical abilities are required to deploy various bits of vocabulary—at what one has to be able to do in order to count as saying something with them—we do not find any special set of these whose practical significance can be understood as pragmatically distinctive of semantically necessary or sufficient conditions.4 Quine thought one could save at least the naturalist program by retreating semantically to the level of reference and truth-conditions. James and 2
3 4
This argument occupies roughly the first half of his classic Empiricism and the Philosophy of Mind. His critique of the phenomenalist version of empiricism can be found in “Phenomenalism”. Cf. “Pragmatics and Pragmatisms”. Ibid.
Towards an Analytic Pragmatism
17
Dewey appeal to the same sort of methodological pragmatism in support of more sweeping sorts of semantic revisionism—pursuing programs that Rorty, for instance, argues should be understood as more rejectionist than properly revisionist. And under the banner “Don’t look to the meaning, look to the use,” Wittgenstein further radicalizes the pragmatist critique of semantics. Pointing out to begin with that one cannot assume that uses of singular terms have the job of picking out objects, nor that declarative sentences are in the business of stating facts, he goes on to deny, in effect, that such uses even form a privileged center, on the basis of which one can understand more peripheral ones. (“Language has no downtown.”) I take it that Wittgenstein also takes the home language-game of the concept of meaning to be explanation of how expressions are correctly used. And he is profoundly skeptical about the utility or applicability of the model of postulation, explanation, and theoretical systematization in the case of discursive practices—about the possibility of systematically deriving aspects of use from assigned meanings. Seen from this perspective, the idea of the classical project of analysis is to codify, using logical vocabulary, the meanings expressed by one vocabulary—from which we are to derive proprieties of its use—from the meanings expressed by some other vocabulary—from which we can derive proprieties of its use. One idea, I think is that this enterprise makes sense only if we think of the uses as species of a genus—of them all being the same general kind of use, say stating facts, or representing states of affairs. This may seem plausible if we focus on a very restricted set of uses— just as, in the case of tools, we might be impressed to notice that hammer and nails, screwdriver and screw, glue, gluepot and brush all have the function of attaching more-or-less flat things to one another. So we can think of declarative sentences as stating empirical, physical, normative, modal, and intentional facts, making claims about such states of affairs (even if we then find ourselves metaphysically puzzled about the nature of the fact-kinds to which we have thereby committed ourselves). But if we think of the uses as very different—if we think also about the carpenter’s pencil, ruler, level, and tool-belt—if we think of linguistic practice as a motley, of uses as not coming in a simple or systematic or even determinate variety, then the very idea that there is such a thing as meanings that permit the codification of proprieties of quite disparate kinds of use—even with liberal use of logical elaboration of the meanings—
18
Robert Brandom
becomes contentious and in need of justification both in general and in each particular case. More specifically, Wittgenstein uses the image of “family resemblances” to urge that the kinds into which linguistic practices and the vocabularies caught up in them are functionally sorted—what belong together in boxes labeled ‘game’, ‘name’, ‘assertion’, ‘observation’ and so on—do not typically admit of specification in terms of underlying principles specifiable in other vocabularies, whether by genus and differentia(e) or any other kind of explicit rule or definition. It is easy to understand this line of thought as entailing a straightforward denial of the possibility of semantic analysis in the classical sense. I think that the thought underlying these observations about the unsystematic, unsurveyable variety of kinds of uses of expressions and about the uncodifiable character of those kinds concerns the essentially dynamic character of linguistic practice. I think Wittgenstein thinks that an absolutely fundamental discursive phenomenon is the way in which the abilities required to deploy one vocabulary can be practically extended, elaborated, or developed so as to constitute the ability to deploy some further vocabulary, or to deploy the old vocabulary in quite different ways. Many of his thought-experiments concern this sort of process of pragmatic projection of one practice into another. We are asked to imagine a community that uses proper names only for people, but then extends the practice to include rivers. There is no guarantee that interlocutors can master the extended practice, building on what they can already do. But if they can, then they will have changed the only sessences proper-name usage could be taken to have had.5 In the old practice it always made sense to ask for the identity of the mother and father of the named item; in the new practice, that question is often senseless. Again, we are asked to imagine a community that talked about having gold or silver in one’s teeth, and extends that practice to talk about having pain in one’s teeth. If as a matter of contingent fact the practitioners can learn to use the expression ‘in’ in the new way, building on but adapting the old, they will have fundamentally changed the smeanings of ‘in’. In the old practice it made sense to ask where 5
Cf. Quine’s remark (in “Two Dogmas of Empiricism”): “Meaning is what essence becomes when it is detached from the thing and attached to the word.”
Towards an Analytic Pragmatism
19
the gold was before it was in one’s tooth; in the new practice asking where the pain was before it was in the tooth can lead only to a distinctively philosophical kind of puzzlement. At every stage, what practical extensions of a given practice are possible for the practitioners can turn on features of their embodiment, lives, environment, and history that are contingent and wholly particular to them. And which of those developments actually took place, and in what order can turn on any obscure fact. The reason vocabulary-kinds resist specification by rules, principles, definitions, or meanings expressed in other vocabularies is that they are the current time-slices of processes of development of practices that have this dynamic character—and that is why the collection of uses that is the cumulative and collective result of such developments-by-practical-projection is a motley. 6 If that is right, then any codification or theoretical systematization of the uses of those vocabulary-kinds by associating with them meanings that determine which uses are correct will, if at all successful, be successful only contingently, locally, and temporarily. Semantics on this view is an inherently Procrustean enterprise, which can proceed only by theoretically privileging some aspects of the use of a vocabulary that are not at all practically privileged, and spawning philosophical puzzlement about the intelligibility of the rest.7 On this conception, the classical project of analysis is a disease that rests on a fundamental, if perennial, misunderstanding—one that can be removed or ameliorated only by heeding the advice to replace concern with meaning by concern with use. The recommended philosophical attitude to discursive practice is accordingly descriptive particularism, theoretical quietism, and semantic pessimism.
6
7
A patient and detailed investigation of the mechanisms of this phenomenon in basic descriptive and scientific concepts, and an extended argument for its ubiquity can be found in Mark Wilson’s exciting and original Wandering Significance. I would be happy if those who dance with his texts find affinities here with Hegel’s insistence that the metaconceptual categories of Verstand must be replaced by those of Vernunft. It is characteristic of his philosophical ambition that draws the opposite of Wittgenstein’s conclusions from an appreciation of the dynamics of conceptual development and its sensitivity to arbitrary contingent features of the practitioners, devoting himself to elaborating what he insists is the logic of such processes and the conceptual contents they shape.
20
3
Robert Brandom
EXTENDING THE PROJECT OF ANALYSIS: PRAGMATICALLY MEDIATED SEMANTIC RELATIONS
On this account Wittgenstein is putting in place a picture of discursive meaningfulness or significance that is very different from that on which the classical project of analysis is predicated. In place of semantics, we are encouraged to do pragmatics—not in the sense of Kaplan and Stalnaker, which is really the semantics of token-reflexive expressions, nor again in the sense of Grice, which addresses conversational heuristics in terms that presuppose a prior, independent, classical semantics—but ‘pragmatics’ in the sense of the study of the use of expressions in virtue of which they are meaningful at all. To the formal, mathematically inspired tradition of Frege, Russell, Carnap, and Tarski, culminating in model-theoretic semantics, is opposed an anthropological, natural-historical, social-practical inquiry aimed both at demystifying our discursive doings, and at deflating philosophers’ systematic and theoretical ambitions regarding them. I think that philosophers of language have tended to draw this opposition in the starkest possible terms, treating these approaches as mutually exclusive, hence as requiring that a choice be made between them. Those who are moved by the pragmatist picture generally accept the particularist, quietist conclusions Wittgenstein seems to have drawn from it. And those committed to some version of the project of semantic analysis have felt obliged to deny the significance of pragmatics in this sense, or at the least to dismiss it as irrelevant to properly semantic concerns. In the most extreme cases, the attitudes of anti-pragmatist philosophers of language to Wittgenstein’s picture verges on that of the Victorian lady to Darwin’s theory: One hopes that it is not true, and that if it is true, at least that it not become generally known. But I do not think we are obliged to choose between these approaches. They should be seen as complementing rather than competing with one another. Semantics and pragmatics, concern with meaning and concern with use, ought to be understood as aspects of one more comprehensive picture of the discursive. Pragmatist considerations do not oblige us to focus on pragmatics to the exclusion of semantics; we can deepen our semantics by the ad-
Towards an Analytic Pragmatism
21
dition of pragmatics. If we extract consequences from the pragmatists’ observations somewhat more modestly and construe the analytic project somewhat more broadly, the two will be seen not only as compatible, but as mutually illuminating. If we approach the pragmatists’ observations in an analytic spirit, we can understand pragmatics as providing special resources for extending and expanding the analytic semantic project from exclusive concern with relations among meanings to encompass also relations between meaning and use. In its most ambitious form, such an enterprise would aspire to articulate something like a logic of the relations between meaning and use. If we leave open the possibility that the use of some vocabulary may be illuminated by taking it to express some sort of meaning or content—that is, if we do not from the beginning embrace semantic nihilism—then the most important positive pragmatist insight will be one complementary to the methodological pragmatism I have already identified. The thought underlying the pragmatist line of thought is that what makes some bit of vocabulary mean what it does is how it is used. What we could call semantic pragmatism is the view that the only explanation there could be for how a given meaning gets associated with a vocabulary is to be found in the use of that vocabulary: the practices by which that meaning is conferred or the abilities whose exercise constitutes deploying a vocabulary with that meaning. To broaden the classical project of analysis in the light of the pragmatists’ insistence on the centrality of pragmatics, we can focus on this fundamental relation between use and meaning, between practices or practical abilities and vocabularies. We must look at what it is to use locutions as expressing meanings—that is, at what one must do in order to count as saying what the vocabulary lets practitioners express. I am going to call this kind of relation “practice-vocabulary sufficiency”—or usually, “PV-sufficiency,” for short. It obtains when engaging in a specified set of practices or exercising a specified set of abilities is sufficient for someone to count as deploying a specified vocabulary. Of course it matters a lot how we think about these content-conferring, vocabulary-deploying practices or abilities. The semantic pragmatist’s claim that use confers meaning (so talk of practices or the exercise of abilities as deploying vocabularies) reverts to triviality if we are allowed to talk about “using the tilde to express negation,” “the ability to mean red by the word ‘red’,” or “the capacity to refer to electrons by the word ‘electron’,”—or, I think,
22
Robert Brandom
even the intention so to refer. And that is to say that the interest of the PVsufficiency of some set of practices or abilities for the deploying of a vocabulary is quite sensitive to the vocabulary in which we specify those practices-orabilities. Talk of practices-or-abilities has a definite sense only insofar as it is relativized to the vocabulary in which those practices-or-abilities are specified. And that means that besides PV-sufficiency, we should admit a second basic meaning-use relation: “vocabulary-practice sufficiency,” or just “VP-sufficiency,” is the relation that holds between a vocabulary and a set of practicesor-abilities when that vocabulary is sufficient to specify those practices-orabilities. VP-sufficient vocabularies that specify PV-sufficient practices let one say what it is one must do to be engaging in those practices or exercising those abilities, and so to deploy a vocabulary. PV-sufficiency and VP-sufficiency are two basic meaning-use relations (MURs). In terms of those basic relations, we can define a more complex relation: the relation that holds between vocabulary V’ and vocabulary V when V’ is VP-sufficient to specify practices-or-abilities P that are PV-sufficient to deploy vocabulary V. This VV-relation is the composition of the two basic MURs. When it obtains I will say that V’ is a pragmatic metavocabulary for V. It allows one to say what one must do in order to count as saying the things expressed by vocabulary V. We can present this relation graphically in a meaninguse diagram (MUD): Meaning-Use Diagram #1: Pragmatic Metavocabulary
V Res 1:VV-1,2
V'
2: VP-suff
1: PV-suff
P
Towards an Analytic Pragmatism
23
The conventions of this diagram are: – Vocabularies are shown as ovals, practices-or-abilities as (rounded) rectangles. – Basic meaning-use relations are indicated by solid arrows, numbered and labeled as to kind of relation. – Resultant meaning-use relations are indicated by dotted arrows, numbered, and labeled as to kind and the basic MURs from which they result. The idea is that a resultant MUR is the relation that obtains when all of the basic MURs listed on its label obtain. Being a pragmatic metavocabulary is the simplest species of the genus I want to introduce here. It is a pragmatically mediated semantic relation between vocabularies. It is pragmatically mediated by the practices-or-abilities that are specified by one of the vocabularies (which say what counts as doing that) and that deploy or are the use of the other vocabulary (what one says by doing that). The relation that is established thereby between the two vocabularies is of a distinctive sort, quite different from, for instance, definability, translatability, reducibility, and supervenience. My basic suggestion for extending the classical project of analysis so as to incorporate as essential positive elements the insights that animate the pragmatist critique of that project is that alongside these classical semantic relations between vocabularies we consider pragmatically mediated ones, of which the relation of being a pragmatic metavocabulary is a paradigm. Under what circumstances would this simplest pragmatically mediated semantic relation be philosophically interesting, when considered in connection with the sorts of vocabularies that have been of most interest to classical analysis? At least one sort of result that could be of considerable potential significance, I think, is if it turned out that in some cases pragmatic metavocabularies exist that differ significantly in their expressive power from the vocabularies for the deployment of which they specify sufficient practices-orabilities. I will call that phenomenon “pragmatic expressive bootstrapping.” If one vocabulary is strictly weaker in expressive power than the other, I will call that strict expressive bootstrapping. We are familiar with this sort of phenomenon in ordinary semantics, where sometimes a semantic metalanguage differs substantially in expressive power from its object language—for instance, where we can produce an extensional metalanguage for intensional languages, as in
24
Robert Brandom
the case of possible worlds semantics for modality. One example of a claim of this shape in the case of pragmatically mediated semantic relations—though of course it is not expressed in terms of the machinery I have been introducing—is Huw Price’s pragmatic normative naturalism.8 He argues, in effect, that although normative vocabulary is not reducible to naturalistic vocabulary, it might be possible to say in wholly naturalistic vocabulary what one must do in order to be using normative vocabulary. If such a claim about the existence of an expressively bootstrapping naturalistic pragmatic metavocabulary for normative vocabulary could be made out, it would evidently be an important chapter in the development of the naturalist core program of the classical project of philosophical analysis. It would be a paradigm of the sort of payoff we could expect from extending that analytic project by including pragmatically mediated semantic relations. The meaning-use diagram of the pragmatically mediated semantic relation of being a pragmatic metavocabulary illustrates an analysis of that relation. It exhibits that relation as the resultant, by composition, of the two basic meaning-use relations of PV-sufficiency and VP-sufficiency. A complex MUR is analyzed as the product of operations applied to basic MURs. This is meaning-use analysis. The same analytic apparatus applies also to more complex pragmatically mediated semantic relations. Consider one of the pragmatist criticisms that Sellars addresses to the empiricist core program of the classical analytic project. It turns on the assertion of the pragmatic dependence of one set of practices-or-abilities on another. Because one cannot withhold a commitment that one cannot undertake, given that he thinks part of what one is doing in saying how things merely appear is withholding a commitment to their actually being that way, Sellars concludes that one cannot have the ability to do that unless one also has the ability to say how things actually are. In effect, this Sellarsian pragmatist critique of the phenomenalist form of empiricism consists in the claim that the practices that are PV-sufficient for ‘is’-φ talk are PP-necessary for the practices that are PV-sufficient for ‘looks’-φ talk. That pragmatic dependence of practices-or-abilities then induces a resultant pragmatically mediated semantic relation between the vocabularies. The meaning-use diagram for this claim is: 8
Cf. “Naturalism without Representationalism”.
Towards an Analytic Pragmatism
25
Meaning-Use Diagram #2: Pragmatically Mediated Semantic Presupposition
Vlooks-φ
Res 1: VV 1,2,3
3: PV-suff
Plooks-φ
Vis-φ 1: PV-suff
Pis-φ
2: PP-nec
The resultant MUR here is a kind of complex, pragmatically mediated, VVnecessity, or presupposition. In fact, although Sellars’s argument for the crucial PP-necessity relation of pragmatic dependence of one set of practices-or-abilities on another is different, his argument against the observational version of empiricism—the claim that purely non-inferential, observational uses do not form an autonomous discursive practice, but presuppose inferential uses—has exactly the same form: Meaning-Use Diagram #3: Pragmatically Mediated Semantic Presupposition Vobse rvational
Res 1: VV 1,2,3
3: PV-suff
P observational
Vinferential
1: PV-suff
2: PP-nec
Pinferential
For these cases, we can say something further about the nature of the pragmatically mediated semantic relation that is analyzed as the resultant MUR in
26
Robert Brandom
these diagrams. For instead of jumping directly to this VV resultant MUR, we could have put in the composition of the PP-necessity and second PVsufficiency relation, yielding a kind of complex pragmatic presupposition:
Meaning-Use Diagram #4: Composition
Vlooks-φ 3: PV-suff
Plooks-φ
Vis-φ Res 2: PV 2,3
1: PV-suff
Pis-φ
2: PP-nec
If this diagram were completed by an arrow from Vis-φ to Vlooks-φ such that the same diagonal resultant arrow could represent both the composition of relations 2 and 3 and the composition of relation 1 and the newly supplied one, then category theorists would say that the diagram commutes. And the arrow that needs to be supplied to make the diagram commute they call the retraction of relation 1 through the composition Res2: Meaning-Use Diagram #5: Composition and Retraction
Vlooks-φ 3: PV-suff
Plooks-φ
Retraction of 1 through Res 2
Res 2: PV 2,3
2: PP-nec
Vis-φ 1: PV-suff
Pis-φ
Towards an Analytic Pragmatism
27
After composition, then, the next most complex form of resultant MUR is retraction. Analyzing the structure of Sellars’s pragmatist arguments against empiricism requires recognizing the pragmatically mediated semantic relation he claims holds between phenomenal and objective vocabulary as the retraction of a constellation of more basic meaning-use relations.
4
AUTOMATA: SYNTACTIC PV-SUFFICIENCY AND VP-SUFFICIENCY
Now this is all extremely abstract. To make it more definite, we need to fill in (at least) the notions of vocabulary, practice-or-ability, PV-sufficiency, and VP-sufficiency, which are the fundamental elements that articulate what I am calling the “meaning-use analysis” of resultant meaning-use relations—in particular, the pragmatically mediated semantic relations between vocabularies that I am claiming we must acknowledge in order to pursue the classical project of philosophical analysis. We can begin to do that by looking at a special case in which it is possible to be unusually clear and precise about the things and relations that play these metatheoretic roles. This is the case where ‘vocabulary’ takes a purely syntactic sense. Of course, the cases we eventually care about involve vocabularies understood in a sense that includes their semantic significance. But besides the advantages of clarity and simplicity, we will find that some important lessons carry over from the syntactic to the semantic case. The restriction to vocabularies understood in a spare syntactic sense leads to correspondingly restricted notions of what it is to deploy such a vocabulary, and what it is to specify practices-or-abilities sufficient to deploy one. Suppose we are given an alphabet, which is a finite set of primitive sign types—for instance, the letters of the English alphabet. The universe generated by that alphabet then consists of all the finite strings that can be formed by concatenating elements drawn from the alphabet. A vocabulary over such an alphabet—in the syntactic sense I am now after—is then any subset of the universe of strings that alphabet generates. If the generating alphabet is the
28
Robert Brandom
English alphabet, then the vocabulary might consist of all English sentences, or all possible English texts.9 What can we say about the abilities that count as deploying a vocabulary in this spare syntactic sense?10 The abilities in question are the capacity to read and write the vocabulary. In this purely syntactic sense, ‘reading’ it means being able practically to distinguish within the universe generated by the vocabulary, strings that do, from those that do not, belong to the specified vocabulary. And ‘writing’ it means practically being able to produce all and only the strings in the alphabetic universe that do belong to the vocabulary. We assume as primitive abilities the capacities to read and write, in this sense, the alphabet from whose universe the vocabulary is drawn—that is, the capacity to respond differentially to alphabetic tokens according to their type, and to produce tokens of antecedently specified alphabetic types. Then the abilities that are PV-sufficient to deploy some vocabularies can be specified in a particularly simple form. They are finite-state automata (FSAs). As an example, suppose we begin with the alphabet {a, h, o, !}. Then we can consider the laughing Santa vocabulary, which consists of strings such as ‘hahaha!’, ‘hohoho!’, ‘hahahoho!’ ‘hohoha!’, and so on.11 Here is a graphical representation of a laughing Santa finite-state automaton, which can read and write the laughing Santa vocabulary:
9
10 11
Computational linguists, who worry about vocabularies in this sense, have developed metalanguages for specifying important classes of such vocabularies: the syntactic analogues of semantic metalanguages in the cases we will eventually address. So, for instance, for the alphabet {a,b}, ‘anbn’ characterizes the vocabulary that comprises all strings of some finite number of ‘a’s followed by the same number of ‘b’s. ‘a(ba)*b’ characterizes the vocabulary that comprises all strings beginning with an ‘a’, ending with a ‘b’, and having any number of repetitions of the sub-string ‘ba’ in between. Here we can safely just talk about abilities, without danger of restricting the generality of the analysis. In the syntactic metalanguage for specifying vocabularies that I mentioned in the note above, this is the vocabulary (ha/ho)*!
Towards an Analytic Pragmatism
29
The Laughing Santa Automaton
a 1
h
2
o
3
!
4
h The numbered nodes represent the states of the automaton, and the alphabetically labeled arcs represent state-transitions. By convention, the starting state is represented by a square (State 1), and the final state by a circle with a thick border (State 4). As a reader of the laughing Santa vocabulary, the task of this automaton is to process a finite string, and determine whether or not it is a licit string of the vocabulary. It processes the string one alphabetic character at a time, beginning in State 1. It recognizes the string if and only if (when and only when) it arrives at its final state, State 4. If the first character of the string is not an ‘h’, it remains stuck in State 1, and rejects the string. If the first character is an ‘h’, it moves to State 2, and processes the next character. If that character is not an ‘a’ or an ‘o’, it remains stuck in State 2, and rejects the string. If the character is an ‘a’ or an ‘o’, it moves to State 3. If the next character is an exclamation point, it moves to State 4, and recognizes the string ‘ha!’ or ‘ho!’— the shortest ones in the laughing Santa vocabulary. If instead the next character is an ‘h’, it goes back to State 2, and repeats itself in loops of ‘ha’s and ‘ho’s any number of times until an exclamation point is finally reached, or it is fed a discordant character. As a writer of the laughing Santa vocabulary, the task of the automaton is to produce only licit strings of that vocabulary, by a process that can produce any and all such strings. It begins in its initial state, State 1, and emits an ‘h’ (its only available move), changing to State 2. In this state, it can produce ei-
30
Robert Brandom
ther an ‘a’ or an ‘o’—it selects one at random12—and goes into State 3. In this state, it can either tack on an exclamation point, and move into its final state, State 4, finishing the process, or emit another ‘h’ and return to State 2 to repeat the process. In any case, whenever it reaches State 4 and halts, the string it has constructed will be a member of the laughing Santa vocabulary. I hope this brief rehearsal makes it clear how the constellation of nodes and arrows that makes up this directed graph represents the abilities to read and write (recognize and produce arbitrary strings of) the laughing Santa vocabulary.13 What it represents is abilities that are PV-sufficient to deploy that vocablary—that is, read and write it, in the attenuated sense appropriate to this purely syntactic case. And the digraph representation is itself a vocabulary that is VP-sufficient to specify those vocabulary-deploying abilities. That is, the digraph representation of this finite-state automaton is a pragmatic metavocabulary 12
13
As a matter of fact, it can be shown that every vocabulary readable/writeable by a non-deterministic finite-state automaton—such as the laughing Santa automaton— is also readable/writeable by a deterministic one. For practice, or to test one’s grip on the digraph specification of FSAs, consider what vocabulary over the same alphabet that produces the laughing Santa is recognized/produced by this automaton:
The "I'll Have What She's Having" Automaton
o
h
2
o 1
4
a a
3
h o
!
5
Towards an Analytic Pragmatism
31
for the laughing Santa vocabulary. The relation between the digraph vocabulary and the laughing Santa vocabulary is, then, a pragmatically mediated—not now semantic, but syntactic—relation between vocabularies. It may seem that I am stretching things by calling the digraph form of representation a ‘vocabulary’. It will be useful, as a way of introducing my final point, to consider a different form of pragmatic metavocabulary for the laughing Santa vocabulary. Besides the digraph representation of a finite-state automaton, we can also use a state-table representation. For the laughing Santa automaton this is:
a h o !
State 1 Halt 2 Halt Halt
State 2 3 Halt 3 Halt
State 3 Halt 2 Halt 4
In read mode, the automaton starts in State 1. To see what it will do if fed a particular character, we look at the row labeled with that character. The LSA will Halt if the input string starts with anything other than an ‘h’, in which case it will change to State 2. In that state, the automaton specified by the table will halt unless the next character is an ‘a’ or an ‘o’, in which case it changes to State 3, and so on. (There is no column for State 4, since it is the final state, and accepts/produces no further characters.) Clearly there is a tabular representation corresponding to any digraph representation of an FSA, and vice versa. Notice further that we need not use a two-dimensional table to convey this information. We could put the rows one after another, in the form: aHalt3Halth2Halt2oHalt3Halt!HaltHalt4. This is just a string, drawn from a universe generated by the alphabet of the LSA, together with ‘Halt’ and the designations of the states of that automaton. The strings that specify finite-state automata that deploy vocabularies defined over the same basic alphabet as the LSA then form a vocabulary in the
32
Robert Brandom
technical syntactic sense we have been considering. And that means we can ask about the automata that can read and write those state-table encoding vocabularies. The meaning-use diagram for this situation is then: Meaning-Use Diagram #6: Specifying the Automaton that Deploys the Laughing Santa Vocabulary
VLaughing Santa Res 1: VV 1,2 1: PV-suff
VLSA StateTable
2: VP-suff
PLaughing Santa Automaton
3: PV-suff
PLSA State-Table Automaton
5
THE CHOMSKY HIERARCHY AND A SYNTACTIC EXAMPLE OF PRAGMATIC EXPRESSIVE BOOTSTRAPPING
Restricting ourselves to a purely syntactic notion of a vocabulary yields a clear sense of ‘pragmatic metavocabulary’: both the digraph and the state-table vocabularies are VP-sufficient to specify practical abilities articulated as a finitestate automaton that is PV-sufficient to deploy—in the sense of recognizing and producing—the laughing Santa vocabulary, as well as many others. Perhaps surprisingly, it also offers a prime example of strict pragmatic expressive bootstrapping. For in this setting we can prove that one vocabulary that is expressively weaker than another can nonetheless serve as an adequate pragmatic
Towards an Analytic Pragmatism
33
metavocabulary for that stronger vocabulary. That is, even though one cannot say in the weaker vocabulary everything that can be said in the stronger one, one can still say in the weaker one everything that one needs to be able to do in order to deploy the stronger one. Here the relevant notion of the relative expressive power of a vocabulary is also a purely syntactic one. Already in the 1950’s, Chomsky offered mathematical characterizations of the different sets of strings of characters that could be generated by different classes of grammars (that is, in my terms, characterized by different kinds of syntactic metavocabularies) and computed by different kinds of automata. The kinds of vocabulary, grammar, and automata lined up with one another, and could be arranged in a strict expressive hierarchy: the Chomsky hierarchy. It is summarized in the following table: Vocabulary Regular
Grammar AÆaB AÆa AÆ
Automaton Finite State Automaton Context-Free Push-Down Automaton Context-Sensitive c1Ac2Æc1c2 Linear Bounded Automaton Recursively Enumer- No Restrictions on Ru- Turing Machine able les (= 2 Stack PDA) The point I want to make fortunately does not require us to delve very deeply into the information summarized in this table. A few basic points will suffice. The first thing to realize is that not all vocabularies in the syntactic sense we have been pursuing can be read and written by finite-state automata. For instance, it can be shown that no finite-state automaton is PV-sufficient to deploy the vocabulary anbn, defined over the alphabet {a,b}, which consists of all strings of any arbitrary number of ‘a’s followed by the same number of ‘b’s. The idea behind the proof is that in order to tell whether the right number of ‘b’s follow the ‘a’s (when reading) or to produce the right number of ‘b’s (when writing), the automaton must somehow keep track of how many ‘a’s have been processed (read or written). The only way an FSA can store in-
34
Robert Brandom
formation is by being in one state rather than another. So, it could be in one state—or in one of a class of states—if one ‘a’ has been processed, another if two have, and so on. But by definition, a finite-state automaton only has a finite number of states, and that number is fixed in advance of receiving its input or producing its output. Whatever that number of states is, and whatever system it uses to code numbers into states (it need not be one-to-one—it could use a decimal coding, for instance), there will be some number of ‘a’s that is so large that the automaton runs out of states before it finishes counting. But the vocabulary in question consists of arbitrarily long strings of ‘a’s and ‘b’s. In fact, it is possible to say exactly which vocabularies finite-state automata (specifiable by digraphs and state-tables of the sort illustrated above) are capable of deploying. These are called the ‘regular’ vocabularies (or languages). The next point is that slightly more complex automata are capable of deploying vocabularies, such as anbn, that are not regular, and hence cannot be read or written by finite-state automata. As our brief discussion indicated, intuitively the problem FSAs have with languages like anbn is that they lack memory. If we give them a memory, we get a new class of machines: (nondeterministic14) push-down automata (PDAs). In addition to being able to respond differentially to and produce tokenings of the alphabetic types, and being able to change state, PDAs can push alphabetic values to the top of a memory-stack, and pull such values from the top of that stack. PDAs can do everything that finite-state automata can do, but they can also read and write many vocabularies, such as anbn, that are not regular, and so cannot be read and written by FSAs. The vocabularies they can deploy are called “context-free”. All regular vocabularies are context-free, but not vice versa. This proper containment of classes of vocabularies provides a clear sense, suitable to this purely syntactic setting, in which one vocabulary can be thought of as “expressively more powerful” than another: the different kinds of grammar can specify, and the different kinds of automata can compute, ever larger classes of vocabularies. Context-free vocabularies that are not regular require more powerful grammars to specify them, as well as more powerful automata to deploy 14
By contrast to FSAs, there need not in general be for every vocabulary computable by a non-deterministic PDA, some deterministic PDA that reads and writes the same vocabulary.
Towards an Analytic Pragmatism
35
them.15 FSAs are special kinds of PDAs, and all the automata are special kinds of Turing Machines. Recursively enumerable vocabularies are not in general syntactically reducible to context-sensitive, context-free, or regular ones. And the less capable automata cannot read and write the all the vocabularies that can be read and written by Turing Machines. Nonetheless, if we look at pragmatically mediated relations between these syntactically characterized vocabularies, we find that they make possible a kind of strict expressive bootstrapping that permits us in a certain sense to evade the restrictions on expressive power enforced for purely syntactic relations between vocabularies. The hierarchy dictates that only the abilities codified in Turing Machines—two-stack push-down automata—are PV-sufficient to deploy recursively enumerable vocabularies in general. But now we can ask: what class of languages is VP-sufficient to specify Turing Machines, and hence to serve as sufficient pragmatic metavocabularies for recursively enumerable vocabularies in general? The surprising fact is that the abilities codified in Turing Machines—the abilities to recognize and produce recursively enumerable vocabularies—can quite generally be specified in context-free vocabularies. It is demonstrable that context-free languages are strictly weaker in syntactic expressive resources than recursively enumerable languages. The push-down automata that can read and write only context-free languages cannot read and write recursively enumerable languages in general. But it is possible to say in a context-free language what one needs to do in order to deploy recursively enumerable languages in general. The proof of this claim is tedious, but not difficult, and the claim itself is not at all controversial—though computational linguists make nothing of it, having theoretical concerns very different from those that lead me to underline this fact. (My introductory textbook leaves the proof as an exercise to the reader.16) General-purpose computer languages such as Pascal and C++ can specify the algorithms a Turing Machine, or any other universal computer, uses to compute any recursively enumerable function, hence to recognize or produce any recursively enumerable vocabulary. And they are invariably con15
16
Regular languages are those that can be specified entirely by production rules of the form: AÆaB and AÆa. Context-free languages relax these restrictions, allowing production rules of the form: AÆ. Thomas A. Sudkamp. Languages and Machines 2nd edition, Chapter 10.
36
Robert Brandom
text-free languages—in no small part just because the simplicity of this type of grammar makes it easy to write parsers for them. Yet they suffice to specify the state-table, contents of the tape (or of the dual stacks), and primitive operations of any and every Turing Machine. Here is the MUD characterizing this pragmatically mediated relation between syntactically characterized vocabularies: Meaning-Use Diagram #7: Syntactic Pragmatic Expressive Bootstrapping
VRecursively Enumerable Res 1: VV 1,2 1: PV-suff
VContextFree
2: VP-suff
P Turing Machine
3: PV-suff
PPush-Down Automaton
I called the fact that context-free vocabularies can be adequate pragmatic metavocabularies for recursively enumerable vocabularies in general ‘surprising’, because of the provable syntactic irreducibility of the one class of vocabularies to the other. But if we step back from the context provided by the Chomsky hierarchy, we can see why the possibility of such pragmatic expressive bootstrapping should not, in the end, be surprising. For all the result really means is that context-free vocabularies let one say what it is one must do in order to say things they cannot themselves say, because the ability to deploy those context-free vocabularies does not include the abilities those vo-
Towards an Analytic Pragmatism
37
cabularies let one specify. Thus, for instance, there is no reason that an FSA could not read and write a vocabulary that included commands such as “Push an ‘a’ onto the stack,”—and thus specify the program of a PDA—even though it itself has no stack, and could not do what the vocabulary it is deploying specifies. A coach might be able to tell an athlete exactly what to do, and even how to do it, even though the coach cannot himself do what he is telling the athlete to do, does not have the abilities he is specifying. We ought not to boggle at the possibility of an expressively weaker pragmatic metavocabulary having the capacity to say what one must do in order to deploy an expressively stronger one. We should just look to see where this seems in fact to be possible for vocabularies we care about, and what we can learn from such relations when they do obtain.
6
SEMANTIC EXAMPLES OF PRAGMATIC EXPRESSIVE BOOTSTRAPPING AND FURTHER BASIC AND RESULTANT MURS Let us recall what motivated this rehearsal of some elements of automaton theory and introductory computational linguistics. I suggested that a way to extend the classical project of semantic analysis so as to take account of the insights of its pragmatist critics is to look analytically at relations between meaning and use. More specifically, I suggested focusing to begin with on two in some sense complementary relations: the one that holds when some set of practices-or-abilities is PV-sufficient to deploy a given vocabulary, and the one that holds when some vocabulary is VP-sufficient to specify a given set of practices-or-abilities. The composition of these is the simplest pragmatically mediated semantic relation between vocabularies: the relation that holds when one vocabulary is a sufficient pragmatic metavocabulary for another. It is a paradigm of the infinite, recursively generable class of complex, pragmatically mediated semantic relations that I propose to lay alongside the other semantic relations between vocabularies that have been investigated by analytic philosophers (for instance those who address the core programs of empiricism and naturalism): relations such as analyzability, definition, translation, reduction, truthmaking, and supervenience. I suggested further that pragmatic metavocabu-
38
Robert Brandom
laries might be of particular interest in case they exhibited what I called “expressive bootstrapping”—cases, that is, in which the expressive power of the pragmatic metavocabulary differs markedly from that of the target vocabulary, most strikingly, when the metavocabulary is substantially expressively weaker—a phenomenon Tarski has led us not to expect for semantic metavocabularies. We have now seen that all of these notions can be illustrated with particular clarity for the special case of purely syntactically characterized vocabularies. The abilities that are PV-sufficient to deploy those vocabularies, in the sense of the capacity to recognize and produce them, can be thought of as various sorts of automata. There are several well-established, different-butequivalent vocabularies that are known to be VP-sufficient to specify those automata. In this special syntactic case we can accordingly investigate the properties of pragmatic metavocabularies, and when we do, we find a striking instance of strict expressive bootstrapping in a pragmatically mediated syntactic relation between vocabularies. Of course, the cases we really care about involve semantically significant vocabularies. Are there any interesting instances of these phenomena in such cases? I have indicated briefly how some of Sellars’s pragmatist criticisms of various ways of pursuing the empiricist program can be understood to turn on pragmatically mediated semantic relations. And I mentioned Huw Price’s idea that although normative vocabulary is not semantically reducible to naturalistic vocabulary, naturalistic vocabulary might suffice to specify what one must do—the practices-or-abilities one must engage in or exercise—in order to deploy normative vocabulary. Here is another example that I want to point to, though I cannot develop the claim here. For roughly the first threequarters of the twentieth century, philosophers who thought about indexical vocabulary took for granted some version of the doctrine that a tokening n of an expression of the type ‘now’ was synonymous with, definable or semantically analyzable as, ‘the time of utterance of n,’ and similarly for ‘here’ and ‘the place of utterance of h,’ and so on. During the 1970’s philosophers such as John Perry, David Lewis, and G. E. M. Anscombe, by focusing on the use of indexicals in modal and epistemic contexts, showed decisively that this cannot be right: what is expressed by indexical vocabulary cannot be expressed equivalently by non-indexical vocabulary. This fact seems so obvious to us
Towards an Analytic Pragmatism
39
now that we may wonder what philosophers such as Russell, Carnap, and Reichenbach could have been thinking. I want to suggest that the genuine phenomenon in the vicinity is a pragmatically mediated semantic relation between these vocabularies. Specifically, in spite of the semantic irreducibility of indexical to nonindexical vocabulary, it is possible to say, entirely in nonindexical terms, what one must do in order to be deploying indexical vocabulary correctly: to be saying essentially indexical things. For we can formulate practical rules such as: If, at time t and place , speaker s wants to assert that some property P holds of , it is correct to say “P holds of me, here and now.” Non-indexical vocabulary can serve as an adequate pragmatic metavocabulary for indexical vocabulary. The fact that one nonetheless cannot say in nonindexical terms everything that one can say with indexical vocabulary just shows that these vocabularies have different expressive powers, so that the pragmatically mediated semantic relation between them is a case of pragmatic expressive bootstrapping. Besides pragmatically mediated semantic relations between vocabularies, there is another sort of pragmatic analysis, which relates one constellation of practices-or-abilities to another. It corresponds to another basic meaning-use relation: the kind of PP-sufficiency that holds when having acquired one set of abilities means one can already do everything one needs to do, in principle, to be able to do something else. One concrete way of filling in a definite sense of “in principle” is by algorithmic elaboration, where exercising the target ability just is exercising the right basic abilities in the right order and under the right circumstances. As an example, the ability to do long division just consists in exercising the abilities to do multiplication and subtraction according to a particular conditional branched-schedule algorithm. The practical abilities that implement such an algorithmic PP-sufficiency relation are just those exercised by a finite-state automaton. Indeed, automata are defined by a definite set of meta-abilities: abilities to elaborate a set of primitive abilities into a set of more complex ones, which can accordingly be pragmatically analyzed in terms of or decomposed into the other.
40
Robert Brandom
To get a usefully general concept of the PP-sufficiency of a set of basic abilities for a set of more complex ones, we need to move beyond the purely syntactic automata I have described so far. One way to do that is to replace their specialized capacities to read and write symbols—in the minimal sense of classifying tokens as to types and producing tokens of specified types—by more general recognitional and productive capacities. These are abilities to respond differentially to various non-symbolic stimuli (for instance, the visible presence of red things), corresponding to reading, and to respond by producing performances of various non-symbolic kinds (for instance, walking north for a mile), corresponding to writing. What practically implements the algorithmic elaboration of such a set of basic differential responsive abilities is a finite state transducing automaton (and its more sophisticated push-down brethren). A Finite-State Transducing Automaton S1:R7 1
S4:R6 2
__:R7 S7:__ 3
5 4
S1:R3
6
S3:__
This is a diagram of an FSTA that has an initial set of stimuli to which it can respond differentially, and an initial set of responses it can differentially produce. And the diagram indicates that in its initial state, if presented with a stimulus of kind 1, it will produce a response of kind 7 and shift to state 2, and if presented instead with a stimulus of kind 7 it will produce no response, but will shift to state 3. It is important to note that although the recognitive and performative abilities that such an automaton algorithmically elaborates are to be considered as ‘primitive’ or ‘basic’ with respect to such elaboration,
Towards an Analytic Pragmatism
41
this does not mean that they are so in any absolute sense. The stimulusresponse formulation by itself does not keep us from considering as ‘primitive’ capacities the abilities to keep ourselves at a suitable distance from a conversational partner, distinguish cubist paintings done by Braque from those done by Picasso, drive from New York to San Francisco, or build a house. The notion of the algorithmic decomposability of some practices-orabilities into others suggests in turn a pragmatic generalization of the classical program of artificial intelligence functionalism—which, though a latecomer in the twentieth century, deserves, I think, to count as a third core program of classical semantic analysis, alongside empiricism and naturalism. AI functionalism traditionally held itself hostage to a commitment to the purely symbolic character of intelligence in the sense of sapience. But broadening our concern from automata as purely syntactic engines to the realm of transducing automata, we are now in a position to see automaton functionalism as properly concerned with the algorithmic decomposability of discursive (that is, vocabularydeploying) practices-and-abilities. What I will call the ‘pragmatic’ thesis of artificial intelligence is that the ability to engage in some autonomous discursive practice—a language game one could play though one played no other—can be algorithmically decomposed into non-discursive abilities. By non-discursive abilities, I mean abilities each of which can in principle be exhibited by something that does not engage in any autonomous discursive practice. (Without that restriction on the primitive abilities out of which discursive ones are to be algorithmically elaborated, the claim would be trivial, since the null algorithmic decomposition is also a decomposition.) The capacity to talk-andthink as I am addressing it is the capacity to deploy an autonomous vocabulary. But unlike classical symbolic AI, the pragmatic thesis of artificial intelligence does not presume that the practical capacities from which some transducing automaton can elaborate the ability to engage in an autonomous discursive practice must themselves consist of symbol-manipulating abilities, never mind ultimately syntactic ones. The algorithmic practical elaboration model of AI gives a relatively precise shape to the pragmatist program of explaining knowing-that in terms of knowing-how: specifying in a nonintentional, non-semantic vocabulary what it is one must do in order to count as deploying some vocabulary, hence as making intentional and semantic vo-
42
Robert Brandom
cabulary applicable to the performances one produces. In particular, it offers a construal of the basic claim of AI-functionalism as an expressive bootstrapping claim about computer languages as pragmatic metavocabularies for much more expressively powerful autonomous vocabularies, namely natural languages. The arguments for and against this pragmatic version of AIfunctionalism accordingly look quite different from those arrayed on the opposing sides of the debate about the prospects of symbolic AI. Combining the notion of PP-sufficiency that holds between two constellations of practices-or-abilities when one can be algorithmically elaborated from the other with the two sorts of basic meaning-use relations out of which we previously constructed the notion of expressively bootstrapping pragmatic metavocabularies—namely, a set of practices-or-abilities being PV-sufficient to deploy a vocabulary and a vocabulary being VP-sufficient to specify a set of practices-or-abilities—makes it possible to define further kinds of pragmatically mediated semantic relations. As an example, consider the relation between logical vocabulary—paradigmatically, conditionals—and ordinary, nonlogical, empirical descriptive vocabulary. I take it that every autonomous discursive practice must include performances that have the pragmatic significance of assertions and inferences (which I would argue come as an indissoluble package). I actually think this PP-necessary condition on any practices PVsufficient for autonomously deploying a vocabulary can usefully be treated as sufficient as well—that is, as what distinguishes discursive practices as such. But nothing in what follows turns on that further commitment. To count as engaging in such practices, practitioners must exercise an ability, however fallible, to assess the goodness of material inferences: to sort them into those they accept and those they reject. This is part of what one must do in order to say anything. But it is easy to say how those recognitional and performative abilities, counted for these purposes as primitive, can be algorithmically elaborated into the capacity to use conditionals. An algorithm VP-sufficient to specify an automaton that practically implements such a pragmatic elaboration or PPsufficiency relation is the following: Assert the conditional ‘if p then q’ if one endorses the inference from p to q; Endorse the inference from p to q if one asserts the conditional ‘if p then q’.
Towards an Analytic Pragmatism
43
These rules of usage codify introduction and elimination rules for the conditional. So the capacity to use conditionals can be algorithmically elaborated from the capacities to make assertions and assess inferences. This is the composition of a PP-sufficiency relation with a PV-sufficiency relation, and is expressed in the following meaning-use diagram: Elaborating Conditionals Vconditionals
Res1:VV 1-4
V1 1: PV-suff
4: PV-suff
2: PV-nec P conditionals
PAlgEl 3: PP-suff
Pinferring/ asserting
5: PV-suff
P ADP
VAlgorithm
The complex resultant meaning-use relation indicated by the dotted arrow at the top of the diagram is a further pragmatically mediated semantic relation. The diagram indicates exactly what constellation of sub-claims about basic meaning-use relations must be justified in order to justify the claim that this relation obtains between two vocabularies, and hence the diagram graphically presents a distinctive kind of meaning-use analysis of it. In fact, if we think further about this example, by filling in another basic meaning-use relation that obtains in this case, we can define an even more articulated pragmatically mediated semantic relation between vocabularies. For when conditionals are deployed with the practical circumstances and consequences of application specified in the algorithm stated above, they let practitioners say what otherwise they could only do; that is, they express explicitly, in the form of a claimable, hence propositional, content, what practitioners are implicitly doing in endorsing some material inferences and rejecting others. This is a VP-sufficiency relation: conditionals let one specify the practices of
44
Robert Brandom
taking-or-treating inferences as materially good or bad. Adding in this explicating relation between conditionals and the practices-or-abilities they make explicit yields a new pragmatically mediated semantic relation that conditionals stand in to every autonomously deployable vocabulary. Its meaning-use diagram is this: Elaborated-Explicating (LX) Conditionals Vconditionals
4: PV-suff
Res1:VV 1-5
5: VP-suff
V1 1: PV-suff 2: PV-nec
P conditionals
PAlgEl 3: PP-suff 6: PV-suff
P inferring P ADP
VAlgorithm
The capacity to deploy conditionals (that is, something PV-sufficient for their use) both can be elaborated from practices PP-necessary for every ADP, and explicates those practices (in the sense of being VP-sufficient for them). It is elaborated-explicative relative to every autonomous vocabulary. We say, it is LX for every AV, hence for every vocabulary (since the use of any vocabulary presupposes, and in that sense is parasitic on, the capacity to use some autonomous vocabulary). I believe that this complex resultant pragmatically mediated semantic relation is important for understanding the distinctive semantic role played by logical vocabulary generally: not just conditionals, but also negation (which makes explicit a central feature of our practice of treating claims as materially incompatible), and even modal vocabulary (which makes explicit a central feature of our practice of associating ranges of counterfactual robustness with material inferences). In my initial characterization of the classical semantic project of philosophical analysis, I pointed to the special status that is accorded to logi-
Towards an Analytic Pragmatism
45
cal vocabulary in that project. What I called “semantic logicism” is its commitment to the strategy of using logical vocabulary to articulate the semantic relations between vocabularies that is its goal—paradigmatically in connection with the core projects of empiricism, naturalism, and functionalism. One interesting way to vindicate that commitment (that is, at once to explain and to justify it) would be to appeal to the fact that logical vocabulary is elaborated from and explicating of every autonomously deployable vocabulary whatsoever. For that means that the capacity to use logical vocabulary is both in this very clear and specific sense implicit in the capacity to use any vocabulary, and has the expressive function of making explicit something already present in the use of any vocabulary. I won’t say anything more here about how such a vindication might proceed, contenting myself with the observation that insofar as there is anything to an account along these lines, supplementing the traditional philosophical analytical concern with relations between the meanings expressed by different kinds of vocabulary by worrying also about the relations between those meanings and the use of those vocabularies in virtue of which they express those meanings is not so much extending the classical project of analysis as unpacking it, to reveal explicitly a pragmatic structure that turns out to have been implicit in the semantic project all along. For the conclusion will be that it is because some vocabularies are universal pragmatically elaborated and explicitating vocabularies that semantic analysis of the logicist sort is possible at all. I don’t claim to have entitled myself to that conclusion here, only to have introduced some conceptual machinery that might make it possible to do so—and so at least to have sketched a way in which the insights of the pragmatist tradition can be assembled and developed so as to be constructively helpful to, rather than destructively critical of, the classical project of philosophical semantic analysis, and so to open the way to extending that project in promising new directions.
REFERENCES Brandom, Robert 2002. “Pragmatics and Pragmatisms”, in: Hilary Putnam: Pragmatism and Realism, ed. By James Conant and Urszula M. Zeglen, London: Routledge, pp. 40-58.
46
Robert Brandom
Translated as “Pragmatik und Pragmatismus“, in: Die Renaissance des Pragmatismus, ed. by M. Sandbothe, Velbrück Wissenschaft, 2000, pp. 29-58. Price, Huw 2004. “Naturalism without Representationalism”, in: Naturalism in Question, ed. by Mario de Caro and David Macarthur, Cambridge, Mass.: Harvard University Press, pp. 71-90. Sellars, Wilfrid 1997. Empiricism and the Philosophy of Mind, Cambridge, Mass.: Harvard University Press. Sellars, Wilfrid 1963. “Phenomenalism”, in: Science, Perception, and Reality, London: Routledge Kegan Paul. Sudkamp, Thomas A. 1998. Languages and Machines 2nd edition, Addison Wesley Longman. Wilson, Mark 2006. Wandering Significance, Cambridge, Mass.: Harvard University Press.
BRANDOM ON KNOWLEDGE AND ENTITLEMENT
Sebastian Schmoranzer, Ansgar Seide
Abstract: Robert Brandom spends a significant part of Making It Explicit discussing our practice of knowledge ascription and the nature of epistemic entitlement. We are interested in finding out what concept of empirical knowledge Brandom is implicitly or explicitly defending. We think that Brandom’s remarks commit him to the thesis that knowledge is justified true belief and that the only kind of empirical justification there is is externalist. Having defended our claim we will finally come to the conclusion that Brandom has no answer to second-order skepticism and that there is no difference between default entitlement and all things considered entitlement. Keywords: knowledge, externalism, entitlement, default and challenge, inferentialism, skepticism
Making It Explicit is a book on semantics. Nonetheless, Robert Brandom spends a significant part of it discussing our practice of knowledge ascription and the nature of epistemic entitlement—the kind of entitlement that is necessary for knowledge. Therefore, we are interested in finding out what concept of empirical knowledge Brandom is implicitly or explicitly defending. We think that Brandom’s remarks commit him to the thesis that knowledge is justified true belief and that the only kind of empirical justification there is is externalist. In order to show this we will proceed in two steps: Firstly, we will argue that Brandom holds a traditional view on knowledge as justified true belief. Secondly, we will try to show that there is no room for internal justification in Brandom’s conception of epistemic entitlement. Having defended our claim, we will finally come to the conclusion that Brandom has no answer to second-order scepticism and that there is no difference between default entitlement and all things considered entitlement.
48
Sebastian Schmoranzer, Ansgar Seide
I On reading Brandom’s epistemological remarks in Making it Explicit, one quickly realizes that he is not so much talking about knowledge as about our practices of knowledge ascription. His main interest consists in answering the question of what we are doing whenever we attribute knowledge to someone. Trying to replace the usual semantic vocabulary, Brandom describes the case of knowledge ascriptions as follows: If I say that Jim knows that vultures are scavengers I first of all treat Jim as committing himself to that claim and thereby make him responsible for it. Secondly, I treat Jim as being entitled to it. This means that I accord his claim a certain authority. And finally, I agree with Jim’s judgment and undertake a corresponding commitment myself, which means that I myself believe that vultures are scavengers. So far, this way of telling the story seems to be compatible with the traditional conception of knowledge. Brandom’s observations could be interpreted as follows: Jim knows that vultures are scavengers if and only if my ascription was correct in the following sense: Jim actually commits himself to that claim, he is in fact entitled to it and my own commitment is correct as well, i.e. vultures are in fact scavengers. Retranslating this into the usual vocabulary we have a traditional definition of knowledge as justified true belief. Nonetheless, one might wonder whether Brandom is telling an old story in new words, or whether he has something different in mind. Maybe Brandom is not only translating the old definition into his vocabulary but trying to change it by first talking about ascriptions of knowledge in pragmatical terms and then defining knowledge in terms of knowledge ascriptions. However, this would lead to an implausible position. For in that case Brandom would have to defend something like the following definition: A person S knows something if and only if someone attributes knowledge to her. But attributions of knowledge are neither sufficient nor necessary for knowledge. Knowledge is factive. Jim cannot know that polar bears are brown if they are not. Therefore, even if someone attributes knowledge to Jim, shares his view on polar bears and thinks that Jim is right in committing himself to that claim, Jim still does not know that polar bears are brown.
Brandom on Knowledge and Entitlement
49
Modifying the definition by saying that Jim knows that p if all or all relevant members of the speech community treat him as knowing that p is implausible as well. As Brandom readily agrees, we could all think that polar bears are brown and still be wrong (MIE Ch. 8). Therefore, even if all of us accord knowledge to Jim he still does not know that polar bears are brown. Maybe it is at least necessary that someone attribute knowledge to the epistemic subject for her to have knowledge. At least, Brandom sometimes seems to defend this view: “Just as the truth condition on knowledge requires that the attributor of knowledge undertake, as well as attribute, commitment to the content of the knowledge claim, so satisfying the entitlement condition by mere reliability requires that the attributor of knowledge undertake (but not necessarily attribute) commitment to the propriety of the reliability inference.” (MIE 221, emphasis added) Taken literally, that cannot be quite right. Poor Jim is lost in the desert and sees vultures flying around. He believes that there are vultures and is justified in doing so because he can see them and his judgment is in fact reliable. But Jim himself is not quite sure whether his judgment is to be trusted given the unfavorable conditions. In this case no one—not even Jim himself— attributes knowledge to him. Still, Jim knows that there are vultures. He would be completely justified in directing his actions on the basis of the corresponding belief. Now, that knowledge does not presuppose knowledge attributions can be obscured by the view that knowledge presupposes reliability and that reliability in turn is a context-sensitive concept. In Goldman’s famous barn façade example (Goldman 1976) someone— call him Peter—is in barn-façade-county. There are 99 barn façades and one real barn. Looking at the real barn, Peter comes to believe that there is a barn. Is his belief the result of a reliable belief-forming mechanism? Asked this way, the question cannot be answered because there is no reliability simpliciter. If we define reliability along the success rate which Peter’s belief-forming method would have in the whole country with 10.000 real barns and 99 trompes l’œil, it is reliable. If, instead, we define it along the success rate
50
Sebastian Schmoranzer, Ansgar Seide
Peter’s method would have in barn-façade-county, it is not reliable. Reliability is always relative to a given reference class. But it would be a mistake to think that there is only one legitimate reference class relative to which reliability is to be defined. Which reference class is to be taken into account seems to depend on the interests and purposes of those who evaluate Peter’s judgement. For this reason, reliability and thereby knowledge seems to be attributorrelative. We have to be careful here, however. The foregoing considerations do not imply that someone actually has to attribute knowledge to the epistemic subject for the subject to have knowledge. In our example, Jim knows that there are vultures relative to ordinary standards of reliability, even if no one in fact attributes knowledge to him. We can therefore conclude that Brandom is best interpreted as holding a traditional view on knowledge.
II The question now is what kind of epistemic entitlement respectively what kind of epistemic justification is available to us. One has to distinguish two kinds of epistemic justification: external justification and internal justification. A belief is epistemically justified if and only if someone is justified in accepting the belief relative to the maxim of increasing the number of true beliefs and keeping the number of wrong beliefs as low as possible. A belief that p is internally justified for someone if and only if there are accessible epistemic reasons for the subject that justify the subject’s belief. A belief that p is externally justified if and only if the circumstances of belief-acquisition are favorable to forming true beliefs of that kind, regardless of whether the subject is aware of it. We think that Brandom is committed to pure externalism in the sense that if a belief is epistemically justified, it is non-internally so—solely externally, that is. Why is Brandom committed to this view? Here is our argument: (1) According to Brandom, all entitlement of empirical claims is based on beliefs that qualify as entitled by default.
Brandom on Knowledge and Entitlement
51
(2) Beliefs that are epistemically entitled by default are only externally justified. (3) If all entitlement is based on solely externally justified beliefs, all beliefs are only externally justified. ________________________________________________________ All our epistemically justified empirical beliefs are externally justified. (1) According to Brandom, there are three ways in which one can be entitled to a knowledge claim: one can be entitled inferentially, deferentially or by default. One can justify the claim by giving reasons for it. Secondly, one can appeal to the authority of another asserter: By referring to someone else’s testimony, I “pick up the justificatory check” (MIE 175) of the original asserter. In those cases, however, a regress lurks: Giving reasons for a claim consists in making other claims that also stand in need of justification. Appealing to the authority of another asserter merely postpones the question because the authority of the original asserter can in turn be called into question. To face this well-known regress problem, Brandom presents a model he calls the default and challenge structure of entitlement (MIE 177). The underlying idea is that the regress problem arises from the questionable premise that entitlement can only be attributed if it has been demonstrated. When we realize that entitlement is a status we can have by default, i.e. that we can be entitled to certain claims without having demonstrated our entitlement, the regress problem dissolves. Such default entitlements are not immune to criticism and can be brought into question, but challenges have no privileged status: Challenges are themselves treated as assertions, namely as assertions that are incompatible with the challenged claim, so “their entitlement is on the table along with that of what they challenge.” (MIE 178) All entitlement, therefore, eventually goes back to default entitlement. No commitment, either inferentially or deferentially justified, is justified unless it is based on a default entitlement. (2) According to Brandom, there are two sorts of entitlements that qualify as default entitlements. The first sort are certain sentence types that are treated as “free moves” by the members of a speech community, for example “Red is a
52
Sebastian Schmoranzer, Ansgar Seide
color” or “There have been black dogs”. These sentence types qualify as prima facie justified first premises just because they are nearly trivial commonplaces. The second sort is observational or noninferential reports. A noninferential report is treated by an attributor as a claim to which the reporter is entitled in case the attributor takes the reporter to be a reliable observer under the observing conditions (MIE 222). An important question arises concerning the nature of default entitlement. Default entitlements can be understood in two ways: They can be thought of as social statuses, i.e. as statuses attributed by community members to other members of the community. The other possibility is to understand them as an entitlement someone has because she in fact arrived at her belief in a way that confers a positive epistemic status on her belief, for example, through a reliable belief forming mechanism. With this question in mind we can now take a look at the two sorts of default entitlement. Brandom does not say much about “free moves”, but it seems that he takes them to be governed by the community. After listing a few examples, Brandom tells us: “These are treated as “free moves” by the members of our speech community—they are available to just about anyone any time to use as premises, to assert unchallenged.” (MIE 222) So it seems that “free moves” are justified because they are accepted in my speech community. The important point now is that there seems to be no reason to count this community-governed sort of entitlement as a form of epistemic entitlement: The fact, by itself, that I can assert some sentences without being challenged by members of my community neither gives me a hint to the truth of my claim nor does it follow that I formed my belief in a reliable way. At this point one might object: “If everyone agrees with me that p isn’t it very unlikely that they are all wrong? And if so, do I not thereby have a reason to accept that p?” Two questions are worth distinguishing here: Firstly, is it a reliable beliefforming method to accept what all accept because they accept it? Secondly, do I have a reason to accept what all accept just because they accept it?
Brandom on Knowledge and Entitlement
53
As to the first question, it should be observed that there is no definite answer. If we were all hooked off from reality, it would be a rather unreliable way to form one’s beliefs. As to the second question, one has to admit that I have no reason to follow the other persons’ opinions unless I have a reason to believe that we are not all wrong. The question concerning the nature of the default entitlement of noninferential reports cannot be answered as easily. Does their status stem from the fact that they are treated as reliable, or from the fact that they are reliable? Brandom is not clear on this issue. On the one hand he writes: “Noninferential reports can function as unjustified justifiers: claimings that are treated as having a defeasible default status.” (MIE 222) Brandom seems to suggest that we can base our justifications on beliefs that are accepted as unjustified justifiers by our speech community. But, on the other hand, only a few sentences later he adds: “Noninferential reports […] have their default entitlement status as a result of the way in which the report tokening […] is elicited through the exercise of a reliable differential responsive reporting disposition.” (MIE 222) This seems to suggest that justified noninferential reports are entitled by default because the reporter is in fact reliable. Let us examine both possibilities in turn. If I am allowed to base my claims on noninferential beliefs because they are accepted as unjustified justifiers by the members of my speech community, again the question arises why this form of entitlement-conferring practice can be seen as constituting epistemic entitlement. That I am treated as a reliable reporter gives me no epistemic right to believe something. If, on the other hand, a noninferential reporter is entitled to his claim by default because he is in fact a reliable reporter concerning the kind of matters reported, the default entitlement of my noninferential report is externalist.
54
Sebastian Schmoranzer, Ansgar Seide
Therefore, external justification is the only kind of epistemic default justification there is. (3) What consequences does this have for the beliefs that are deferentially or inferentially justified? These are cases in which entitlement can be inherited but not produced (MIE 176). This can easily be explained when one considers the case of inferential justification. A belief is inferentially justified if it is defended on the basis of other justified claims standing in an inferential relation to the belief in question. The justification for the premises is thereby conferred on the conclusion. But one can only confer what is already there, as it were. If one of the premises is externally justified so is the conclusion. The reason to hold the conclusion is not accessible to the epistemic subject if the reason for one of the premises is inaccessible. Consider the following case: In his youth, Jim was trained to identify vultures. Unfortunately, he forgot about that. He has no idea that he is a reliable reporter of vultures. He sees one and claims: “That is a scavenger.” Asked how he knows this he answers: “That is a vulture. Vultures are scavengers. Therefore, that is a scavenger.” His belief is inferentially justified. The conclusion is correctly inferred from justified premises. But it is not internally justified since he has no idea why it should be true. And this is so because he neither has a reason to consider himself a reliable reporter of vultures nor can he refer to distinguishing marks. This observation should give us a pause when we read that Brandom wants to defend a middle position between justificatory internalism and reliabilist externalism (MIE 217). In “Insights and Blindspots of Reliabilism” (AR Ch. 3) he holds the view that sometimes an epistemic subject can be justified without being able to give reasons for her belief as long as she is in fact a reliable reporter. This is what Brandom calls the “Founding Insight of Reliabilism”. Against a radical form of reliabilism Brandom argues convincingly that notwithstanding this insight there are no noninferentially justified beliefs without there being inferentially justified beliefs. This is what he describes as the “Conceptual Blindspot of Reliabilism”. Brandom’s argument is elegantly brief: In order to hold a belief one has to understand the proposition be-
Brandom on Knowledge and Entitlement
55
lieved. This in turn requires linguistic competence. One does not understand the term “polar bear” if one does not know that polar bears are white furry animals. Therefore, if one is noninferentially justified in believing that there is a polar bear on the hill one has to know in addition that polar bears are white furry animals. And both claims together—given that justification is closed under known entailment—inferentially justify the belief that there is a white furry animal (AR 97-110). We agree with Brandom on this point. But that does not show that he is holding a middle position between externalism and internalism. One would have to presuppose that inferentially justified beliefs are thereby internally justified. That this is wrong has been shown above. All Brandom has shown is that he holds an intermediate position between epistemic inferentialism and epistemic noninferentialism. This is compatible with our conclusion that all beliefs—if justified—are only externally justified.
III What if we are right and the only plausible conception of epistemic entitlement to be found in Brandom is purely externalist? One worry that immediately comes to mind is second-order scepticism. In a way pure externalists have an easy answer to the challenge. According to them, knowledge only presupposes external justification. Now, neither the skeptic nor we can tell whether we are reliable reporters of truth. Therefore the skeptical thesis that we do not have knowledge because we are never externally justified cannot be defended. Unfortunately, the thesis that we do have knowledge cannot be defended either because we do not have any internal reason to believe that we are reliable reporters of truth and therefore qualify as knowers. So scepticism triumphs at second order. Another consequence would be that the concept of default entitlement is of no use in Brandom’s epistemology.1 The idea is that a belief is justified by default if it is prima facie justified but that it can lose that status and not be justified all things considered. This means that after reasonable doubts are on 1
Although it might still be useful for his semantics.
56
Sebastian Schmoranzer, Ansgar Seide
the table the belief in question loses its positive justificatory status unless it can be defended by further arguments. But if all justified beliefs are externally justified that can never happen. Let us suppose that I hold the noninferentially justified belief that there is a vulture sitting in the tree. If it is justified, it is so in virtue of the reliability of my belief-forming mechanism. But this would still be the case after a usually trustworthy person challenges my claim by saying that I am not to be trusted on that matter. If my belief was reliable before, it will remain so thereafter. And my reply “Well, I was raised with vultures” would not change anything either because the positive justificatory status of my original belief consists solely in its being reliable. The distinction between default entitlement and all things considered justification is an empty one once justification is a purely external matter.
IV We have tried to show that Brandom is best interpreted as holding a traditional view on knowledge that should be qualified as externalist because the only notion of plausibly maintained epistemic entitlement is that of external justification. Unfortunately, a solely externalist conception of justification makes the dialectical tool of default entitlement epistemically useless and it leaves us with the problem of second-order scepticism. So Brandom ends up defending a well known epistemological position facing the same old and notorious problems. However, his new perspective on the old stories provokes fruitful rethinking of the relation between knowledge and knowledge ascription or between inferentialism and noninferentialism on the one hand and internalism and externalism on the other. Furthermore, it is worth thinking about adapting his default and challenge structure of entitlement within an internalist framework. Therefore, even if Making It Explicit is, first of all, a book on semantics, it is a valuable contribution to epistemology as well.
Brandom on Knowledge and Entitlement
57
REFERENCES Brandom, Robert 1994. Making It Explicit: Reasoning, Representing, and Discursive Commitment, Cambridge, Mass.: Harvard University Press, (MIE). Brandom, Robert 2000. Articulating Reasons: An Introduction to Inferentialism, Cambridge, Mass.: Harvard University Press, (AR). Goldman, Alvin 1976. „Discrimination and Perceptual Knowledge“, in: Journal of Philosophy 73, pp. 771-791.
OF μ-MESONS AND ORANGES SCRUTINIZING BRANDOM’S CONCEPT OF OBSERVABILITY
Jochen Apel, Simone Bahrenberg, Carolin Köhne, Bernd Prien, Christian Suhm
Abstract: The main concern of our paper is to scrutinize Brandom’s concepts of observation and observability. We provide two arguments in order to show that, contrary to Brandom’s view, there is an epistemically significant difference between observable and unobservable entities. The first argument hinges on the fact that scientific and common sense terms differ in regard to the continuity of ontological tenets related to them. Our second argument throws light on the fact that not all observation sentences can function as regressstoppers in discussion. Consequently, there seems to be a remarkable difference between scientific and everyday observations with respect to justification. Keywords: inferentialism, observation, observability, theoretical entities
INTRODUCTION In our paper we will be concerned with Brandom’s concept of observability which he introduced in his book Making It Explicit as well as in a couple of papers. In section 1 we will have a look at some basic prerequisites of his concept of observation reports and examine the concept itself. Section 2 is devoted to an analysis of the adequacy of Brandom’s account of scientific observation for scientific practice. We will acknowledge that Brandom’s notion of observability is probably adequate to scientific parlance, but will object that it does not capture what in fact is involved in scientific practice, especially in scientific observation. In sections 3 and 4 we will provide two arguments for the inappropriateness of Brandom’s concept of observability. We will argue, in particular, that it does not help to explain two striking differences between
60
Apel, Bahrenberg, Köhne, Prien, Suhm
theoretical terms on the one side and common sense terms on the other in regard to their respective historical continuity and stability. In our opinion, the concept of observability, as Brandom sees it, suffers from a lack of explanatory power, a power which opposing approaches to observation and observability, notably those which allow for an epistemically relevant difference between observable and unobservable entities, do have.
1
BRANDOM ON OBSERVATION AND OBSERVABILITY
Brandom is a proponent of inferentialism. According to this position, an assertion owes its specific content to its inferential correlations with other assertions. To Brandom’s mind, a necessary, though not sufficient, condition for the use of concepts is to have the capacity to respond differentially to distinct environing stimuli.1 Before people are able to make observation reports they must have acquired a reliable differential responsive disposition. We, as genuine knowers, share this capacity with artifacts and measuring instruments. The difference between possessors of genuine observational knowledge and merely reliably differentiating responders is based on the fact that candidates of observational knowledge are disposed to do so by applying concepts. This means that some sort of understanding is involved in observation in case a reporter’s response to a stimulus amounts to making a move in the game of giving and asking for reasons (cf. MIE 88). The latter, in turn, involves knowing the inferential correlations between the uttered assertion and other assertions (the claim that something is red, for example, entails that it is colored and is incompatible with the claim that it is green).2 Why, however, is it important for Brandom to be concerned with observation reports? He characterizes discursive practices as being implicitly normative and inferentially articulated, since they entail certain structures of commitment and entitlement. What we do when we make an assertion or claim, is to take the responsibility to justify it if it is challenged, and to be prepared to justify other claims (cf. MIE 167f). At this point, observation reports 1 2
Cf. MIE 87; TMD 349; NIK 96. Cf. MIE 88f; TMD 349ff; NIK 96.
Of μ-Mesons and Oranges
61
come into play as regressstoppers, because they can function as unjustified justifiers (cf. MIE 222). The significance of observational knowledge is best understood if one considers its role in the default-and-challenge structure of entitlement. Whereas inferential justifications always lead to further assertions and thereby evoke a justificatory regress, observation reports rejoice in being treated as claims which are, though defeasible, noninferential and justified by default. Obviously, a philosophically interesting regress problem only occurs if one presupposes that being entitled to an assertion implies being able to demonstrate the entitlement. In our everyday language, however, there are plenty of assertions such as ‘Red is a color’ to which we are entitled by default, i.e. without giving reasons (cf. MIE 222). According to Brandom, observation reports likewise fall in the category of assertions justified by default. We now want to take a closer look at Brandom’s concept of observability. In accordance with Sellars, Brandom believes that theoretical and observable entities do not differ in kind, but only in how we come to know about them. Accordingly, we have only inferential cognitive access to theoretical entities, whereas non-theoretical entities can be observed noninferentially (cf. TMD 362). It is important to distinguish two senses of observability that are involved in Brandom’s account. In a narrow sense, only such entities are observable that are noninferentially reportable in a particular historical and scientific context. According to this sense, superstrings are unobservable since today we do not dispose of suitable experimental devices for the detection of eleven-dimensional objects, let alone physicists properly trained for their observation. In a broad or fundamental sense, however, all entities are observable, since it is possible for the line between theoretical and observable objects to be shifted—for example by developing new measuring instruments. Brandom’s argument suggests that nothing is unobservable in principle and, consequently, no epistemically insurmountable gap between the observable and unobservable parts of physical reality exists. “[T]here is no particular line to be drawn between what is in principle observable and what is not. The only constraints are what a reporter can be trained under some circumstances reliably to differentiate […].” (NIK 96)
62
Apel, Bahrenberg, Köhne, Prien, Suhm
In consequence of his concept of observability, Brandom maintains that “[…] a properly trained physicist, who can respond systematically differently to differently shaped tracks in a cloud chamber will, if responding by non-inferentially reporting the presence of mu mesons, count as genuinely observing those subatomic particles. The physicist may start out by reporting the presence of hooked vapor trails and inferring the presence of mu mesons, but if the physicist then learns to eliminate the intermediate response and respond directly to the trails by reporting mesons, the physicist will be observing them.” (NIK 96) Although the observation of μ-mesons clearly falls within a restricted area of highly specialized experimental practice, according to Brandom it does not differ in a fundamental manner from the observation of everyday middlesized objects. “[T]he community for whom ‘mu meson’ is an observation predicate will be much smaller and more highly specialized than the community for whom ‘red’ is one. But these are differences of degree, rather than kind.” (NIK 97) Finally, Brandom argues that the fact that physicists can retreat to inferences in order to justify the upshot of scientifically ambitious observations, as in the case of μ-mesons, should not seduce us to conclude that initial observation reports on the existence of μ-mesons themselves result from inferences (cf. NIK 97).
2
WHAT SCIENTISTS SAY AND WHAT THEY DO
In our view, Brandom’s concept of observability is implausible in the light of scientists’ experimental practice. We do not think that theoretical entities are observable like, for example, oranges. Before we go into the details of our arguments in the next two sections, we will anticipate a possible response to them that is related to the scientific parlance. Though we acknowledge that
Of μ-Mesons and Oranges
63
Brandom’s notion of observability is adequate to scientific parlance, we doubt whether it adequately captures experimental practices and their role in the overall scientific endeavor. In our opinion, the expression ‘I am observing a μ-meson’—which physicists possibly utter when looking at specifically hooked vapor trails—is to be interpreted as an abbreviatory parlance. Physicists do not observe any elementary particles, though they might say so. In fact, they infer from the observation of specifically hooked vapor trails to the presence of, say, a μmeson. The only difference between a well-trained μ-meson physicist and his non-trained colleague consists in different parlances. The experienced physicist is, in fact, not trained to the observation of μ-mesons, but rather to the uttering of ‘There’s a μ-meson’ should she observe suitable cloud chamber trails. If she is forced to justify her belief in the presence of a μ-meson, she will retreat to inferences, probably concerning the functioning of the measuring devices, in exactly the same way as her non-trained colleague does in that situation. She will not, that is to say, simply respond “But look, there really is a μ-meson in the cloud chamber! You just have to learn to observe it.” as she would do, if she were asked to justify her belief that an orange is lying on the table. A similar case occurs when scientists utter that they are observing a theoretical measure as, for example, the melting point of a metal. Strictly speaking, they are not observing the melting point in question, but rather, concluding it from sufficiently rich data by an inductive inference. In a nutshell, they will ascertain the sought-after measure by calculating an average temperature from the experimentally determined melting temperatures of various substances of the same metal.3 Obviously, this implies an inference, namely from a multitude of experimental data to a general claim about the melting point of a particular chemical element. In light of this example, it seems again plausible to presume that theoretical entities are unobservable and that scientists’ parlance as to the contrary is to be seen as a shortcut for the inferences they require in order to justify their belief in the existence of unobservables. At any rate, we maintain that there is an epistemically relevant difference between observable and unobservable entities. The difference does not, con3
Cf. Bogen and Woodward 1988, 308f.
64
Apel, Bahrenberg, Köhne, Prien, Suhm
trary to what Brandom claims, hinge on the methodological or practical issues of developing suitable measuring instruments and of training the appropriate observation techniques of scientists. It is, rather, based on the very nature of physical objects and their causal interactions with us.
3
THE ARGUMENT FROM THE HISTORY OF SCIENCE
We will now try to develop two arguments for the thesis that there is an epistemically significant difference between observable and unobservable entities. The first one is a remote relative of Larry Laudan’s famous pessimistic metainduction which he put forward in the debate on scientific realism (cf. Laudan 1996). Similar to Laudan’s anti-realistic challenge, our first argument hinges on the striking evidence from the history of science for the ubiquity of the thoroughgoing and conceptually momentous revolutions that led to radical revisions of approved scientific ontologies. In particular, theoretical entities like phlogiston or the electromagnetic ether, once fully accepted members of the scientific world picture, are nowadays seen as chimeras that plainly do not exist.4 In contrast to theoretical entities in science, however, middle-sized objects around us in our daily life do not seem to undergo such dramatic changes. We still believe in the existence of tables, human beings, clouds and trees, just as our Medieval or Greek ancestors did. Though there may be cases when our common sense theories shift in relevant aspects so that we rethink the ontological furniture of the observable reality (think, for example, of witches and witchcraft), it seems obvious that there is a remarkable difference between science and common sense with respect to the stability of basic ontological tenets. Whereas, in the history of science, theoretical shifts and revolutions seemed to be on the agenda nearly all the way, drastic changes usually do not occur in everyday contexts. 4
Brandom’s favorite example of an observable elementary particle, the μ-meson, is, ironically, also a good example of some sort of ontological shift in particle physics, since some time after its detection it turned out to be a lepton (called muon) rather than a meson.
Of μ-Mesons and Oranges
65
To our minds, the difference between science and common sense as to the longevity of their respective ontological commitments is of great importance and cannot be dismissed lightly in the contexts of observation and observability. We are in need of an explanation of the fact that putative observations in science very often go astray, whereas their common sense siblings usually rejoice in long-term success. If confronted with this explanandum, however, Brandom’s concept of observability fails to provide an explanans. If one holds that there is nothing other than a methodological difference between scientific and everyday observations related to the respective demands for an adequate training of observers, no theoretical means seems to be at hand. In Brandom’s view, every object is observable in principle, i.e. observable if appropriate observation instruments are available and provided that the relevant sort of observation training is done. Consequently, his approach leaves open the question of why many scientific posits are epistemically peculiar, namely insofar as they are unobservable objects susceptible to theoretical revision. In contrast to Brandom, empiricists do have the theoretical resource to explain the asserted difference between science and common sense. According to Bas van Fraassen’s constructive empiricism, for example, the observability of an entity is a function of our physiological makeup and is, therefore, not at all dependent on and alterable by technical devices or some sort of observation training (cf. van Fraassen 1980, 4-19). Though we do not want to argue for van Fraassen’s position, it strikes us that Brandom’s concept of observability, if compared with its empiricist rivals, comes off badly with respect to Laudan-style records of scientific and common sense histories.
4
THE ARGUMENT FROM INFERENTIAL PRACTICE
Our second argument takes its cue from Brandom’s distinction between inferential and noninferential justifications. Whereas many sentences are justifiable only by other sentences, i.e. inferentially, observation sentences are noninferentially justified by observations and therefore function typically as regressstoppers in discussion. If, however, one considers the case when a belief
66
Apel, Bahrenberg, Köhne, Prien, Suhm
expressed in an observation sentence is challenged, a further noteworthy difference between science and common sense gets obvious. On the one hand, beliefs based on everyday observations—think of the visual perception of a ripe orange—are not usually justified by further beliefs, in case they are challenged, but by setting the challenger’s mind to the observed entity and by asking more or less rhetorical questions like “But look, don’t you see the orange on the table?”. On the other hand, beliefs about theoretical entities such as μ-mesons are typically not defended by pointing to cloud chambers in the particle physicists’ laboratories, but by giving reasons for believing in the existence of the posited entities. If a physicist claims that she has observed a μ-meson, she will, if her claim is doubted, refer to the proper functioning of the measuring apparatuses and to the theories which describe the causal processes of the observation rather than draw the attention of her discussion partner to the specific trace in the cloud chamber—even if her claim about the observation of a μ-meson is challenged by another physicist who is well acquainted with traces of μ-mesons. The distinction between inferential and noninferential justifications with respect to scientific and everyday contexts obtrudes even more if cases of conflict between the different sorts of justification are taken into account. Think, for example, of a blind test of fruits and their colors. If you taste the flavor of a ripe orange without seeing it and infer that you are in contact with an orange-colored fruit, but what you see after opening your eyes is a purple fruitlike object (cf. MIE 225), your noninferential observation trumps your inference—you will be sure that you are aware of a purple thing. In sharp contrast to this situation, the inferential reasoning will outweigh the noninferential observation in an analogous laboratory-case. If a scientist claims to have observed a μ-meson while her colleague infers to the contrary by reason of inappropriate measuring conditions or a malfunctioning of technical devices involved in the measuring process, there will be no doubt about the respective epistemic merits of the claims. Obviously, we are again faced with a remarkable difference between scientific and everyday observations. It seems inevitable from the above considerations that scientific observation sentences mostly and typically do not serve as regressstoppers. They are, if challenged, in need of inferential justification just as many other non-observational sentences are, namely by referring to the
Of μ-Mesons and Oranges
67
adequateness and proper functioning of the observation means. In contrast, beliefs based on everyday observations do serve as regressstoppers in discussion. The belief that an orange is lying on the table is not usually expected to be justified by referring to the conditions of visual sight or some piece of optical theory. Similar to the conclusion of our first argument, we get the result that there is an epistemically relevant difference between scientific and everyday observations or, more precisely, between scientific and everyday observation sentences with respect to their role in contexts of justification. And once again it seems doubtful whether Brandom’s concept of observability provides a plausible explanation of this fact.
5
CONCLUSION
In the light of our two arguments, it seems questionable whether, in principle, there is no epistemically relevant distinction between observable and unobservable entities, as it follows from Brandom’s concept of observability. First, studies from the history of science indicate that theoretical concepts are unstable and error-prone, especially if compared with observational common sense concepts. Second, with regard to justification and the foundation of beliefs, noninferential observations do not have the same epistemic status in scientific as in common sense contexts. Both arguments pull in the same direction. In consequence, there seems to be a remarkable epistemic difference between the observable and the unobservable parts of physical reality which is not captured in Brandom’s account of observation. Unlike some of its philosophical rivals, it suffers from a lack of explanatory power as to two basic differences between observation in science and in everyday life. If one takes a liking to inferences to the best explanation, as Brandom probably does, one could finally conclude that both differences pivotal to the arguments presented in the preceding sections are best explained by an epistemically relevant distinction between observable and unobservable entities, the former accessible to noninferential observation, the latter only ascertainable by inferences. If Brandom is unwilling to accept this consequence he seems to owe us an even better explanation.
68
Apel, Bahrenberg, Köhne, Prien, Suhm
REFERENCES Bogen, James and Woodward, James 1998. „Saving the phenomena“, in: The Philosophical Review 97, pp. 303-352. Brandom, Robert 1994. Making It Explicit: Reasoning, Representing, and Discursive Commitment, Cambridge, Mass.: Harvard University Press, (MIE). Brandom, Robert 2002. Tales of the Mighty Dead: Historical Essays in the Metaphysics of Intentionality, Cambridge, Mass.: Harvard University Press, (TMD). Brandom, Robert 2002. “Non-Inferential Knowledge, Perceptual Experience, and Secondary Qualities. Placing McDowell’s Empiricism”, in: Reading McDowell. On Mind and World, ed. by Nicholas H. Smith, London: Routledge, pp. 92-105, (NIK). Laudan, Larry 1996. “A Confutation of Convergent Realism”, in: The Philosophy of Science, ed. by David Papineau, Oxford: Oxford University Press, pp. 107-138. van Fraassen, Bas C. 1980. The Scientific Image, Oxford: Clarendon Press.
SUCCESSFUL ACTION AND TRUE BELIEFS Nikola Kompa, Rudolf Owen Müllan, Bernd Prien, David P. Schweikard
Abstract: We discuss the classical pragmatist thesis that truth and success are connected. Whyte has proposed a sophisticated version of this thesis, called success semantics, according to which an assertion is true if all of the actions that “follow” from an assertion via practical syllogisms succeed. Against this, Brandom has argued that truth and success need not coincide because practical inferences are non-monotonous. We argue that Whyte’s success semantics can be modified so as to avoid Brandom’s objections. The significance of success semantics stems from the fact that it offers a pragmatist conception of the relation between our discursive practices and the world. This conception could serve as an alternative to Brandom’s, which relies on the process of rational rectification and in which practical inferences play no role. Keywords: truth, success, practical syllogism, semantics
In this paper, we want to defend the classical pragmatist idea that we can determine the truth of our beliefs by looking at their role as bases for successful action.1 In his 1990-paper “Success Semantics”, Jamie Whyte defended a sophisticated version of this idea, to which Robert Brandom has raised a couple of objections. Whyte suggests the following principle concerning the truthconditions of our beliefs: From a belief together with a desire (i.e. one type of pro-attitude) one can infer an intention to perform an action. So far, this is just a version of the classical practical syllogism. Whyte’s principle now maintains that the belief B is true if and only if the following holds: If I take a desire D, from D and B practically infer the intention to do A, then the actual performance of A will lead to the satisfaction of D. Drawing on Frank Ramsey he formulates it as follows: 1
By calling it the ‘classical pragmatist idea’ we do not mean to insinuate a specific historical claim with reference to any particular author. Rather, we take the idea we are working with to be common to a wide variety of pragmatist views of semantics.
70
Kompa, Müllan, Prien, Schweikard
“A belief’s truth condition is that which guarantees the fulfilment of any desire by the action which that belief and desire would combine to cause.” (Whyte 1990, 150) Now, in most cases, many beliefs figure as premises in practical reasoning. This means that in the first instance, Whyte’s principle yields only truthconditions for conjunctions of beliefs. However, as Brandom acknowledges, this problem can be overcome by a process of factoring out: If I, for instance, have a desire that the room be illuminated and further believe B1) that the electricity is working and B2) that there is a bulb in the lamp, I can (normally) assume the beliefs B1) and B2) to be the guarantors of the success of my turning on the light. Should the action not fulfill my desire, one of these beliefs must be false. I can test my belief B1 that the electricity is working by using it as a premise in other practical syllogisms. For example, I can assess my belief that the electricity is working by conjoining it with the belief that this is a water boiler and the desire for hot water to make tea. This would make it rational to flip the switch of the water boiler. If this action succeeds in getting me hot water, I have pretty good evidence that the electricity is working. I can then conclude that my desire that the room be illuminated was not satisfied by flipping the light switch because my other belief (that there is a bulb in the lamp) is false. So, the central tenet of Whyte’s Success Semantics is that if all of the beliefs figuring in a practical syllogism are true and these beliefs together with a certain desire D make it rational to perform an act A, then A must lead to the satisfaction of D. If it doesn’t, we can conclude that at least one of the beliefs must be false (assuming that the inference was valid).
1
THE SIGNIFICANCE OF THE ISSUE
Before we turn to Brandom’s objections to this thesis, we would like to comment briefly on its significance. In our view, the main idea of Success Semantics is the claim that the contact between our discursive practice and the world is established by the practical syllogism. By understanding action in
Successful Action and True Beliefs
71
terms of the practical syllogism, we transfer the way the world constrains the success of our actions to the way the world constrains the truth of our beliefs. In Brandom’s view, on the other hand, our practices are constrained quite differently by the world. Suppose, for example, that we have a concept ‘acid’ with the following circumstances and consequences of application (cf. MIE 2.V): if a liquid tastes sour, one is entitled to apply the term ‘acid’ to it. And if one is committed to calling something ‘acid’, then one is also committed to its turning litmus paper red. Now, supposing I observe that a certain liquid tastes sour. Then I am entitled to apply the term ‘acid’ to it, which, in turn, commits me to apply ‘turns litmus paper red’. Suppose further that I also observe that the liquid turns litmus paper blue. I am then entitled to incompatible commitments. When this happens, I am obliged to modify my assertional practices in such a way as to avoid this predicament in the future. In this way, the world can constrain our inferential practices. The point of discussion is the following: We will try to defend what we take to be the main idea of Success Semantics, i.e. the thesis that practical inferences play an indispensable role in establishing contact between intentionally acting beings and the world. Brandom, on the other hand, holds that no practical inferences are required for the possibility of ending up with incompatible commitments. All that is needed are non-inferential entitlements and a two-sided semantics. In “Pragmatics and Pragmatisms” Brandom raises two objections to Success Semantics: 1) He notes that the account of truth-conditions formulated by Whyte presupposes the notion of contentful desires. But he doubts that this can be accounted for without falling into the myth of the given (cf. PaP 52). 2) Even if this concern is bracketed, the thesis about truthconditions is not true: Even if all the doxastic premises of a practical inference are true, the action need not lead to the satisfaction of the desire involved (cf. PaP 53f and US). In this paper, we will try to answer only the second objection. We see the dialectical situation as follows: There are two possible accounts of how the world can constrain our practices, Brandom’s theory of incompatibilities and Success Semantics. Brandom argues that the latter account is unworkable. We will here try to counter one of the two objections Brandom has raised and
72
Kompa, Müllan, Prien, Schweikard
thus attempt to take the first step towards showing that Success Semantics is a viable alternative for an account of worldly constraint.
2
BRANDOM’S OBJECTION IN “UNSUCCESSFUL SEMANTICS”
In his article “Unsuccessful Semantics”, Brandom raises the following objection against Success Semantics: Even if a practical inference is valid, and all the doxastic premises in it are true, the inferred act A can still fail to satisfy desire D. Brandom considers the following example: X wants to have cookies, and X believes (truly) that cookies are in the cupboard. X now concludes (validly) that she should go to the cupboard. Brandom says that this action might still fail to satisfy X’s desire for any of infinitely many reasons, for example if wet weather has swelled the doors of the cupboard shut. According to Brandom, X has no belief concerning this last point, i.e. neither believes that the doors are swelled shut, nor that the doors are not swelled shut. Further, Brandom may concede that if X accepts as a further premise ‘The doors are swelled shut’, she could (and would) no longer conclude that she should go to the cupboard. But that does not mean, according to Brandom, that the intention ‘I should go to the cupboard’ only follows if one accepts the premise ‘The doors are not swelled shut’. Rather, the above practical inference is materially good as it stands. It can, however, be defeated by the acceptance of a further premise, such as ‘The doors are swelled shut’. Thus, Brandom’s view of our material practical inferences is that they are non-monotonous.2 Given a valid material practical inference from a desire D and a set of beliefs B there are beliefs C such that B and C together no longer imply the consequence. If practical inferences are in fact non-monotonous, then there are cases in which a speaker/agent accepts only true premises, draws only valid practical inferences and in which the actions thus arrived at fail to satisfy the desire involved. If the inference is non-monotonous, the failure of the action need not be due to the falsity of one of the doxastic premises, but can also be due to the fact that conditions obtain which defeat the inference. 2
He holds the same view about theoretical material inferences.
Successful Action and True Beliefs
3
73
A MODIFIED VERSION OF SUCCESS SEMANTICS
Now the defender of Success Semantics has several options: First, he could try to argue that practical inferences are monotonous after all. This can be done in two ways: Either one argues that the negations of all the assumptions that Brandom classifies as possible defeasors of an inference are required as premises for the inference to go through, or one argues that some sort of ceteris paribus clause is needed as an extra premise. Brandom discusses both of these strategies in “Unsuccessful Semantics” and we agree with him that neither looks very promising. A second option is to modify the central tenet of Success Semantics in such a way as to avoid Brandom’s objections. This is what we will try to do here. We accept Brandom’s claim that material practical inferences are nonmonotonous, as well as Brandom’s claim that we have no beliefs about many background or normality conditions. We also accept Brandom’s conclusion that under these conditions the central tenet of Success Semantics is false. Therefore, we propose to modify this tenet in the following way: Whereas the original version claims that at least one of the premises of the practical inference must be false if the action A does not satisfy the desire D, our modified version claims that the set of sentences in which one has to look for the culprit is larger: If the action A does not satisfy the desire D, then at least one of the premises or one of the normality assumptions of which the agent knows that their negation would defeat his inference must be false. Let us explain this a little: In the example given above, the agent infers from the doxastic premises (‘Cookies are in the cupboard’ and ‘The cupboard is in the kitchen’) and her desire to eat cookies that she should go to the kitchen. This inference is non-monotonous insofar as the following sentences would, if taken as further premises, defeat the inference: “the cupboard having been nailed shut, moved, mined, encased in glass, shrunk to microscopic size, and so on.” (US 177) Now, we can assume that the agent knows that these claims would defeat his inference, or at least we can assume that this is generally known in the linguistic community. Knowing what would defeat an inference is simply part of knowing the contents of the assertions involved. In
74
Kompa, Müllan, Prien, Schweikard
any case, in this way we get a reasonably precise definition of a set of normality conditions known to be relevant, N for short. The set N contains sentences that describe what is normally the case, e.g., that objects do not suddenly shrink to microscopic size. The negations of the sentences in N are defeasors of practical inferences such as the one in our example. The central tenet of this modified version of Success Semantics is then: Suppose the practical inference from a set of beliefs B and a desire D to the intention to do A is materially good and that N is the set of sentences the negations of which are known to defeat this inference. Suppose further that the action A, if carried out, would not lead to the satisfaction of D. Then at least one sentence from the union of B and N must be false. In order to find out which of the sentences in B and N are false, we have to engage in the process of factoring out that Whyte has described in “Success Semantics”. We have to form a hypothesis as to which of the sentences from B and N might be responsible for the failure, combine it with other sentences and desires, practically infer other actions, carry them out and see if they succeed or not. In this process, we treat sentences from N in the same way as we treat sentences from the set of beliefs B. If we for some reason suspect that a normality condition fails to obtain, we can test it in the same way that we can test any other assumption. It does not matter here that we had no beliefs about the normality condition when we drew the inference.
4
CONNECTIONS WE DO NOT KNOW ABOUT
This modification obviously tackles the counterexamples Brandom mentions in “Unsuccessful Semantics”. However, Brandom might try to strengthen his objection to Success Semantics by noting that the above modification only takes care of normality conditions whose negations are known to defeat an inference. But what about normality conditions that are not known to be relevant? Suppose there is a normality condition that the air contains no XYZ. Suppose further that, as a matter of fact, if XYZ is in the air, cookies dissolve, but that this is not known in the community. Now, if XYZ is in fact in the air, we have a case in which the action of going to the kitchen will not satisfy the desire for cookies, even though all of the premises are true and all
Successful Action and True Beliefs
75
of the normality conditions whose negations are known to defeat the inference are in place. Here we have a case that also refutes the modified version of Success Semantics. However, we think that this case can be accounted for by a principle of closure under formal implication. Even if an agent has no beliefs about the normality conditions in N, it can be said that he implicitly accepts them. He implicitly accepts these conditions insofar as one would mention the acceptance of those conditions in a rational reconstruction of his behavior. One would be entitled to attribute these accepted assumptions on the grounds that (i) the agent acts like someone who believes that these conditions obtain, (ii) the agent draws practical inferences which would not be valid if these conditions did not obtain, and (iii) he will say that these conditions obtain upon request. (iv) Furthermore, these assumptions have default entitlement status. Consequently, acceptance of a given assumption is what can rationally be attributed to the agent in question, i.e. it is what he rationally ought to believe, given his other beliefs and his behavior. Take a defeasor p, for example that there is XYZ in the air. Then there is a particular normality condition n ∈N: such that p → ~n. Now we take it that if we implicitly accept the conditions in N, then we can also be said to accept claims that follow from sentences in N. In other words, we take it that acceptance is closed under logical implication: If someone accepts q and q entails r, then he also accepts r. This closure principle would be implausible for belief. But given that acceptance captures what we ought to believe, the closure principle for acceptance enjoys a certain plausibility. Now, note that sentences of the form ~(p & (p→~n)) for any p and any n in N belong to this closure.3 Call these sentences p-sentences. So while we may not accept that ~p, for any defeasor p, we nonetheless accept the respective p-sentence, i.e. ~(p & (p→~n)) (For we can accept the latter without accepting ~p.) Our proposal is to further modify Success Semantics by requiring agents to look for the reason of their action’s failure in this part of N’s closure as well. When an action fails, this can also be due to the fact that one of the psentences is false, i.e. that a sentence of the form p & (p→~n) is true for a particular p and a particular n in N. 3
(p & (p→~n)) implies ~n, so n implies ~(p & (p→~n)).
76
Kompa, Müllan, Prien, Schweikard
Applied to the example considered here, this means the following: When the action fails (i.e. the agent gets no cookies by going to the kitchen) and the agent has ruled out beyond reasonable doubt that one of the premises was false and that one of the background conditions whose negation is known to defeat the inference did not obtain, then he has to consider (among other possibilities) the possibility that the sentence ‘XYZ is in the air and if XYZ is in the air, cookies dissolve’ is true.4 Again, the truth conditions of this conditional can be ascertained in basically the same way as that of any other assertion: We have to experiment with XYZ and cookies and this process of experimentation will itself be guided by the modified version of Success Semantics.
5
CONCLUDING REMARK
One might wonder whether the modifications we have proposed do not disfigure Success Semantics beyond recognition. One thing is clear: Whereas the original version held the promise of a relatively principled path to the truth, this is no longer the case with our modified version. The modified version requires considering those connections between states of affairs of which we are so far utterly ignorant5. It turns out that finding out the truth is much more difficult than Whyte’s Success Semantics might lead one to expect. We think, however, that the central idea of Success Semantics is still preserved in our modified version. As we explained earlier in this paper, we take it to be one of the main ideas of Success Semantics that the ultimate point of contact of our discursive practices with the world around us is established by practical inferences and the success of the actions thus arrived at. The world can register protest against our discursive practices by letting actions fail that should succeed given what we believe. This is to say that the world can ‘force’ 4
5
There is no guarantee that the culprit will be found, though, as there may be no recursive procedure for how to go about updating the system of beliefs and normality assumptions in the light of unforeseen circumstances. We are not in this sense ignorant of the normality conditions in N. We have no beliefs about them, but if the subject somehow came up, we would know what to believe.
Successful Action and True Beliefs
77
us to revise our beliefs. This idea is clearly preserved by our modified version of Success Semantics.
REFERENCES Brandom, Robert 1994. Making It Explicit: Reasoning, Representing, and Discursive Commitment, Cambridge, Mass.: Harvard University Press, (MIE). Brandom, Robert 1994b. “Unsuccessful Semantics”, Analysis 54, pp. 175-178, (US). Brandom, Robert 2002. “Pragmatics and Pragmatisms”, in: Hilary Putnam: Pragmatism and Realism, ed. by James Conant and Urszula M. Zeglen, London: Routledge, pp. 40-59. Whyte, J.T. 1990. “Success Semantics”, Analysis 50, p. 149-157, (PaP). Whyte, J.T. 1991. “The Normal Rewards of Success”, Analysis 51, pp. 65-73. Whyte, J.T. 1997. “Success Again: Replies to Brandom and Godfrey-Smith”, Analysis 57, pp. 84-88.
ARE FUNDAMENTAL DISCURSIVE NORMS OBJECTIVE?
Sebastian Laukötter, Bernd Prien, Till Schepelmann, Christian Thein
Abstract: Robert Brandom’s book Making It Explicit can be read as a theory of social norms. Since our rationality consists in our being engaged in norm-governed discursive practices, social norms play a very central role for us. In our paper, we are going to draw two distinctions in the domain of social norms: First one can distinguish between objective and non-objective norms, and second one can distinguish between those norms in the game of giving and asking for reasons that correspond to inferences, and those norms in that game that do not. Brandom himself stresses the importance of the first distinction, but he does not even mention the second. Having drawn these distinctions, we want to defend the thesis that the norms that do not correspond to inferences (i) are fundamental for our discursive practices and (ii) that Brandom has to treat them as non-objective. We will argue that (iii) this view is objectionable. Keywords: social norm, objectivity, game of giving and asking for reasons, rationality
1
TWO KINDS OF NORMS
We want to start by introducing the distinction between objective and nonobjective norms. What we mean by objectivity of a norm is more adequately put by saying that the norm gives rise to attitude-transcendent correctness. A norm gives rise to correctness insofar as it declares certain actions to be correct and others incorrect. The attitudes that are said to be transcended are the deontic attitudes of the members of the community, i.e. whether they take an action to be correct or not. Now, the correctness engendered by a norm is attitude-transcendent if and only if it is possible for all of the members of the community to take an action to be correct although it in fact violates the norm. So the correctness of an action according to a norm can transcend the
80
Laukötter, Prien, Schepelmann, Thein
deontic attitudes of the community members in the sense that the correctness is independent of those attitudes. Brandom claims that both attitude-transcendent and non-attitude-transcendent norms exist. Examples of the latter kind are conventional norms. “Whatever the Kwakiutl treat as an appropriate greeting gesture for their tribe […] is one; it makes no sense to suppose that they could collectively be wrong about this sort of thing.” (MIE 53) In cases like this, what the large majority of the community-members takes to be correct is correct. Therefore, it makes no sense to assume that everyone could be wrong about the appropriateness of a gesture. Or, to take another example: When everybody accepts that male bank employees should wear ties and not pearl-necklaces it makes no sense to assume that they are mistaken. While Brandom concedes that there are norms of this sort, he insists that conceptual norms are not of this sort (cf. MIE 53). Brandom claims that the norms that underwrite the inferences constituting the conceptual contents of assertions give rise to attitude-transcendent correctness. Even if all of the members that ever did or ever will belong to the community took a certain inference to be good, it could still be that the inference is not good.1 For instance, it is possible that everybody who is, was, or will be a member of our linguistic community takes the inference from “There is lightning” to “There will be thunder” to be materially good, while in fact this inference is not good.
2
HOW CAN COMMUNAL PRACTICES INSTITUTE TRANSCENDENT CORRECTNESS?
A full discussion of how social practices can institute attitude-transcendent correctness would be beyond the scope of this paper. However, as this will be important for our discussion of fundamental discursive norms later on, we now want to describe what we take to be a central feature of Brandom’s account. This is the idea that our discursive practices are solid, corporeal, even 1
Brandom takes it to be one of the prime tasks of MIE to show how norms of this sort can be implicit in our practices. How can norms be instituted by the actual normative attitudes in a community and, at the same time, transcend these attitudes?
Are Fundamental Discursive Norms Objective?
81
lumpy and thick, as Brandom calls it at different places in his writings.2 This means that our practices do not merely consist in the manipulation of symbols, but rather, include the objects to which they refer. When describing the discursive practices we also have to mention the objects we talk about because our practices include language entry and exit transitions. As Brandom says in lecture six of BSD, discursive practices are nothing that would “fall on the ‘word’ side of a word/world gulf.” (BSD 6.1, p. 3) Because of this, “[t]he way the world is, constrains proprieties of inferential, doxastic, and practical commitments in a straightforward way from within those practices.” (MIE 332) Brandom then goes on to illustrate this constraint with his acid example:3 “So if I perceive a liquid as tasting sour, infer that it is an acid, infer further that it will therefore turn litmus paper red, and, intending to match a red pigment sample, accordingly dip litmus paper in the liquid, I may nonetheless subsequently acquire perceptually a commitment to the result being a blue, rather than a red, piece of paper […]. In this way I can find myself with incompatible commitments (which need to be sorted out if I am to remain entitled to any of my commitments in the vicinity).” (MIE 332)4 This constraint by the world on our inferential practices provides an essential part of an explanation of how our actual attitudes can institute correctnesses that transcend these attitudes. The world can tell me, as it were, that some of the inferences I endorse are not really good. I may take it that one can infer from ‘sour tasting liquid’ to ‘acid’, but it may in fact be that one can only infer from ‘sour tasting clear liquid’ to ‘acid’. 2 3 4
Cf. MIE 332, MIE 632, BSD 6.1. The fact that this example keeps reappearing in Brandom’s writings testifies to the importance of the process described here. It is interesting that in later writings (for example, the Hegel-papers in TMD) Brandom speaks of an obligation to remove such incompatibilities, instead of a commitment. This is surprising because in MIE, Brandom explains that he prefers the term ‘commitment’ to ‘obligation’ because of the connotations of commands or edicts from a superior involved in the latter term (cf. MIE 160).
82
Laukötter, Prien, Schepelmann, Thein
This constraint by the world on the inferential proprieties opens up the possibility of attitude-transcendent correctness of inferences: Even if everyone in the community were to infer from ‘acid’ to ‘turns litmus paper red’, it could still be that this inference is not good because, as a matter of fact, there are sour tasting liquids that do not turn litmus paper red. In that case it would not be correct to draw the above inference, even if it is universally endorsed.
3 FUNDAMENTAL DISCURSIVE NORMS Having dealt with the distinction concerning the objectivity of norms we now turn to the distinction between inferential norms and fundamental discursive norms. This is a distinction in the domain of the norms that define the game of giving and asking for reasons, and it is one that Brandom, as far as we can see, never mentions. The game of giving and asking for reasons contains norms that correspond to inferences, for example, the commitment-preserving inference from ‘There is lightning’ to ‘There will be thunder soon’. Corresponding to this semantic relation there is, on the pragmatic level, a norm according to which one is ceteris paribus committed to utter the words ‘There will be thunder’ if one has uttered the words ‘There is lightning’. Similarly, there are norms corresponding to entitlement-preserving inferences and relations of incompatibility. Since these conceptual norms correspond to inferences, we shall refer to them as inferential norms. Now, even if Brandom does not say so explicitly, we think it is clear that the game of giving and asking for reasons also contains norms that do not correspond to an inference or an incompatibility relation. Here are three examples: 1) The obligation to justify one’s claims:5 When a speaker acknowledges a commitment, she incurs an obligation to demonstrate her entitlement if appropriately challenged. Like being committed to further claims, this 5
Interestingly, Brandom uses the term ‘obligation’ in this context: “implicit obligation to vindicate the commitment undertaken by demonstrating one’s entitlement to it.” (MIE 227)
Are Fundamental Discursive Norms Objective?
83
obligation belongs to the normative consequences of undertaking a commitment, but one would not say that there is an inference corresponding to this consequence. 2) The obligation to remove incompatibilities: As the example of the concept ‘acid’ discussed above shows, it can happen that we run into incompatible commitments. When something like this occurs, we are obliged to modify at least one of our inferential commitments. Neither does the norm giving rise to this obligation correspond to an inference. One might try to argue that this obligation is already implicit in the incompatibility of two claims because part of what it is to be incompatible is that this obligation obtains. This may be so, but one still has to distinguish between the norms that correspond to the incompatibility of two claims and the norm according to which one is obliged to do something in this case. 3) The entitlement to ascribe beliefs in the de re mode: If a speaker claims that an object t is F and I am committed to the claim that s and t are identical, than I can draw a substitutional inference to the claim that s is F. Thus far, my discursive actions are legitimized by norms corresponding to inferences. However, for the practice of de re belief ascription, the next step is crucial: Even if the speaker were to deny explicitly that s and t are identical, I am entitled to say that she believes that s is F and criticize her for saying that. The entitlement to do so must be due to a norm that is not already implicit in the inferential norms.6 We shall call norms like these Fundamental Discursive Norms because they have to be present in every practice that is to count as the game of giving and asking for reasons. Moreover, the content of these norms—what they require us to do under which conditions—is not determined empirically as in the case of inferential norms. We have seen in the previous section that the content of inferential norms is constrained by the way the world is, and that we are obliged to modify our inferential commitments if we find out that they do not conform to the way the world is. The content of the fundamental discursive norms, on the other hand, is not determined empirically, and we would not 6
The norm about de re belief ascription is crucial for Brandom’s argument for a representational dimension of conceptual content.
84
Laukötter, Prien, Schepelmann, Thein
modify them in the face of recalcitrant experience. If we did that, we would stop playing the game of giving and asking for reasons.
4
ARE FUNDAMENTAL DISCURSIVE NORMS OBJECTIVE?
Brandom stresses that inferential norms are objective and that this distinguishes them from conventional norms, which are not. We now want to argue that Brandom is committed to the view that the fundamental discursive norms are not objective, either, and that they therefore seem to be mere conventions. At the very least, his theory in MIE lacks the resources to establish a difference between fundamental discursive norms and conventional norms. The following two considerations lead us to this conclusion: 1) Brandom cannot argue for the objectivity of the fundamental discursive norms in the same way that he has argued for the objectivity of material inferential norms. As we have seen above, in the last analysis this objectivity is due to the fact that there is a worldly constraint on what we can take these norms to be. That the world can constrain our inferential norms depends on the following three conditions: i) We can become entitled to claims because of observation, ii) some claims are incompatible with one another, so that we can become entitled to incompatible claims, iii) there is a norm according to which we are obliged to modify our inferential commitments in this case. It is clear that the worldly constraint on our norms does not apply to the fundamental discursive norms because in their case condition iii) is not satisfied. We would not modify the fundamental discursive norms in cases of incompatibilities.7 2) There is a line of argument in MIE that we think can be interpreted as an attempt to establish a difference between fundamental discursive norms and conventional norms. Brandom says that rationality is what distinguishes us from the non-us around us. And since playing the game of giving and ask7
In the case of the norm to remove incompatibilities, there is the further difficulty that this norm is required for the account of the objectivity of the inferential norms. The attempt to account for the objectivity of this norm in the above way would therefore be circular.
Are Fundamental Discursive Norms Objective?
85
ing for reasons constitutes our rationality, one can say, as Brandom does in a slightly different context: “Conversation is the great good for discursive creatures.” (MIE 644) However, the assumption that rationality defines us seems unfounded. Could one not just as well assume that what defines us is faith in God and that worshiping Him is the great good for us? In that case one could argue that the norms that are indispensable for our worshiping practices have a special status while the norms that are indispensable for the discursive practices would sink to the status of mere conventions. So the appeal to rationality does not, in the end, suffice to distinguish the fundamental discursive norms from mere conventions. For all Brandom says in MIE, it could be a mere convention that we should be rational. Since no other way of arguing for the objectivity of the fundamental discursive norms suggests itself, it seems that in MIE Brandom has to accept the consequence that it is up to the community to decide whether there is an obligation to remove incompatibilities, whether one can ascribe beliefs in the de re mode, or whether speakers have to justify their claims if challenged. We think that this consequence is objectionable because at least two of the fundamental discursive norms mentioned here seem to correspond to features of the world. Unlike inferential norms, however, they seem to correspond to what one might call formal features of the world, whereas inferential norms correspond to material features of the world. Let us first explain this regarding the obligation to remove incompatibilities: As Brandom would probably agree, it is a formal feature of the world that it contains properties that are incompatible with one another because it is in this that the determinateness of properties consists.8 That two properties are incompatible in turn means that one and the same object cannot exhibit both of them. There is a correspondence between this formal feature of the world and the obligation of speakers to modify their inferential commitments if they become entitled to attribute incompatible properties to an object. Similarly, one can argue that the entitlement to ascribe beliefs in the de re mode is due to a formal feature of the world, namely that all speakers are talking about one and the same world. When S claims that an object t is F, I can 8
Cf. “Holism and Idealism in Hegel’s Phenomenology”, TMD 179f.
86
Laukötter, Prien, Schepelmann, Thein
use ‘t is identical with s’ as a collateral premise to infer ‘s is F’, even though S would not accept this identity claim. What is more, I can assess S’s claim that t is F by assessing that ‘s is F’. This practice corresponds to the formal fact that S and I are both talking about one and the same world, i.e. that S’s claim ‘t is F’ and my claim ‘t is identical with s’ are supposed to be true of one and the same world. Thus there seems to be an objective reason why we are obliged to remove incompatibilities and why we are entitled to ascribe beliefs in the de re mode. Still, at least in MIE, Brandom ultimately says nothing that could satisfactorily account for this objectivity and distinguish the fundamental discursive norms from conventional norms. We think that this issue is ultimately left unresolved in MIE. However, the question of the significance of one of the fundamental discursive norms—the obligation to remove incompatibilities—is taken up again in “Holism and Idealism in Hegel’s Phenomenology” (TMD ch. 6) and especially in the sixth Locke Lecture. There, Brandom distinguishes between two senses of incompatibility. On the one hand, two properties are incompatible in the objective sense if one object cannot exhibit both of them. On the other hand, two claims can be incompatible in the subjective sense if speakers are not permitted to endorse both of them. Brandom then mentions two ways in which one might try to account for the objectivity of the obligation to remove incompatibilities. The first strategy (called objective pragmatism) assumes that there are objectively incompatible properties (i.e. properties that cannot both be exhibited by one object) and argues that because of this, there is an obligation to remove subjective incompatibilities. The second strategy (called subjective pragmatism) argues, in a way reminiscent of Kant’s transcendental idealism, that because our discursive practice requires that there are commitments which cannot both be undertaken by one and the same speaker, there must be possible properties of objects that cannot both be exhibited by a single object. In other words, the second strategy assumes that there are incompatibilities in the subjective sense and argues that this somehow constitutes the formal feature of the world that there are incompatibilities in the objective sense (cf. BSD 6.4, p. 25f). We think that the chances of getting either strategy to work are very slim. But if
Are Fundamental Discursive Norms Objective?
87
they could be made to work, they would indeed account for the objectivity of the obligation to remove incompatibilities. In any case, Brandom thinks that one cannot adopt either of these explanatory strategies because they presuppose either that one can understand incompatibility in the objective sense apart from incompatibility in the subjective sense, or vice versa. Against this, Brandom argues that both senses of incompatibility are reciprocally sense-dependent, that one can understand each only if one also understands the other. This may be so, but we think that this still leaves the question of the objectivity of the obligation to remove incompatibilities unresolved. Brandom’s thesis about reciprocal sense-dependence is a thesis about our grasp of certain concepts. And as such it offers as yet no explanation of why speakers should avoid subjective incompatibilities.
REFERENCES Brandom, Robert 1994. Making It Explicit: Reasoning, Representing, and Discursive Commitment, Cambridge, Mass.: Harvard University Press, (MIE). Brandom, Robert 2002. Tales of the Mighty Dead: Historical Essays in the Metaphysics of Intentionality, Cambridge, Mass.: Harvard University Press, (TMD). Brandom, Robert 2006. Between Saying and Doing: Towards an Analytic Pragmatism, Locke Lectures held at Oxford University, Trinity Term 2006, to appear 2008 from Oxford University Press. Unpublished Manuscript at http://www.pitt.edu/~brandom/ locke/index.html, (BSD).
REALIST AND IDEALIST INTERPRETATIONS OF BRANDOM’S ACCOUNT OF OBJECTIVITY
Michael Pohl, Raja Rosenhagen, Arne M. Weber
Abstract: The text deals with the question of how to understand Brandom’s strong notion of objectivity. The first section outlines Brandom’s views on two important concepts closely linked with the concept of objectivity, namely: truth and reference. Keeping in mind those views, two ways of interpretation suggest themselves, a realist and an idealist reading of the notion of objectivity. The two following sections are devoted to analyses and an assessment of the viability of these readings, respectively. The authors draw the conclusion that only the idealist reading suggested is in line with Brandom’s overall philosophical project. In the last section it is shown that, if one accepts the idealist reading as appropriate, a satisfactory clarification of the concept of causality and of the use of causal vocabulary is a task yet to be accomplished. Keywords: objectivity, idealism, realism, reference, transcendental argument
1
INTRODUCTION
When reading Robert Brandom’s works, we felt that some fundamental questions concerning an appropriate interpretation of his notion of objectivity remained yet to be answered. In order to remove that feeling, we shall try to attribute to Brandom an ontological position which both lines up best with his overall philosophical project and leads to an interpretation of the notion of objectivity which meets the criteria of adequacy Brandom sets up for it. In the first section we start by introducing Brandom’s view on some important concepts, ending up at a point where two different ways of interpreting objectivity come into view—a realist and an idealist reading. Accordingly,
90
Michael Pohl, Raja Rosenhagen, Arne M. Weber
in the second section we analyze the viability of a realist reading of his notion of objectivity, which involves a commitment to some sort of ontological realism. In the third section we are going to pay attention to the other, idealist interpretation, in search for an adequate conception of objectivity implicit in what Brandom says about our linguistic practices. At the end of our inquiry another critical issue comes into play. In the fourth section we claim that if one follows the interpretation we recommend, a sufficient clarification of the concept of causality and of the use of causal vocabulary does not yet seem to be provided by Brandom. Thus, although we will offer a tentative suggestion as to how Brandom’s account of objectivity might be developed in order to provide the clarification needed, we nonetheless see the necessity of further explication and of tying up some loose ends which are still dangling in the air.
2
PRELIMINARIES—TRUTH, REFERENCE, FACTS
In the fifth chapter of Making It Explicit, Brandom develops a deflationary account of truth and reference. Sophistically refining a prosentential contentredundancy theory of truth, he tries to show that the expressions ‘true’ and— in a parallel fashion—‘refers’ and its cognates could and should be construed as functioning anaphorically (MIE 283). The interpretation of ‘refers’ comes— from a traditional perspective—“in two flavors” (MIE 306), as Brandom calls it, i.e. as an intralinguistic word-word-relation and as an extralinguistic wordworld-relation (cf. MIE 306). In line with his deflationist likings, Brandom, however, claims that “an analysis in terms of anaphoric mechanisms can provide the resources for a purely intralinguistic account of the use of the English sentences by means of which philosophers make assertions about extralinguistic referential relations” (MIE 306). As part of the explicating and thus—according to Brandom’s original account of logic—logical locutions, ‘true’ and ‘refers’ “play the expressive role of mak-
Realist and Idealist Interpretations of Brandom’s Account of Objectivity
91
ing explicit the [purely intralinguistic] anaphoric relations that make it possible for unrepeatable tokenings to play indirectly inferential roles and so to have conceptual content.” (MIE 498) Without delving into the details of the account of anaphora, deixis and substitution on which Brandom’s strategy depends, we shall now take a look at the advantage Brandom sees in following his approach: As Brandom has it, the reification of ‘true’ and ‘refers’ amounts to a fatal philosophical error motivated by a mistaken grammatical analogy in which ‘true’ is taken to be a predicate and ‘refers’ is understood as indicating the existence of a relation. Accordingly, “truth and reference are philosophers’ fictions, generated by grammatical misunderstandings” (MIE 324), by drawing a “mistaken grammatical analogy to predicates and relational expressions.” (MIE 323, see also MIE 499). Clearly, it is precisely this philosophical error which Brandom seeks to avoid. This strategy is motivated by his overall pragmatist policy, which is opposed to the representationalist approach and involves a “characteristic shift of explanatory attention […] from what is represented to representings of it.” (MIE 292) The general idea is that “representeds are explained in terms of the representings, instead of the other way around.” (MIE 292) Interestingly, in defending the closely related view that if one calls something a fact one just takes it to be true and thereby endorses it, Brandom concedes that there is no need to deny “the general possibility of expressing semantic content truth-conditionally” (MIE 329). Rather, what is doubted is the possibility of “explaining semantic content truth-conditionally” (MIE 329) and—hence—of “treating representational locutions as basic in the order of semantic explanation” (MIE 496, emphasis added). Following this pragmatic line of thought one might worry that—as Brandom himself puts it— “a semantic idiom that identifies facts with true claims (via the identification of taking to be a fact with taking to be true […]) must inevitably slose the worlds—trading its solidity for a froth of words […] [and ending up in a] threatening idealism of linguistic practice [that] seems to be implicit in such an identification.” (MIE 330f) In this situation, wicked linguistic idealism seems to lurk just around the corner. But it is one of Brandom’s own criteria of adequacy for his overall phi-
92
Michael Pohl, Raja Rosenhagen, Arne M. Weber
losophical project that it should allow for a strong notion of objectivity in the sense that the correctness or successfulness of our empirical claims does not depend on what anybody (an expert, a scientist, or even the whole community) takes to be the case, but on what really is the case. For there are empirical and practical constraints on our discursive practice and it “is not up to us which claims are true (that is, what the facts are).” (MIE 331, emphasis added) One point of view opposed to idealism is realism. So in the next section we will examine whether it is possible to construe the strong notion of objectivity Brandom seeks to establish in a realist fashion, i.e. whether it is possible to interpret Brandom as trying to avoid the threat of linguistic idealism by adopting a realist perspective.
3
THE REALIST READING
On the realist construal of Brandom’s position pursued in this section the strong notion of objectivity has to be understood as involving a commitment to ontological realism. By ‘ontological realism’ we mean the claim that there is an objective, (mostly) language-independent world,1 which is what we are talking about when we make, for example, empirical claims. The question we want to address now is whether Brandom—presupposing that he is at least implicitly committing himself to such a claim—can, in order to escape the threatening idealism, convincingly show that such an objective world exists that has a certain authority over us in the sense that the correctness of our representational claims depends on the way the world really is. A clue to what the answer to this question might be can be extracted from Brandom’s account of practices. He claims that “[d]iscursive practices incorporate actual things. They are solid—as one might say, corporeal: they involve actual bodies” (MIE 332). It is in the same spirit when Brandom—in his more recent Locke Lectures—asserts that “practices and abilities are not the kind of thing that can be separated from the objects they involve […]. Engaging in discursive practices and 1
For sake of simplicity we will from now on use the expression ‘objective world’.
Realist and Idealist Interpretations of Brandom’s Account of Objectivity
93
exercising discursive abilities is using words to say and mean something, hence to talk about items in the world. Those practices, the exercise of those abilities, those uses, establish semantic relations between words and the world.” (BSD 6.1, p. 3) The idea seems to be that directedness to and constraint by an objective world is somehow built into our discursive practices (at least when they involve empirical concepts). In Brandom’s pragmatist theory, at least some of our discursive practices are referred to as feedback-governed ‘thick’ practices, using “ ‘thick’, in the sense of essentially involving objects, events, and worldly states of affairs. [Thus, b]its of the world are incorporated in such practices, in the exercise of such abilities.” (BSD 6.1, p. 4) That practices are feedbackgoverned means, according to Brandom, that they are constantly further developed by repeatedly going through a cycle of perception, performance, assessment of the performance and further performance.2 This cycle works as follows: A subject is reacting to a state of affairs by performing a certain speech-act (here perception and performance take place). If a kind of inferential incompatibility between the concepts the subject is using in performing that speech-act exists (which becomes apparent in the stage of assessing the performance), then she is committed to changing the concepts she was using (which, when done, is itself a performance and leads to further performances modified accordingly). In this way, the facts are setting up authority over the subject’s concepts.3 That the subject is committing herself to correct or develop her conceptuse whenever she comes across material incompatibilities between those concepts is—on the realist construal pursued here—just what it means that the subject is answerable to an objective world. Brandom writes: “[A]cknowledging material inferential and incompatibility relations essentially involves representing objects as having properties […] that stand in corresponding relations to one another. In drawing inferences 2 3
Cf. BSD 6.1, p. 3f. This cycle is also referred to as “Test-Operate-Test-Exit (TOTE) cycle” (ibid.). For a concrete example that illustrates how the cycle works, see BSD 6.3, pp. 12-13.
94
Michael Pohl, Raja Rosenhagen, Arne M. Weber
and ‘repelling’ incompatibilities, one is taking oneself to stand in representational relations to objects that one is talking about.” (BSD 6.3, p. 15) If we follow the realist line of interpretation, at this point the impression seems to be forced upon us that Brandom—if he really wants to be a realist—relies on a transcendental argument in order to establish a strong notion of objectivity. As this argument runs, it is just because it is not possible to describe our linguistic practices or to be engaged in them without presupposing the conception of an objective world, and because it is futile to question whether we (and others) are engaged in such a practice, that there has to be such a world. There are, however, three problems with the realist reading. First, there are significant doubts whether the line of thought presented in the last paragraph really suffices to show what it is meant to, i.e. that there is an objective world of facts that has a certain kind of authority over us. At best, it shows that we have to presuppose the existence of such a world when we describe discursive practices of other subjects, or when we ourselves are engaged in discursive practices. Even if this is taken to be sufficient to show that one cannot coherently question a realist ontology, it certainly does not show that such a realist ontology is correct. Transcendental arguments show us something about the structure of our mind, in this case about the ontological commitments we might necessarily have to undertake in our actual discursive practices, but these arguments do not tell us anything about the objective world itself. So, as it seems, a transcendental argument is a very weak argument in the end.4 Second, it seems that Brandom would rather reject having made use of a transcendental argument at all. Take for example the following quotes: “[T]he acknowledgement of the existence of conceptually structured facts to which our practices […] answer […] is not meant to have any explanatory value except what can be cashed out in terms of the deontic and social-perspectival articulation of our discursive practices. In particu4
Cf. Stroud 1968; or the articles recently published in Stern 1999.
Realist and Idealist Interpretations of Brandom’s Account of Objectivity
95
lar, it is not intended to explain so much as the possibility of that articulation – rather, the other way round.” (Brandom 2000a, 361) “Our practices of comparing, assessing, and correcting different repertoires of commitments one with respect to another – those we attribute to others and those we undertake ourselves – are what make them intelligible as perspectives, views of something, ways in which a perspectiveindependent reality can appear.” (Brandom 2000b, 431) These quotes indicate that Brandom does not want to say that the existence of an objective world is necessary to explain the possibility of our discursive practice, as a transcendental argument (unsuccessfully) purports to, but that in turn our discursive practices explain the possibility of the conception of such a world. In fact, it even seems that the conception of an objective world is for Brandom just a product of our discursive practice. We will come back to that line of thought in the next section. The third problem is this: It seems that Brandom cannot really choose the realist option anyway. The use of the notion of an objective world— realistically construed—seems to involve commitment to the existence of objects that we might—if we are lucky—actually succeed in referring to. So if the answer to the question of what the facts are takes in an allusion to such a notion of the objective world, this answer threatens to conflict directly with Brandom’s deflationary account of the use of ‘refers’. For being committed to the existence of such objects allows for the possibility of making sense of an extralinguistic use of ‘refers’ which cannot be accounted for in terms of the purely intralinguistic account Brandom offers.5
5
Allowing such extralinguistic uses of ‘refers’ that cannot be analyzed in terms of an intralinguistic account would make it difficult to see how to avoid the conclusion of Quine’s that “what matters semantically is not meaning but reference”, a conclusion Brandom refuses, as it involves giving up “on contents as inferential roles, in favor of a different sort of primitive.” (MIE 481f)
96
Michael Pohl, Raja Rosenhagen, Arne M. Weber
4
THE IDEALIST STRATEGY
In the previous section we tried to show that there are significant reasons to doubt that Brandom’s strong notion of objectivity should be understood in a realist fashion. But if a realist understanding of the conception of objectivity seems highly problematic, how can Brandom both avoid the threat of linguistic idealism and secure a rather strong notion of objectivity? In the conclusion of MIE, where Brandom explicitly addresses the question of the objectivity of norms, he claims: “[C]onceptual norms implicit in the practices attributed to a community outrun the nonnormatively specifiable behavioral discriminations members of that community are disposed to make. For this reason, conceptual norms can be understood as objective, and so as binding alike all members of a discursive community, regardless of their particular attitudes.” (MIE 631) The uptake seems to be that a strong notion of objectivity is secured by insisting on the fact that the norms to which we actually are committed implicitly by our practices are not identical with what we would take ourselves to be committed to by explicitly formulating rules. In this way, what we are committing ourselves to depends on what (implicitly) is the case and not on what we (explicitly) take to be the case. The question is, then, how the norms that actually are implicit in a community’s practices can come into view. A prima facie plausible answer is that they can only be identified from the third-person perspective of an external interpreter who ascribes objective commitments to members of that community by examining the (causal) connections between the given facts (i.e. what is true from the perspective of the interpreter) and the way communitymembers respond to them. But the interpreter is himself committed to some objective norms governing his practice of assessing. These norms also only come into view from the standpoint of another interpreter. This seems to lead into a regress which, for Brandom, is unsolvable. Thus, he concedes that
Realist and Idealist Interpretations of Brandom’s Account of Objectivity
97
“[t]here is never any final answer as to what is correct; everything, including our assessments of such correctness, is itself a subject for conversation and further assessment, challenge, defense, and correction.” (MIE 647) But then, taking into account the social dimension of our linguistic practices,6 we come to see that each of us is in a third-person perspective to each other member of the community and can make the other’s norms explicit.7 This creates what Brandom calls an interpretative equilibrium, with all members of the community interpreting each other reciprocally, making each other’s commitments explicit (MIE 642). Thus, the asymmetry between the interpreter and the interpreted community collapses. The interpreter and the interpreted are part of the same community (MIE 644). What follows, however, is that we commit ourselves to norms we do not know and perhaps never even could know. The objectivity of our norms could only come into view from a perspective completely external to us—which would have to be a kind of “bird’s eye view” (cf. MIE 601). But such a view is not intelligible, for every interpreter who is taking someone as engaged in linguistic practices and is thereby assessing this someone’s commitments according to her own norms is also taking it that this someone is in turn assessing her according to this someone’s norms. That is, she is taking the interpretee as belonging to her (the interpreter’s) community—and can therefore never adopt a standpoint outside of it. That just is what Brandom means when he says that “there can never be any final answer as to what is correct.” (MIE 647) Now, it seems that Brandom wants us to conceive of ourselves as being confined within the boundaries of our interpretative equilibrium, constantly developing our concepts, but never really being able to get a full grip on what is objectively the case, as this is always just out of our expressive reach, being only implicit in our practices. The idea of a perspective-independent world, as 6 7
The important point is that our practices include the full expressive resources supplied by logical vocabulary (cf. MIE 642f). Thus, “[t]he underlying difference between the two stances [i.e. the stance of the interpreter and the stance of the interpreted] is, not the distinction between communally external and internal attitudes or interpretations, but the distinction between explicit and implicit ones.” (MIE 639)
98
Michael Pohl, Raja Rosenhagen, Arne M. Weber
we might want to remember at this point, does not have any explanatory value. It rather seems to be just some kind of regulative idea, which we daringly reify, thereby falling prey to a misunderstanding. However, it might well be the case that, in the way referred to, a strong notion of objectivity can be secured. But as it seems to us, this involves subscribing to a certain sort of idealism about the world as such.
5
CAUSALITY
As has been already mentioned,8 one criterion of adequacy Brandom sets up for himself is that his position should allow for a strong conception of objectivity. In the second section we showed that there are significant doubts that such a conception can be accounted for by a realist reading, as such a reading would conflict, for several reasons, with other parts of Brandom’s theory. As it turned out in the last section, this does not have to bother Brandom, for a more idealist interpretation seems to allow for a strong notion of objectivity. For, what really is the case is, according to Brandom, not identical with what we take to be the case, but with what we should take to be the case—according to the norms which are implicit in our practices. So, if Brandom really wants us to construe his position as a broadly idealist one, we should be well aware not to fall back into the realist understanding while interpreting other philosophically important notions and the way they figure in the idealist and inferentialist picture Brandom offers. It seems that many of those concepts are well accommodated, for instance ‘truth’ or ‘reference’. One concept, however, which is—at least implicitly— important to Brandom’s position and which, moreover, seems particularly susceptible to a realist misinterpretation is that of causality. Thus, the concept of causality seems to be crucial in the context of Brandom’s account of what he calls ‘RDRDs’—reliable differential responsive dispositions. Thus, Brandom seems to endorse an externalist semantic in a Davidsonian sense.9 Plus, inferring is construed as “a causal process that relates deontic attitudes” (MIE 8 9
Cf. p. 3 in this paper. MIE 632.—For an overview of Davidson’s position within the externalist tradition see his “Epistemology Externalized”.
Realist and Idealist Interpretations of Brandom’s Account of Objectivity
99
260, emphasis added) and reasons, too, “can be causes because deontic scorekeeping attitudes can play both normative and causal roles.” (MIE 271) Of course, Brandom states quite clearly that he has overcome the traditional dualism between concepts and causes (cf. MIE 618-623). Such a dualism contrasts “the conceptual, as the product of cognitive activity, with the nonconceptual impingement on cognitive receptivity in virtue of which that cognitive activity is constrained.” (MIE 620) Hence, on such a “conception of concepts, they function as epistemological intermediaries. They stand between the understanding mind and a world that is the source of their content or matter—a world composed of particulars that are grasped by means of general concepts and that imposes itself causally on a mind obliged somehow to conform itself to those causal impingements.” (MIE 617f) Brandom, however, claims that ”[c]oncepts […] do not serve as epistemological intermediaries […]. This is not because there is no causal order consisting of particulars, interaction with which supplies the material for thought. It is rather because all of these elements are themselves conceived as thoroughly conceptual, not as contrasting with the conceptual.” (MIE 622) Further, when explaining how to understand ‘thick’ practices, Brandom says that “it is wrong to contrast discursive practice with a world of facts and things outside it, modeled on the contrast between words and the things they refer to.” (MIE 332) These passages, among others, indicate that causation, too, should be understood as a relation located completely within the realm of the conceptual and not be construed as relating that realm with something outside it. But, that being so, the question remains, how, exactly, we have to understand causality that way? To us, it is not completely clear, how causality, presupposed that the realist reading of objectivity is bound to fail, can both be understood as something thoroughly conceptual and still play the crucial role it does in providing empirical content, in developing RDRDs, et cetera. Hence, as it
100
Michael Pohl, Raja Rosenhagen, Arne M. Weber
seems, Brandom’s conception of causality is still in need of further clarification. The general strategy we feel Brandom might want to pursue is as follows: Logic, as Brandom has it, should be conceived as the expressive organ of selfconsciousness (cf. MIE 643). That is, logical vocabulary is expressive vocabulary. In using causal vocabulary we make ourselves explicit as beings that make themselves explicit as standing in relation to a world, which is taken to be outside of them. In that way, the causal vocabulary seems to be essential to explicate the conception of ourselves and should thus be construed as an important part of the expressive, i.e. logical vocabulary. We think that developing the conception of causality somehow along these lines might both fit well in the overall idealist interpretation suggested in the third section and avoid falling back into the realist reading criticized in the second section. But, whatever the general merits of such an account in the light of Brandom’s overall project may be, working out the details and answering the ensuing question of what relation holds between causation and inference remains an important task for future research.
REFERENCES Brandom, Robert 1994. Making It Explicit: Reasoning, Representing, and Discursive Commitment, Cambridge, Mass.: Harvard University Press, (MIE). Brandom, Robert 2000a. “Facts, Norms, and Normative Facts: A Reply to Habermas”, in: European Journal of Philosophy 8, pp. 356-374. Brandom, Robert 2000b. “Knowledge and the Social Articulation of the Space of Reasons”, in: Epistemology – An Anthology; ed. by Ernest Sosa, Malden, Mass.: Blackwell, pp. 424-432. Brandom, Robert 2006. Between Saying and Doing: Towards an Analytic Pragmatism, Locke Lectures held at Oxford University, Trinity Term 2006, to appear 2008 from Oxford University Press. Unpublished Manuscript at http://www.pitt.edu/~brandom/ locke/index.html, (BSD). Davidson, Donald 1991. “Epistemology Externalized”, Dialectica 45. Stroud, Barry 1968. Transcendental Arguments, in: The Journal of Philosophy, Vol. LXV, No. 9, pp. 241-256. Stern, Robert 1999. Transcendental Arguments. Problems and Prospects; Oxford: Clarendon Press.
COMPARING BRANDOM’S CRITICAL READING AND HEGEL’S SYSTEMATIC ENTERPRISE
Jana Elisa Falkenroth, Attila Karakuş, David P. Schweikard
Die der Zeit nach letzte Philosophie ist das Resultat aller vorhergehenden Philosophien und muss daher die Prinzipien aller enthalten; sie ist darum, wenn sie anders Philosophie ist, die entfaltetste, reichste und konkreteste. (Hegel)
Abstract: Brandom’s work on Hegel does not only fit into his project of supplementing his systematic account of discursive practices and conceptual normativity with studies on the historical predecessors of his inferentialism, it has also drawn considerable attention in Hegel scholarship. In this paper, however, we object to the assumptions underlying Brandom’s treatment of Hegel’s theory of concepts in the Phenomenology of Spirit and in the Science of Logic. In particular, we show that Hegel, contrary to Brandom’s interpretation, conceived of these two works as pursuing fundamentally different tasks in relation to Hegel’s systematic philosophy. Acknowledging particularly the role of Hegel’s Logic as a categorical ontology and metaphysical theory of subjectivity in the attempt to reappropriate Hegel’s insights into the nature of concepts or the relational constitution of self-consciousness, results in an alternative critical reading, one that does not eschew disclosing the ontological assumptions that shape and pervade Hegel’s system. Keywords: Hegel, empirical concepts, logical concepts, systematic philosophy, ontological commitment
1
INTRODUCTION
If one were to formulate, locate and specify the dimensions of Brandom’s main philosophical projects, it would seem apt to mention two big issues. These two issues are (i) “the explication of concepts” in the sense of “the de-
102
Jana Elisa Falkenroth, Attila Karakuş, David P. Schweikard
velopment and application of expressive tools with which to make explicit that which is implicit in the use of concepts” (Brandom 2001, 77) and (ii) the study of leitmotifs in the history of “the metaphysics of intentionality”.1 Hegel is a major point of reference in the latter project, or rather, a constant inspiration and role model for both of these projects. In a set of more recently published articles,2 Brandom seeks to reconstruct, interpret and elaborate on the central views he takes Hegel to be arguing for. In 1997, Richard Rorty had already characterized Brandom’s work as “an attempt to usher analytic philosophy from its Kantian to its Hegelian stage” (cf. Rorty 1997, 8f). And indeed, this move is suggested even in the concluding remarks of Brandom’s Locke Lectures, in which he identifies pragmatically mediated semantic relations which are crucial to that work as “the essence of the view [Hegel] develops under the heading of ‘idealism’.” (BSD 6, p. 32) In lieu of trying to capture the significance of Brandom’s work on Hegel for analytic philosophy, or of reconstructing and commenting on the importance of these studies on Brandom’s own philosophy, in the following we shall concentrate on some crucial features of the way this systematic appropriation of Hegel is carried out. The best point of reference for this purpose seems to be Brandom’s programmatic essay entitled ‘A Sketch of a Program for a Critical Reading of Hegel—Comparing Empirical and Logical Concepts’ in which the two projects mentioned above coalesce and in which he provocatively sets out to specify one aspect of “what is living and what is dead in Hegel’s philosophy.” (SPCRH 131) In preparing the ground for his reflections on the way he takes Hegel as determining and relating empirical and logical concepts, Brandom distances himself from the view that no treatment of Hegel can bypass the fundamental commitment to monolithic systematicity. Quite obviously, this is an immensely important methodological question on which any interpreter of Hegel’s systematic philosophy has to take a stand. From Rolf-Peter Horstmann’s perspective, in order to make use of any of Hegel’s claims, it is 1 2
See Brandom, Tales of the Mighty Dead—Historical Essays in the Metaphysics of Intentionality. See TMD, chap. 6 and 7, “Selbstbewusstsein und Selbst-Konstitution”, and “Sketch of a Program for a Critical Reading of Hegel—Comparing Empirical and Logical Concepts”, in the following referred to as SPCRH.
Comparing Brandom’s Critical Reading to Hegel’s Systematic Enterprise
103
imperative to accept the claim that the encyclopaedic system of the philosophical sciences developed by Hegel forms a comprehensive systematic whole (cf. Horstmann 1999). In short, it is all or nothing. While Horstmann is very skeptical about the strategy of extracting bits from this whole, since it breaks with the strong holistic character of Hegel’s system, Brandom rejects this picture of the situation. One can, so he says, reject the monolithic systematicity claim and still inquire into “other master-commitments or insights of Hegel’s [that] can be made sense of apart from the systematicity claim, so as to see that they do not in fact require or entail it.” (SPCRH 133) Let us at this stage observe that Horstmann and Brandom seem to be talking at cross purposes regarding the precise nature of Hegel’s strong systematic holism. As Brandom puts it, what is at issue is the semantically monolithic character of Hegel’s philosophical system, where ‘semantic’ stands for the structure and workings of everyday and (especially) philosophical concepts. Horstmann, on the other hand, does not identify Hegel’s term ‘concept’ with the linguistic or logical aspect of its meaning, but claims that what is problematic about accepting Hegel’s legacy is that its “value can only be seen in the peculiarities of Hegel’s theory, which lie in the means he provides for explaining [—for instance—] social and political phenomena. These means undoubtedly consist in the conceptual and logical apparatus Hegel sets down as the rational basis for such explanations.” (Horstmann 1999, 283) By ‘conceptual and logical apparatus’, Horstmann is referring to a set of interconnected metaphysical, methodological and epistemological claims that he takes as underlying every aspect of Hegel’s philosophy. We do agree with Brandom that a systematic appropriation of Hegel’s philosophy, i.e. one that aspires to make sense of Hegel’s insights into the nature of certain fundamental philosophical problems, does not have to take the shape or even aspire to the absolute dimensions of Hegel’s entire system.3 So far, selectivity seems fully justified. But we think Brandom goes too far in chipping off bits of the systematic monolith Hegel has created and thereby not only risks blinding out essential aspects of Hegel’s thought, but even ven3
Alternatively, we could say that we agree to a certain extent that the de re-mode of interpretation—as Brandom specifies it in chapter 3 of TMD—yields more fruitful results philosophically.
104
Jana Elisa Falkenroth, Attila Karakuş, David P. Schweikard
tures to fall back behind Hegel by arguing for a position of a type that Hegel explicitly intended to overcome. In order to substantiate this critique and to pinpoint the scope and the consequences of Brandom’s reading of Hegel, we will first reconstruct the two theses about the relation between empirical and logical concepts that Brandom attributes to Hegel and outline Brandom’s rejection of them (2). Instead of contesting this rejection, we will then look more closely at three more controversial background assumptions that support the attribution of these theses (3). This will motivate the sketch of an alternative program for a critical reading of Hegel that preserves what we take to be both ‘living in Hegel’ and essential to the philosophical enterprise (4).
2
A DISTINCTION AND TWO THESES
Let us start with a brief reconstruction of how Brandom states and treats the point we want to discuss. (i) The point of departure in Brandom’s “Sketch of a Program …” is a distinction between two categories of concepts: empirical and logical concepts. Brandom aims at comparing and contrasting Hegel’s views about the nature of the concepts that belong to these two categories respectively. The distinction as such and definition of these categories is, however, Brandom’s and not Hegel’s. Brandom takes the first category to comprise “ordinary determinate concepts”, or more specifically, “the contents of the concepts expressed by our use of ordinary words, such as ‘red,’ ‘rigid,’ and ‘rotten,’ and of the claims expressed by our use of ordinary declarative sentences containing them.” (SPCRH 134) We use such concepts to articulate both our empirical consciousness, as in ordinary reports about perceptions or intentions, and our empirical self-consciousness. The category of logical concepts is taken to comprise “the logical philosophical concepts and speculative philosophical claims whose development is the subject matter of both the Phenomenology of Spirit and the Science of Logic.” (SPCRH 134) This first step towards a characterization of logical concepts is clearly already part of the reading of Hegel that Brandom wants to set out: it is a localization of Hegel’s treatment of logi-
Comparing Brandom’s Critical Reading to Hegel’s Systematic Enterprise
105
cal concepts. Their systematic definition is given by way of describing their relation to empirical concepts. According to Brandom, and this is a view he also attributes to (or even one he himself inherited from) Hegel,4 this relation is “an expressive one.” And he continues: “To be a logical concept, for Hegel, is to play the distinctive expressive role of making explicit general features of the use and content of ordinary, nonlogical (‘determinate’) concepts. It is to be a certain kind of metaconcept.” (SPCRH 134) In short, we have two conceptual levels, a ground-level of empirical concepts and a meta-level of logical concepts whose role is to express the structural characteristics of contents of ground level concepts and claims. (ii) Jumping straight to the results of Brandom’s analyses, let us now take a look at the two theses he attributes to Hegel: 1. Hegel thinks logical concepts are different from empirical concepts in that because of the difference between their characteristic expressive tasks, while there can for reasons of deep principle be no final, stable, expressively complete set of empirical concepts, there is a final, stable, expressively complete set of logical concepts. 2. Hegel thinks that logical concepts are like empirical concepts in that in spite of the difference between their characteristic expressive tasks, in principle, the only way to understand, specify, or convey the contents of both sorts of concepts is by a rational reconstruction of a history of their development (SPCRH 159). Note that in both theses the thought attributed to Hegel is not specified with respect to the place where Hegel formulates or defends these views. It is clear from Brandom’s analyses that he takes these views to be present in both the Phenomenology of Spirit and the Science of Logic, which he regards as “essentially the same system” (SPCRH 157). We will have to say more about this claim, which we label the Same-System-Claim, later on. (iii) Before we turn to this and other implicit assumptions, it is worth looking briefly at Brandom’s rejection of these theses. For Brandom’s critical reading does not result in an elaboration or even an adoption of exactly the 4
Brandom himself indicates having appropriated this view of logical concepts from Hegel in footnote 5 of SPCRH, where he refers to his own elaboration on this conception in chapter 1 of his Articulating Reasons.
106
Jana Elisa Falkenroth, Attila Karakuş, David P. Schweikard
views he detects in Hegel. Rather, studying the intricacies of Hegel’s ideas serves in part as an interpretative exercise whose function is to help develop tools for expounding systematic views that are independent of the actual position from which the structural insights are derived.—Brandom rejects the first claim on the grounds that he thinks Hegel is mistaken in believing that it is possible to give such a complete set of logical concepts and to “finish the expressive task to which logical vocabulary is addressed.” (SPCRH 159) Consequently, he is not convinced that Hegel, as he claimed, developed once and for all the final and complete set of logical concepts. The latter part of this rejection is based on the fact that since Hegel’s time the logical vocabulary has been significantly extended to include, for instance, iterated quantifiers. The claim that it is impossible to complete this task reaches further into the understanding of the nature of logical vocabulary and, as we are now in a position to say, into the methodology of Brandom’s own elaboration of expressive tools in his Locke Lectures.—Regarding the second claim, Brandom rejects the idea that pursuing what he calls a ‘semantic genealogy’ is the only way to understand the contents of these two sorts of concepts. His own strategy in MIE is, in his own words, to start “with a pragmatic metalanguage, in which we describe the practices that confer and determine conceptual content, and specify in those terms what it is that a particular logical locution makes explicit, introducing it as expressing those features.” (SPCRH 160) In the following, we do not aim to reject Brandom’s rejections or straightforwardly defend the claims he attributes to Hegel. Rather, we shall scrutinize some of the controversial presuppositions that motivate and buttress the attribution of those two claims to Hegel. Our discussion of these presuppositions ultimately not only turns on the reading of Hegel or Hegel’s systematic enterprise, but on Brandom’s understanding of the philosophic enterprise as such.
3
PRESUPPOSITIONS
Without further discussion of that matter, it seems high time to say that we do not intend to object in principle to Brandom’s idea of a critical reading as a
Comparing Brandom’s Critical Reading to Hegel’s Systematic Enterprise
107
method of dealing with classical philosophical texts.5 That is, we do not want to quarrel over the blend of de re- and de dicto-specifications in a reading of Hegel that is meant to be both systematically illuminating and historically accurate. Our aim is to follow the traces detectable in Brandom’s reading of Hegel to his own fundamental philosophical convictions. We will restrict ourselves to three points: (i) the understanding of the role or function of logical concepts, (ii) the relationship between Hegel’s Phenomenology of Spirit and his Science of Logic and (iii) the aim and the subject matter of the Science of Logic. All three points are related to the Same-System-Claim alluded to above.6 Obviously, these points are all controversial and we do not want to engage in a discussion about the ‘correct’ reading of Hegel, assuming there could be one. (i) According to Brandom, Hegel assigns to logical concepts the role of expressing and making explicit the “general features of the use and content of ordinary, nonlogical (‘determinate’) concepts.” (SPCRH 134) On this account, the logical vocabulary forms a kind of metavocabulary with respect to the empirical vocabulary, which is the object of its expressive task. But even though Brandom hastens to indicate the places in Hegel’s writings where this understanding of logical concepts can be made out, it is hard to suppress the impression that this does not capture the entire meaning assigned by Hegel to the predicate ‘logical’. In developing a specifically semantic reading of Hegel, an approach he shares with other Hegel scholars,7 Brandom seems to factor out all ontological connotations of the term ‘logical’. Here, two questions arise: (a) How does this understanding relate to how Hegel conceives his Logic and (b) how can one, on the basis of this reading, account for Hegel’s distinction between logical concepts (in his use of the term) and the concepts of the Realphilosophien. What exactly, to rephrase the latter question, is “the use Hegel makes of his logical vocabulary in the Realphilosophie” (SPCRH 134)? Brandom gives the following answer to both questions: 5 6
7
The most detailed statement of Brandom’s hermeneutics can be found in chapter 3 of TMD, some comparatively brief remarks in SPCRH, p. 158f. Brandom himself notes that this claim is controversial (see SPCRH 157, footnote 29), but he does not take pains to justify it or discuss alternative views of the relation between these two works. Notably Stekeler-Weithofer 1992.
108
Jana Elisa Falkenroth, Attila Karakuş, David P. Schweikard
“The principal point of his logic is to develop conceptual tools that are necessary and sufficient to express explicitly the essential structures that are implicit in our use of ordinary concepts (including those of the empirical sciences) in judgment and action.” (SPCRH 134)8 We will come back to this point. (ii) Concerning the relationship between Hegel’s Phenomenology and his Logic, what provokes us is Brandom’s (admittedly controversial9) Same-SystemClaim according to which they represent ‘essentially the same system’. The first part of this claim—‘essentially the same’—can be read as saying that the commonalities between these two works are what they are essentially about. So the differences between them, if there are any, are not taken to concern essential aspects of the two works. The last term of this controversial claim— ‘system’—is even more striking, for it is, for instance, one of the main labels Hegel chose for the organization of his own work in his late Encyclopedia. What is more, Hegel did expound his view of the relation between the Phenomenology and the Logic, which can also be understood as his conception of his philosophical system, in the preface to the first edition to the ‘doctrine of being’. There he says of the external relationship between the previous work and the one he is introducing that the original plan was to complement the Phenomenology as the system’s first part with a second part which would have comprised the Logik and the Realwissenschaften. In that plan, which turned out to be infeasible, the Logic would have been ‘the first subsequent step’ [“die erste Folge”] to the Phenomenology. As another example of Hegel’s understanding of this relationship, we would like to quote a passage from one of the introductory texts to the first volume of the Science of Logic in which Hegel speaks of the Phenomenology being presupposed by the Logic. 8
9
Obviously, the term ‘logic’ in this passage is not highlighted as to refer to Hegel’s Logic, i.e. the work published in three volumes between 1812 and 1816, so it might be a misunderstanding to identify this characterization with the characterization Brandom would give of the Logic. But if there were significant differences between what Brandom takes to be ‘his logic’ and what he takes to be ‘his Logic’, we would clearly be entitled to ask for an explanation. To our knowledge, Brandom has not yet given such an explanation. Cf. SPCRH 157, footnote 29.
Comparing Brandom’s Critical Reading to Hegel’s Systematic Enterprise
109
“Es ist in der Einleitung bemerkt, daß die Phänomenologie des Geistes die Wissenschaft des Bewußtseins, die Darstellung davon ist, daß das Bewußtsein den Begriff der Wissenschaft, d. i. das reine Wissen, zum Resultate hat. Die Logik hat insofern die Wissenschaft des erscheinenden Geistes zu ihrer Voraussetzung, welche die Notwendigkeit und damit den Beweis der Wahrheit des Standpunkts, der das reine Wissen ist, wie dessen Vermittlung überhaupt enthält und aufzeigt.” (boldface by us)10 We read this as an indication that Hegel conceived of those two works as containing closely connected but separate philosophical projects. At most, we think, they can be called two closely connected but separate parts of one overarching philosophical project, but not as essentially the same. (iii) Having dwelled perhaps a little too long on the ‘external’ relation between the Phenomenology and the Logic, we should now say more about the projects we take Hegel to be pursuing in those works. Brandom may be right in arguing that both works can be taken to represent essentially the same semantic project—where the semantic is one among further aspects—, but we think they differ essentially in their perspectives and systematic tasks. According to our understanding, the Phenomenology as the ‘Science of the Experience of Consciousness’ is first and foremost an epistemological or, more precisely, a meta-epistemological project, whereas the Logic, and this might meet with more protest, is conceived as the fusion of ontology, understood as the theory about what ‘truly’ or ‘really’ is, and metaphysical theory of subjectivity. Once again, let us refer to Hegel’s own description of the respective tasks as given in the Preface to the Logic: “In der Phänomenologie des Geistes habe ich das Bewußtsein in seiner Fortbewegung von dem ersten unmittelbaren Gegensatz seiner und des Gegenstandes bis zum absoluten Wissen dargestellt. Dieser Weg geht durch 10
See Hegel, Wissenschaft der Logik I, Suhrkamp Bd. 5, S. 67. Of course, one could elaborate on this characterization of the relationship between the two works in terms of the concept of ‘Voraussetzung’ in its reflective logical sense. But it would seem odd to claim that Hegel employs systematic logical vocabulary in an introductory text that is located explicitly outside the argument of the main text.
110
Jana Elisa Falkenroth, Attila Karakuş, David P. Schweikard
alle Formen des Verhältnisses des Bewußtseins zum Objekte durch und hat den Begriff der Wissenschaft zu seinem Resultate.”11 Here, the Phenomenology is characterized as the reconstruction of a path which serves the self-cognition [Selbsterkenntnis] of consciousness. This path is depicted with respect to the consciousness that appears therein, the appearing consciousness [das erscheinende Bewusstsein], and reconstructed from the philosophical narrator’s perspective. It is this constellation of a duality of perspectives that may ground the talk of the development of two sorts of vocabulary: one that is present in judgments of the finite (appearing) consciousness, and one that serves to study the structure of this vocabulary. But calling the elements of the latter vocabulary the description of what Hegel calls logical concepts, seems to strip off the ontological meaning Hegel assigns to them. Here is a passage in which Hegel describes how exactly he intends the first two parts of his Logic, which constitute what he calls ‘objective logic’, to replace former conceptions of ontology, while preserving their systematic task. “Die objektive Logik tritt damit vielmehr an die Stelle der vormaligen Metaphysik, als welche das wissenschaftliche Gebäude über die Welt war, das nur durch Gedanken aufgeführt sein sollte. – Wenn wir auf die letzte Gestalt der Ausbildung dieser Wissenschaft Rücksicht nehmen, so ist [es] erstens unmittelbar die Ontologie, an deren Stelle die objektive Logik tritt, – der Teil jener Metaphysik, der die Natur des Ens überhaupt erforschen sollte; das Ens begreift sowohl Sein als Wesen in sich, für welchen Unterschied unsere Sprache glücklicherweise den verschiedenen Ausdruck gerettet hat.” (boldface by us)12 In a systematically parallel passage of the Encyclopedia he writes: “Die Logik fällt daher mit der Metaphysik zusammen, der Wissenschaft der Dinge in Gedanken gefaßt, welche dafür galten, die Wesenheiten der Dinge auszudrücken.”13 11 12 13
Wissenschaft der Logik, Einleitung, Allgemeiner Begriff der Logik, Suhrkamp Bd. 5, S. 42. Wissenschaft der Logik I: Die Lehre vom Sein, Suhrkamp Bd. 5, S. 61. Hegel, Enzyklopädie der philosophischen Wissenschaften I, Suhrkamp Bd. 8, § 24.
Comparing Brandom’s Critical Reading to Hegel’s Systematic Enterprise
111
This, we take it, makes clear that Hegel’s Logic is dedicated to the development of a network of concepts which express not only the workings of empirical concepts, but also—and, indeed, essentially—the basic structure of both thought and the world. Given the preparatory work of the Phenomenology, distinctions such as that between thought and world or between empirical and logical concepts are not presupposed, but only conceivable as aspects or moments of the structure developed in the Logic. According to Hegel, this task can only be fulfilled on the basis of a presuppositionless, immediate beginning, and the shape it takes is necessarily, to borrow Brandom’s description, “a final, stable, expressively complete set of logical concepts.” (SPCRH 159) As we saw, Brandom rejects the claim that there could even be such a final set of logical concepts and he is silent on Hegel’s ontological claims about logical concepts. In response, we want to urge that taking the ontological dimension of Hegel’s Logic and his system seriously does not commit one to accepting or endorsing his claim about the finality of the set of concepts he developed.
4
AN ALTERNATIVE CRITICAL READING
We have seen that part of Brandom’s program for showing ‘what is living and what is dead’ in Hegel’s thought is based on a view of what Hegel is arguing for in his treatment of empirical and logical concepts. With respect to the latter sort of concepts we have quoted some passages in which Hegel specifies not only the relation between his Phenomenology and Logic, but also the meaning of the term ‘logical’ in sketching the enterprise of the Logic as both an ontology and a (metaphysical) theory of subjectivity. The Logic thus conceived is, for Hegel, the basis of a fundamentally monistic philosophical system whose later, Realphilosophie parts are carried out employing the same categorical framework by way of formulating regional principles (such as the principles of necessity, freedom or the will). Given this brief sketch of Hegel’s systematic enterprise, three groups of questions arise: (1) How does Brandom’s program for a critical reading relate to this systematic enterprise? How much of it is preserved and what aspects are excluded from the appropriation? (2) If Brandom factors out the onto-
112
Jana Elisa Falkenroth, Attila Karakuş, David P. Schweikard
logical dimension in his reading of Hegel’s treatment of logical concepts, does that indicate that he regards the idea of ontological inquiry as outdated? (3) Is there an alternative to Brandom’s reading that deserves to be called ‘critical’? Ad (i): Regarding the first question, we want to urge that Brandom’s treatment of logical concepts as constituting solely an expressive vocabulary leaves out a fundamental claim about the task of the Logic and its role in what Hegel calls ‘the system’. This amounts to ignoring the monism arrived at by the end of the Phenomenology, which forms the basis of Hegel’s philosophical system. Brandom does not inquire into the ontological dimension when he talks about the relations between empirical and logical concepts. But a rendering of Hegel’s conception of logic would have to insist that expounding the structure of logical vocabulary just means expressing the structure of both thought and the world. Here we see, and that is our point, the need for further discussion. Ad (ii): This point is pretty straightforward. For here, our aim of tracing elements of Brandom’s reading of Hegel back to his own conception of the philosophical enterprise cumulates. Even if his work so far centers around the ‘explication of concepts’ and the ‘metaphysics of intentionality’, we think it a fair question to ask whether Brandom’s analytic pragmatism is also a form of quietism in the sense that nothing can be said about the true nature of reality. Ad (iii): Finally, is there a chance of developing a critical perspective on Hegel’s philosophy that preserves his fundamental ontological commitments? We think there is. Some approaches to a systematic reappropriation were focused so far on Hegel’s concept of action (cf. Quante 2004) and his epistemology (cf. Halbig 2002), taking his ontological commitments as informative for his views, instead of elaborating on claims extracted from the ontological context of their statement. This treatment of Hegel’s ideas does not—as if on a slippery slope—amount to the acceptance of Hegel’s most ambitious claims about the nature and reality of the Absolute. But it does seek to exploit Hegel’s views on systematic questions, for instance his combination of objective idealism and direct realism in epistemology (Halbig 2002, ch. 9), for the purpose of furthering contemporary philosophical debate. This could be an alternative critical reading of Hegel, could it not?
Comparing Brandom’s Critical Reading to Hegel’s Systematic Enterprise
113
REFERENCES Brandom, Robert 1994. Making It Explicit: Reasoning, Representing, and Discursive Commitment, Cambridge, Mass.: Harvard University Press, (MIE). Brandom, Robert 2001. “Reason, Expression, and the Philosophical Enterprise”, in: What Is Philosophy?, ed. by C. P. Ragland and Sarah Heidt, New Haven: Yale University Press, pp. 74-95. Brandom, Robert 2002. Tales of the Mighty Dead: Historical Essays in the Metaphysics of Intentionality, Cambridge, Mass.: Harvard University Press, (TMD). Brandom, Robert 2004. “Selbstbewusstsein und Selbst-Konstitution”, in: Hegels Erbe, ed. by Christoph Halbig, Michael Quante, Ludwig Siep, Frankfurt am Main: Suhrkamp, pp. 46-77. Brandom, Robert 2005. “Sketch of a Program for a Critical Reading of Hegel— Comparing Empirical and Logical Concepts”, in: International Yearbook of German Idealism 3, ed. by K. Ameriks & J. Stolzenberg, Berlin, New York: de Gruyter, pp. 131161, (SPCRH). Halbig, Christoph 2002. Objektives Denken – Erkenntnistheorie und Philosophy of Mind in Hegels System, Stuttgart-Bad Cannstadt: Frommann-Holzboog. Hegel, G.W.F. 1970. Enzyklopädie der philosophischen Wissenschaften, Theorie Werkausgabe Bd. 8-10, Frankfurt a. M.: Suhrkamp. Hegel, G.W.F. 1969. Wissenschaft der Logik, Theorie Werkausgabe Bd. 5+6, Frankfurt a. M.: Suhrkamp. Horstmann Rolf-Peter 1999. “What is Hegel’s Legacy, and What Should We Do With It?”, in: European Journal of Philosophy 7:2, pp. 275-287. Quante, Michael 2004. Hegel’s Concept of Action, Cambridge: Cambridge University Press. Rorty, Richard 1997. “Introduction”, in: Wilfrid Sellars, Empiricism and the Philosophy of Mind, ed. by Robert Brandom, Cambridge, Mass.: Harvard University Press, pp. 1-12. Stekeler-Weithofer, Pirmin 1992. Hegels analytische Philosophie – Die Wissenschaft der Logik als kritische Theorie der Bedeutung, Paderborn: Schöningh.
BRANDOM AND FREGE Ulrike Kleemeier, Christian Weidemann
Abstract: According to Robert Brandom a careful look at Gottlob Frege’s early work reveals him as one of the most important ancestors of semantic inferentialism. Our paper offers some—partly complementary, but mainly critical—remarks concerning Brandom’s treatment of Fregean themes. We raise some doubts, whether Frege’s famous context principle should be identified with the Kantian idea of the primacy of judgments and ask critically to what extent Frege can be seen as part of the Kantian tradition in general. We also suggest a new reaction to the so called Caesar problem, which avoids some of the difficulties pointed out by Brandom and others. Finally we are trying to show that it is highly questionable to attribute the doctrine of semantic inferentialism to Frege, even with regard to his early Begriffsschrift. Keywords: context principle, Caesar problem, inferentialism
Besides Kant, Wittgenstein and Sellars, Gottlob Frege is the most frequently mentioned author in Robert Brandom’s seminal work Making it Explicit. Frege not only belongs to Brandom’s rather few “personal heroes”, (the young) Frege is even regarded as no less than one of the ancestors of semantic inferentialism. In what follows, we shall offer some—partly constructive, but mainly critical—remarks concerning Brandom’s treatment of Fregean themes. Especially with regard to Frege’s alleged inferentialism, we wonder whether Frege himself would not have thought of Brandom’s philosophical genealogy as part of a hostile takeover.
1
THE CONTEXT PRINCIPLE
1.1
The Content of the Context Principle
The context principle should not be identified with the Kantian idea of the “primacy of judgments” (e.g. TMD 257). According to Dummett’s analysis,
116
Ulrike Kleemeier, Christian Weidemann
the principle must be divided into two, one regarding sense, the other regarding reference. (A) Context principle about sense: The sense of an expression consists in its contribution to the determination of the senses (truth conditions) of sentences in which it may occur. (B) Context principle about reference: If we have fixed the senses (truth conditions) of a certain range of sentences in which an expression may occur, we have fixed a reference for this expression. In both variations, the context principle expresses not only something more precise but also something different from the idea of the primacy of judgments over concepts. The context principle is primarily a thesis about language and only indirectly one about epistemology. 1.2 The Question of Continuity We share Brandom’s opinion (MIE 81f) that Frege most certainly did not give up the context principle as thesis about sense after Grundlagen. That Frege remains faithful to the context principle can be made highly plausible by looking at the following statements: (1) “Die einfachen oder selbst schon zusammengesetzten Namen nun, aus denen der Name eines Wahrheitswerthes besteht, tragen dazu bei, den Gedanken auszudrücken, und dieser Beitrag des einzelnen ist sein Sinn. Wenn ein Name Theil des Namens eines Wahrheitswerthes ist, so ist der Sinn jenes Namens Theil des Gedankens, den dieser ausdrückt.” (GGA I, 51, § 32) (1893) (2) “Das Zeichen ‚(16 – 2)‘ hat also auch einen Sinn, der ein Teil des Sinnes jenes Satzes ist, und dieser Sinn ist das, was das Zeichen ‚(16 – 2)‘ zum Ausdrucke des Gedankens beiträgt.” (Logik in der Mathematik, NS Bd. I, 249) (1914) (3) “Wir haben in diesem Satz [‚Der Ätna ist höher als der Vesuv‘] den Eigennamen ‚Ätna‘, der zum Sinn des ganzen Satzes, zum Gedanken, einen Beitrag liefert. Dieser Beitrag ist ein Teil des Gedankens, er ist der Sinn des Wortes ‚Ätna‘ ”. (Logik in der Mathematik, NS Bd. I, 250) (1914)
Brandom and Frege
117
(4) “Die andern Sätze (Axiome, Grundgesetze, Lehrsätze) dürfen kein Wort enthalten und kein Zeichen, dessen Sinn und Bedeutung oder dessen Beitrag zum Gedankenausdruck nicht bereits völlig feststände, sodass über den Sinn des Satzes, den darin ausgedrückten Gedanken kein Zweifel ist.” (Brief an Hilbert vom 27.12.1899, NS Bd. II, 62) In all these passages, the context principle about sense is used quite explicitly. Since within Frege’s frame of thought there is no useful distinction that can be drawn between an expression having a completely specific sense and its having a reference, Frege certainly also subscribed continuously to the following principle: (C) The reference of an expression consists in its contribution to the determination of the references of sentences in which it may occur. Regarding principle (B), which is taken by Dummett as the context principle about reference, there may be doubts as to whether even in Grundlagen it was ever seriously advanced by Frege. 1.3
The Alleged Conflict between the Context Principle and the Treatment of Sentences as Proper Names as well as the Conception of Truth Values as Objects
A conflict of this kind is perceived by Dummett and other interpreters of Frege. The argument runs as follows: If sentences are nothing but proper names and truth values are therefore objects, the insight into the specific role of sentences in language is eliminated. Contrary to this position, it can be shown that Frege’s later identification of sentences with singular terms and truth values with objects not only does not contradict the context principle, but follows from this principle in combination with other premises. Consider the following line of Fregean thoughts: 1. An expression has a sense and a reference if it makes a contribution to the determination of the senses and references of sentences in which it may occur. (Context principle) 2. The expression “is true” does not make such a contribution.
118
Ulrike Kleemeier, Christian Weidemann
3. 4. 5. 6. 7. 8. 9.
The expression “is true” is a mock predicate. Truth is not a concept. Truth is something. Everything there is is either a function or an object. Truth must be an object. Truth stands in a specific relationship to sentences. Objects stand to sentences in one of the following three relationships: expression—reference; expression—sense; expression—part of sense. 10. Truth is neither the sense of a sentence nor part of the sense of a sentence. 11. Truth is the reference of a sentence. 12. Objects can only be referred to by proper names. 13. Sentences are proper names for truth values. This whole argument builds, among other things, on the context principle.
2
FREGE AND KANT
It is quite misleading to follow Sluga, as Brandom has done to a great extent (TMD 252-262), in construing Frege as a critical thinker in the tradition of Kant’s philosophy. In Grundlagen, Frege introduces the analytic/synthetic distinction in a completely different way from that of Kant (see GLA 14f, § 3), a fact which Brandom did not notice in MIE (475ff). For Kant, this difference is one between the contents of different judgments. A synthetic judgment is one which enlarges the concept of the subject (Subjektbegriff). An analytic judgment only states what is already contained in the concept. For Frege, the distinction between analytic and synthetic judgments does not concern the contents of judgments, but their way of justification. A sentence is analytic if it can be proven solely by means of logical laws and definitions. Otherwise it is synthetic. Frege does not share Kant’s notion of an analytic judgment (see GLA 92f, § 88). For Kant, analytic judgments cannot be informative. For Frege, it is quite essential that analytic sentences may be informative. In the Grundlagen he tries to show exactly this for arithmetic.
Brandom and Frege
119
Frege strongly objects against Kant’s view that no object can be given to us independently from the senses (GLA 93f, §§ 88f.). Our access to numbers and logical objects in general is not dependent on sensuality, because numbers are not empirical objects. In contradistinction to Kant, Frege declares the axioms of geometry to be objective because they can be expressed and thereby communicated in words (see GLA 40, § 26). On the other hand, Frege believed spatial intuition to be subjective, because it cannot be communicated. Therefore the axioms of geometry can in no way be founded, as Kant claims, on pure intuition. These points already suffice to prevent one from assimilating Kant’s and Frege’s positions. Sluga also refers to the fact that in Grundlagen Frege heavily stresses the notion of reason (Vernunft). But Sluga is wrong to conceive Frege’s notion of reason as a Kantian faculty of knowledge (Erkenntnisvermögen). Frege’s use of the term “reason” in Grundlagen makes it highly probable that its meaning is more or less the same as that of “laws of thought” respectively “laws of logic”. Those laws are conceived as completely objective. It is important to add to these considerations that in Frege’s days it was very much en vogue to use Kantian terminology and express one’s own thoughts within this framework. But by no means did this imply sharing Kant’s positions.
3
THE CAESAR PROBLEM
It is incontestable that Frege did not solve the Caesar Problem, i.e. the question of whether Caesar might be a natural number.1 This problem arises in the Grundlagen as well as in the Grundgesetze. Using the terminology of Grundgesetze, the source of the difficulty is that Axiom V is not sufficient to determine the notion of a course of values (Wertverlauf), respectively the concept of an extension of a concept. But perhaps there is a kind of solution at least within the theoretical framework of the Grundlagen. The idea is roughly as follows: In Grundlagen Frege stresses the fact that number words play a double role in language. At least in ordinary language they often work as predicative 1
See Brandom, “Frege’s Technical Concepts” and “The Significance of Complex Numbers for Frege’s Philosophy of Mathematics” in TMD.
120
Ulrike Kleemeier, Christian Weidemann
expressions as in “There are four Jupiter moons.” In arithmetical language the most prominent sort of sentences are identity statements and therefore number terms here mainly perform the role of singular terms as in “The number of Jupiter moons = four”. Frege eventually decides to treat number words as singular terms and therefore numbers as objects. But in fact there is no necessity for this move. As David Bostock (1974) has shown, number terms can also be construed as second level predicates, respectively quantifiers. In the Grundlagen, Frege is very much aware of the twofold role that number words can play. On the ontological level, this awareness is reflected in Frege’s opinion that extensions of concepts (i.e. the objects numbers are to be identified with) are very special objects. What is special about them is that they are somehow able to “represent” concepts, to “stand in” for them. Following this line of thought, numbers are special objects insofar as they can represent concepts. At this point a central difference between number terms and “ordinary” singular terms comes into play. It is impossible to substitute the name “Julius Caesar” for any number word in a predicative statement. For example, it seems to be impossible to understand the use of the term “Caesar” in the sentence “There are Caesar Jupiter moons”. The point is this: the term “Caesar” can never play a predicative role whereas number words behave differently. Now, in a way the Caesar problem is eliminated, because what distinguishes any number from Caesar is that the latter can never be an object which can “stand in” for a concept. So Caesar can never be a number. To conceive of Caesar as being a number means to commit a category mistake. But this is evident only if one refrains from concentrating solely on the use of number words in the language of arithmetic and instead integrates the role of number terms into ordinary language. Even if the Caesar problem could not be solved, Brandom’s statement does not seem to be true, that this puzzle not only concerns abstract objects (numbers, courses of values etc.), but extends to the whole range of Frege’s theoretical concepts, respectively entities such as sense, reference, thought, truth value. Why, for example, should one not reject the idea that Caesar might be the sense of an expression? For Frege, senses are eternal and immutable objects, but Caesar is neither eternal nor immutable. For similar reasons, the idea can be dismissed that Caesar might be a truth value. Truth values are also eternal and immutable objects. Moreover, truth values are associated by
Brandom and Frege
121
Frege with thoughts respectively sentences, whereas precisely this association is lacking in the case of Julius Caesar. Finally, Frege declares the terms “is true” and “truth” to be indefinable. Now, if truth is not definable, it would seem to be impossible to identify any object with a truth value. Brandom of course ascribes to Frege the view that truth values may be defined in terms of the goodness of inferences. But exactly this idea is not Fregean. On the contrary, Frege tends to define the goodness of inferences in terms of truth values. Frege’s attempt to give us objects through logical means alone may have been unsuccessful, but this failure does not affect all his other fundamental concepts.
4
FREGE AND INFERENTIALISM
Brandom ascribes to (the young) Frege a method which he designates as “substitutional—inferential triangulation”. Brandom: “Two claims have the same conceptual content if and only if they have the same inferential role: a good inference is never turned into a bad one by substituting one for the other.” (TMD 58). This approach to Frege is questionable for several reasons. First, there is no sufficient basis for it in Frege’s texts. Even the Frege of the Begriffsschrift is not an inferentialist thinker in Brandom’s sense. The most important passages of the paragraph Brandom particularly loves to cite are as follows: “die Inhalte von zwei Urtheilen [können] in doppelter Weise verschieden sein […]: erstens so, dass die Folgerungen, die aus dem einen in Verbindung mit bestimmten andern gezogen werden können, immer auch aus dem zweiten in Verbindung mit denselben andern Urtheilen folgen; zweitens so, dass dies nicht der Fall ist. Die beiden Sätze: ‚bei Plataeae siegten die Griechen über die Perser‘ und ‚bei Plataeae wurden die Perser von den Griechen besiegt‘ unterscheiden sich in der erstern Weise. Wenn man nun auch eine geringe Verschiedenheit des Sinnes erkennen kann, so ist doch die Uebereinstimmung überwiegend. Ich nenne nun denjenigen Theil des Inhaltes, der in beiden derselbe ist, den begrifflichen Inhalt.
122
Ulrike Kleemeier, Christian Weidemann
[…] im Urtheile [kommt] hier nur das in Betracht […], was auf die möglichen Folgerungen Einfluss hat. Alles, was für eine richtige Schlussfolge nöthig ist, wird voll ausgedrückt; was aber nicht nöthig ist, wird meistens auch nicht angedeutet; nichts wird dem Errathen überlassen.” (BS § 3, Frege’s emphasis) Indeed, Frege stresses here the important role of inference. But he does not identify the content of a judgment with the inferential role of this judgment. All he says is that his concept script will express everything which is necessary for making inferences and nothing else. The context of the cited passage reveals what exactly is necessary for this purpose. It is the assertion of a conceptual content by the sign “|—” which is translated by Frege in Begriffsschrift as “is a fact”. So Frege says that all that is necessary for drawing correct inferences with a judgment is the assertion of its conceptual content. But he does not say that the content of a judgment should be reduced to its inferential role. Quite the contrary, following Frege’s argument in Begriffsschrift, the inferential role of a judgment consists in assertion and asserting presupposes a content to which it can be applied. The content itself is not explained in inferentialist terms. So what is presented as an inferentialist position by Brandom seems to turn out to be the opposite. Secondly, one should ask whether an inferentialist approach to language and logic is consistent with Frege’s later distinction between sense and reference (a distinction which is latently already present in the Begriffsschrift). The sense/reference-distinction requires distinguishing between synonymy and logical equivalence. Synonymy is a relation between sentences: two sentences are synonymous if and only if they express the same thought. Logical equivalence is a relation concerning the thoughts themselves. Therefore it is possible that two different thoughts can stand to each other in the relation of logical equivalence, whereas it is not possible for two sentences to be synonymous if they express different thoughts. We may conclude that for Frege it is not the same whether the two sentences p and q express the same thought or whether they express different, but logically equivalent thoughts. The case of the two sentences “At Plataea the Greeks defeated the Persians” and “At Plataea the Persians were defeated by the Greeks”, which Frege discusses in Begriffsschrift, is indeed an example of synonymous sen-
Brandom and Frege
123
tences. Quite a different example is that of the sentences “2 + 3 = 5” and “5 = 5”. Frege emphasizes that these statements express different thoughts. On the other hand it is hard to think of an inference where the substitution of “2 + 3 = 5” by “5 = 5” (and vice versa) may turn this inference from a good one into a bad one. Also, the two sentences “p” and “(p or q) and not q” do not express the same thought for Frege. But substituting “p” by “(p or q) and not q” in an inference cannot change that inference from a good one into a bad one. Then, Brandom’s statement “Two claims have the same conceptual content if and only if they have the same inferential role” would be false. In fact, two sentences may have different conceptual contents but the same inferential role, namely when the contents are logically equivalent. It has to be remarked that arithmetic seemed to be an interesting discipline to Frege precisely because “2 + 3” and “5” denote the same number but express different senses. In a similar way, it is essential for logic that logical equivalence may exist between different thoughts.
REFERENCES Bostock, David 1974. Logic and Arithmetic. Natural Numbers, Oxford: Clarendon Press. Brandom, Robert 1994. Making It Explicit: Reasoning, Representing, and Discursive Commitment, Cambridge, Mass.: Harvard University Press, (MIE). Brandom, Robert 2002. Tales of the Mighty Dead: Historical Essays in the Metaphysics of Intentionality, Cambridge, Mass.: Harvard University Press, (TMD). Frege, Gottlob 1879. Begriffsschrift, Halle (reprint, Hildesheim: Olms 1964) (BS). Frege, Gottlob 1884. Grundlagen der Arithmetik, Breslau (reprint, Hamburg: Felix Meiner Verlag 1986) (GLA). Frege, Gottlob 1893/1903. Grundgesetze der Arithmetik, Jena (reprographical reprint, Hildesheim: Olms 1962) (GGA). Frege, Gottlob 1969. Nachgelassene Schriften und wissenschaftlicher Briefwechsel, 2 Bde., Hamburg: Felix Meiner Verlag (NS).
BETWEEN NORMATIVE AND MODAL VOCABULARY A FIRST ENCOUNTER WITH BRANDOM’S LOCKE LECTURES
Bernd Prien
Abstract: In his Locke-Lectures (entitled “Between Saying and Doing”, BSD) Brandom pursues two goals, one broader and the other more specific. The broader goal is to extend the project of classical analytic philosophy (cf. title of lecture one), the narrower one is to provide an analysis of the notion of intentionality. In this paper, I will be concerned with the following aspect of this analysis. In the course of his analysis of intentionality, Brandom claims that certain practices are sufficient to deploy normative vocabulary and that certain other practices are sufficient to deploy modal vocabulary. Since normative and modal vocabulary differ significantly in content, the practices that deploy them must differ significantly too. My worry is that the difference between the practices is not sufficient. Keywords: normative vocabulary, modal vocabulary, meaning, use, pragmatism
1
EXTENDING THE PROJECT OF ANALYSIS
Before I turn to Brandom’s analysis of intentionality, let me say a few words about the way in which he wants to extend the project of classical analytic philosophy. Brandom sees it as the central project of analytic philosophy to establish relations between different vocabularies, that is to say between the contents of different vocabularies. The core programs of analytic philosophy, naturalism and empiricism, can be understood in this way. Brandom wants to extend this project. He thinks that it is a mistake merely to look at semantic relations between vocabularies. He argues that one should also look at relations between vocabularies and practices, i.e. between the meaning and the
126
Bernd Prien
use of expressions. Brandom introduces the following two relations that can obtain between a practice of using expressions and a semantically specified vocabulary: 1. PV-sufficiency: this relation obtains when the use of an expression is sufficient for the expression to have a certain meaning. In other words: The practice is such that it confers a certain content on the expression. The game of giving and asking for reasons is sufficient to confer propositional content in general. More specific practices might prove to be sufficient for the conferral of specific contents. 2. VP-sufficiency: The pragmatist philosopher somehow has to specify the practices she takes to confer content on some vocabulary. Vocabulary that can be used to specify practices is VP-sufficient for those practices. One important question in this context is the following: Can we describe the game of giving and asking for reasons in purely naturalistic terms or do we need normative terms? Apart from these two relations between practices and vocabularies, the relation of PP-sufficiency between practices is important for Brandom: Sometimes one action can be carried out by combining simpler actions in an algorithm. When the actions contained in one practice are in this way sufficient to perform the actions in another practice, the former practice is said to be PPsufficient for the latter.
2
THE QUESTION OF INTENTIONALITY
Brandom’s broader goal in BSD is to urge analytic philosophers not to limit their attention to relations between meanings (i.e. VV-sufficiency or necessity), but to consider relations between meaning and use as well. In the second part of BSD, Brandom applies these new methods in order to provide an analysis of the intentional relation. In the course of his analysis of the intentional relation, Brandom specifies certain practices that are supposed to be PV-sufficient for the deployment of normative vocabulary, a paradigm example of which is the claim that S is committed to do A. He also specifies practices that are supposed to be PV-
Between Normative and Modal Vocabulary
127
sufficient for the deployment of modal vocabulary, a paradigm example of which is the claim that some state of affairs is necessary. I have already said that my worry here is that the two practices are not sufficiently different to explain the semantic differences between the two vocabularies. As the following Meaning-Use-Diagram (MUD) indicates, Brandom claims that there are subpractices of the Game of Giving and Asking for Reasons that are PP-sufficient for practices (the boxes to the far left and right of the diagram) that are in turn PV-sufficient to deploy normative and modal vocabulary. VNorm
VModal V Autonomus Vocabulary
PV-suff
PV-suff
PP-suff
PP-suff PNorm
PSubj-Norm
PObj-Modal
PModal
P Game of Giving and Asking for Reasons Simplified version of the MUD in BSD 6.2
In order to indicate the importance of this question, I will now give a brief sketch of the rough idea of Brandom’s analysis of the intentional relation. Brandom starts with the idea that the game of giving and asking for reasons is necessary to deploy any autonomous vocabulary at all. These practices form, as it where, the center of our use of expressions. Without them, no propositional content could be conferred on expressions at all. It does not matter for my purpose here exactly how this game is played. It will be important, however, that it contains the actions of treating others as committed or entitled to claims and the action of treating claims as incompatible.
128
Bernd Prien
Brandom then goes on to claim that, from these actions, practices can be elaborated that are sufficient to confer normative content and practices that are sufficient to confer modal content. So, if I know how to treat speakers as committed or entitled and I know how to treat claims as incompatible, I know everything I need to know in order to be able to use expressions in a way that would confer normative or modal content on them. This means that anyone who can speak at all can in principle use normative and modal vocabulary. The following consideration connects Brandom’s concern with the normative and the modal vocabulary to the question of intentionality: “The basic idea is that normative vocabulary makes explicit important features of what knowing and acting subjects do when they deploy a vocabulary, when they use expressions so as to say something. And modal vocabulary makes explicit important features both of what is said and of the objective world that is talked about.” (BSD 6.2, p. 7f) The idea is that when I say ‘S is committed to p’, I am saying something about the linguistic practices, whereas when I say ‘It is necessary that p’, I am saying something about the world. For this reason, it can be said “that those features [that can be elaborated so as to deploy normative and modal vocabulary] correspond, respectively, to the subjective and the objective poles of intentional relations.” (BSD 6.2, p. 7)
3
MY CONCERN HERE
In this paper, I will not be concerned with the overall soundness of Brandom’s approach to intentionality. Instead, my worry is this: Brandom claims that certain practices are sufficient for the deployment of normative vocabulary and that certain other practices are sufficient for the deployment of modal vocabulary. Now, as a pragmatist, Brandom is committed to pointing out those differences between these two practices which can account for the fact that normative content is conferred in the one case and modal content in the
Between Normative and Modal Vocabulary
129
other. Brandom rightly stresses that the contents of normative locutions differ sharply from the contents of modal locutions. Consequently, there must be a clear difference between the two practices intended to deploy these vocabularies, a difference that can account for this difference in content. In this regard, I think that BSD is unsatisfactory. As we will see, the practices that Brandom takes to be sufficient to deploy normative vocabulary are indeed sufficient to do so. However, I will argue that the practices that are supposed to deploy modal vocabulary are very closely related to the normative practices. For this reason, they cannot be taken to confer modal content, but must rather be taken to confer normative content as well.
4
THE INTRODUCTION OF NORMATIVE VOCABULARY
I will now turn to the practice of using words that accounts for their expressing normative contents. Brandom starts by arguing that any practice sufficient for an autonomous vocabulary must include the actions of treating speakers as committed or entitled to claims (BSD 4.5, p. 31ff). To introduce normative vocabulary, he extends this practice by introducing the new performance of uttering X. This action is to be performed if and only if one would treat the speaker S as committed to p. This use of the utterance X justifies a semantic interpretation, according to which X means: ‘S is committed to p’. Of course, this account of why X means ‘S is committed to p’ depends on the assumption that there are non-verbal practices that should be interpreted as treating someone as committed or entitled. In MIE, Brandom explains what practices have to look like if they are to be interpreted as treating someone as committed or entitled (one has to be disposed to sanction behavior under various conditions; cf. MIE 3.II.3-5). In BSD, Brandom argues in a more abstract manner that a practice sufficient for the deployment of an autonomous vocabulary must contain practices recognizable as treating someone as committed or entitled. Therefore, he does not have to go into the question of what those practices look like here. In any case, for the sake of argument I will accept that there are practices of treating speakers as committed/entitled. These practices can be said to be
130
Bernd Prien
practically directed towards our being committed or entitled.1 On this assumption, it seems reasonable to say that an utterance X that is keyed to these practices via response substitution should be interpreted as meaning ‘S is committed/entitled to p’.
5
THE INTRODUCTION OF MODAL VOCABULARY
As I said at the outset, my worry is that the practices that are supposed to be sufficient for modal vocabulary are so closely related to the practices that are sufficient for normative vocabulary that it becomes difficult to see why these two practices should confer entirely different contents. I now want to make this relatedness visible by describing the practices that are supposed to deploy modal vocabulary. Unfortunately, Brandom is not quite as explicit about them as he is about the normative practices. However, I think his position can be reconstructed thus: In lecture four, Brandom argues that, in order to grasp any concepts at all, one has to associate with each inference a range of counterfactual robustness. For each inference, one has to have an idea of which collateral premises would defeat it and which ones would leave it intact. One can now introduce modal vocabulary because, when a speaker associates a range of robustness with an inference from p to q, she takes q to be necessary if p, at least within a certain range of collateral assumptions. Brandom elaborates on this rough idea in lecture five. As the MUD at the beginning of BSD 5.5 indicates, Brandom starts from the practice of treating speakers as committed or entitled. This is the same practice as the one we have discussed in the previous section: the practice that can be elaborated so as to include the expression ‘S is committed/entitled to p’. From the practices of treating speakers as committed or entitled we can algorithmically elaborate the practice of treating claims as incompatible: To treat two claims p and q as incompatible is to withhold attribution of an entitlement to p (which one would otherwise attribute) because the target of attribution is committed to q. Suppose I would normally attribute entitlement to p 1
For the concept of practical direction toward something cf. BSD 6.2, p. 10.
Between Normative and Modal Vocabulary
131
to a speaker because she is entitled to r. If I then refrained from attributing entitlement to p because the speaker is committed to q, I would be treating p and q as incompatible. So, the action of (non-verbally) treating two claims as incompatible is a combination of already existing actions in a conditional algorithm.
8: VV-suff
(3)
V Modal Logical: N, K, L
VIncompat V Autonomous 7: PV-suff
(2)
2: PV-nec
P Incompat
4: PP-suff
(1)
P Com/Ent
3: PP-suff
P
Giving and Asking for
Reasons
5: PV-suff
PADP
V Norm: Com/Ent
Simplified version of the MUD in BSD 5.5. The steps considered here concern the boxes (1)-(3) on the left.
We can then further extend these practices by introducing another new action, namely uttering the words ‘p and q stand in relation R’. This expression
132
Bernd Prien
is to be used if and only if one were to treat the two commitments p and q as incompatible. This last step in the elaboration is of the same type as the one that led to normative vocabulary: We simply extend the practices by introducing an utterance as a new action and connect that new action to the existing practices by a norm. Brandom calls this type of elaboration ‘response substitution’. The question now is how this newly introduced utterance should be interpreted. In choosing the expression ‘p and q stand in relation R’, I have already indicated that I think it should be interpreted as claiming a relation between two claims. But the crucial question here is: what kind of relation? Does the use of this expression, as I have described it, vindicate interpreting the utterance as ‘It is impossible that p and q both obtain’? In that case, the newly introduced utterance would belong to the modal vocabulary. Or should we rather interpret the utterance as ‘One should not be committed both to p and q’, as saying that speakers are not permitted to endorse both claims? In that case, the newly introduced utterance would belong to the normative vocabulary. Brandom himself distinguishes between incompatibility in a subjective sense and in an objective sense.2 In the subjective sense incompatibility means that speakers should not be committed to both p and q, while in the objective sense it means that it is impossible that p and q both obtain. So, in the first, subjective sense, incompatibility is a normative notion, while in the second, objective sense, it is a modal notion. Using this distinction, the question I have just raised can be formulated thus: Should R in ‘p and q stand in relation R’ be interpreted as meaning incompatibility in the objective or in the subjective sense? To answer this question, we have to remember that the practices of treating speakers as committed or entitled are practically directed toward proprieties of actions. This assumption was crucial for the introduction of normative vocabulary. Now, it is clear that the algorithmic elaboration that yields the action of treating two claims as incompatible cannot change this—this action is still practically directed toward proprieties of linguistic actions and not 2
Cf. “Holism and Idealism in Hegel’s Phenomenology”, TMD 182, TMD 196, and BSD 6.4.
Between Normative and Modal Vocabulary
133
to possibilities or necessities of facts. The same goes, of course, for the utterance ‘p and q stand in relation R’ that is keyed to this action. Consequently, R should be interpreted as meaning ‘incompatible’ in the subjective sense and not in the objective sense. The newly introduced utterance means ‘Speakers should not be committed to both p and q’ and not ‘p and q cannot both be the case’ as Brandom claims.3
6
CONCLUDING REMARKS
My argument here depends on the pragmatist principle that different semantic properties of expressions (different contents) have to be explained by different uses of the expressions.4 At the end of lecture six (BSD 6.4, p. 24), there is an MUD that seems to contradict this principle. According to that MUD, the practice of rational rectification is PV-sufficient both for the vocabulary of modal incompatibility and the vocabulary of normative incompatibility.
V Normative
V Modal
Incompatibility
Incompatibility
PV-suff
PV-suff P Rational Rectification
MUD from BSD 6.4 3
4
That Brandom takes VIncompat in the above MUD to belong to the modal vocabulary is clear because he says that it is the objective sense of incompatibility that is “used as the basis of a formal semantics” (BSD 6.4, p. 20). Semantic pragmatism is the view that “whenever [the semantic theorist] associates with expressions some semantically relevant whatsis as its content or meaning, she undertakes an obligation to explain what it is about the use of that expression that establishes in practice the association between it and the semantically relevant whatsis.” (PaP 43f)
134
Bernd Prien
For this brief remark, it is not important what exactly rational rectification is. I just want to note that I do not see how a pragmatist can claim that one and the same practice is sufficient for both the normative and the modal notion of incompatibility which, as Brandom rightly emphasizes “are clearly different notions of incompatibility” (BSD 6.4, p. 21). My second concluding remark is that there seems to be a discrepancy between what Brandom says in the text and what the MUD says. Brandom writes: “But in practically acknowledging an obligation to rectify or repair a set of commitments, one is doing something that can be specified not just by using one or the other, but, crucially, by using both. [notions of incompatibility].” (BSD 6.4, p. 23) So, the text speaks of a VP-sufficiency relation: What I am doing when I am rectifying commitments can be specified in two ways, i.e., by using modal vocabulary and by using normative vocabulary. Given what ‘specify’ means in BSD, I would concede this. According to the MUD, however, relations of PV-sufficiency and not of VP-sufficiency obtain.
REFERENCES Brandom, Robert 1994. Making It Explicit: Reasoning, Representing, and Discursive Commitment, Cambridge, Mass.: Harvard University Press, (MIE). Brandom, Robert 2002. Tales of the Mighty Dead: Historical Essays in the Metaphysics of Intentionality, Cambridge, Mass.: Harvard University Press, (TMD). Brandom, Robert 2002b. “Pragmatics and Pragmatisms”, in: Hilary Putnam: Pragmatism and Realism, ed. by James Conant and Urszula M. Zeglen, London: Routledge, pp. 40-59, (PaP) Brandom, Robert 2006. Between Saying and Doing: Towards an Analytic Pragmatism, Locke Lectures held at Oxford University, Trinity Term 2006, to appear 2008 from Oxford University Press. Unpublished Manuscript at http://www.pitt.edu/~brandom/ locke/index.html, (BSD).
HOW TO KRIPKE BRANDOM’S NOTION OF NECESSITY
Benedikt Paul Göcke, Martin Pleitz and Hanno von Wulfen
Abstract: In this paper we discuss Brandom’s definition of necessity, which is part of the incompatibility semantics he develops in his fifth John Locke Lecture. By comparing incompatibility semantics to standard Kripkean possible worlds semantics for modality, we motivate an alternative definition of necessity in Brandom’s own terms. Our investigation of this alternative necessity will show that— contrary to Brandom’s own results—incompatibility semantics does not necessarily lead to the notion of necessity of the modal logic S5. Keywords: incompatibility semantics, modal logic, necessity, possible worlds, linguistic ersatzism
1
INTRODUCTION
In his fifth John Locke Lecture “Incompatibility, Modal Semantics and Intrinsic Logic” (BSD 5) Brandom picks up a project he had already started in his 1985 text “Varieties of Understanding”1: developing a formal semantics which fits his inferentialist theory of meaning. This formal system is based on the single concept of incompatibility between sentences. According to Brandom, incompatibility is symmetric and persistent. It is symmetric since if p is incompatible with q, then q is incompatible with p. It is persistent since we cannot obliterate the incompatibility of two sentences by adding a third one. The propositional content of a sentence is represented by the set of sentences incompatible with it (cf. BSD 5.3, p. 8). Thus, the meaning of any sentence is dependent on the meaning of every other sentence—just as it should be in 1
Brandom also mentions this project in his Making it Explicit and Articulating Reasons.
136
Benedikt Paul Göcke, Martin Pleitz, Hanno von Wulfen
Brandom’s inferentialist semantics. One sentence is said to (incompatibility-) entail another just as everything incompatible with the second is incompatible with the first. The basic notion of incompatibility and the defined notions of propositional content and entailment are the framework of incompatibility semantics. In this framework, Brandom goes on to define the operators of propositional logic (negation, conjunction etc.) and the alethic modalities of necessity and possibility. In what follows, we will focus on the notion of necessity. In incompatibility semantics, the meaning of “necessarily p” has to be given as the set of sentences incompatible with it. Therefore Brandom’s definition of necessity and later our own alternative definition take the form of a characterization of those sentences which are incompatible with “necessarily p”: “To be incompatible with necessarily-p is to be (self-incompatible or) compatible with something that does not entail p. For anything compatible with something that does not entail p is compatible with something that does not necessitate p, and so leaves open the possibility that p is not necessary.” (BSD 5.4, p. 17) To show that this definition does indeed capture the intuitive notion of necessity, Brandom proves (in the extensive appendix to the fifth lecture) that this definition implies that what is necessarily the case is actually the case, that there are contingent truths and other theorems commonly held to be characteristic of necessity. However, something astonishing happens: Brandom’s incompatibility modal logic turns out to be equivalent to S5. And Brandom does not seem to be too happy with this. “This is the system in which it is true both that whatever is necessary is necessarily necessary and that whatever is possible is necessarily possible. In the usual Kripke semantics, this is the modal logic generated by accessibility relations among possible worlds that are reflexive, symmetric, and transitive. In the tangled jungle of modal-logical systems, this is the unexciting, well-studied, well-behaved, plain-vanilla modal analogue of the classical non-modal propositional calculus.” (BSD 5.4, p. 17f, emphasis added)
How To Kripke Brandom’s Notion of Necessity
137
The goal pursued in this paper is to show that incompatibility semantics does not necessarily lead to the unexciting modal logic of S5. Starting from Kripkean intuitions about the semantics of modality, we develop an alternative but still incompatibilistic notion of necessity which—in “the tangled jungle of modal-logical systems” (BSD 5.4, p. 18)—gives us the modal logic of K, T, and B, but neither of S4 nor of S5. We hope that, apart from the formal details, our investigation will shed some light on Brandom’s project of incompatibility semantics and the notion of necessity in general.
2
AN ALTERNATIVE DEFINITION OF NECESSITY
It is characteristic of Brandom’s definition of necessity that he does not need additional primitives like possible worlds, accessibility between possible worlds and truth in a possible world, which in Kripke-semantics for modal logic are essential to get started at all. In Brandom’s eyes this is an advantage because in contrast to possible worlds semantics “semantics done in terms of incompatibility is directly modal.” (BSD 5.3, p. 11) For the sake of argument, we accept this scepticism about possible worlds conceived as basic entities. Nevertheless, we think it promising to contrast Brandom’s approach to necessity with the Kripkean approach based on possible worlds. A first tentative observation on Brandom’s definition is the following: Whereas Kripke semantics fixes the truth condition of ‘necessarily p’ in a particular possible world w in a local way as the truth of p in all worlds accessible from w, Brandom defines necessity in a global way by talking about the entirety of all sentences at once. Thus, Brandom’s account lacks anything like a particular perspective from which a formula like ‘necessarily p’ is evaluated. As S5 is the only modal logic in which Kripke’s locality intuition can be ignored (as necessity in S5 is simply truth in all possible worlds), this first observation gives raise to the suspicion that Brandom presupposes S5 from the very beginning by defining necessity globally, while ignoring the locality that can be given by the framework of possible worlds and accessibility. But as we—like Brandom—do not want to take possible worlds as basic entities, we need to define the concept of a possible world. In the philosophy of modality, an attempt has been made to represent possible worlds as maxi-
138
Benedikt Paul Göcke, Martin Pleitz, Hanno von Wulfen
mally consistent propositions (for instance by Plantinga, Adams and Stalnaker (cf. Plantinga 1974, Stalnaker 1979)). In incompatibility semantics, this would amount to representing a possible world by a maximally compatible set of sentences: A set such that every sentence either is included or is incompatible with it. This is in accordance with Plantinga’s characterization of a possible world: “Let us say that a state of affairs S includes a state of affairs S’ if it is not possible […] that S obtain and S’ fail to obtain […] a state of affairs S precludes a state of affairs S’ if it is not possible that both obtain […] a state of affairs S is complete or maximal if for every state of affairs S’, S includes S’ or S precludes S’. And a possible world is simply a possible state of affairs that is maximal.” (Plantinga 1974, 44f) To obtain a full framework for possible worlds, we have to define the concepts of truth-in-a-world and accessibility in terms of incompatibility. Possible World: The set of sentences W is a possible world iff W is coherent and for any set of sentences X: W⎥= X or W∪X is incoherent. Truth-in-a-world: The sentence p is true in the possible world W iff W⎥= p. Accessibility: The possible world W2 is accessible from the possible world W1 iff for every p satisfying W2⎟= p there is an X with W1⎥= X so that X∪{p} is coherent. That truth-in-a-world as defined in terms of entailment is implied in the intuitive understanding of our Plantinga-inspired definition of possible worlds. Thus, with respect to a possible world, any sentence is either true in it or incompatible with it. With regard to our notion of accessibility we can again build on Plantinga. He states that “to say that W is possible relative to W’ is to say that W would have been possible if W’ had been actual; alternatively, it is to say that every propo-
How To Kripke Brandom’s Notion of Necessity
139
sition true in W is possible in W’, or that every state of affairs that obtains in W is possible in W’.” (Plantinga 1974, 51f) In our definitions we only use Brandom’s concepts of the incompatibility of sentences and of incompatibility entailment. We are thus able to represent the framework of possible worlds in incompatibility semantics. *** Let us start with Kripke Brandom’s notion of necessity and present an alternative incompatibilistic definition of necessity—one which captures what we have called Kripke’s locality intuition about modality and therefore does not presuppose the modal logic of S5 right from the start. To motivate our alternative definition of necessity, we will analyse “q is incompatible with necessarily p” in standard Kripke semantics. Let us start by formulating the plain incompatibility of q and p in Kripke semantics. As Brandom himself notes, the incompatibility of q and p is the impossibility of the conjunction q and p (“non-compossiblity”) (cf. BSD 5.4, p. 13). Therefore: 1 2 3
q incompatible with p ⇔ ¬M(q ∧ p) ⇔ ¬∃W (q true in W ∧ p true in W)
That is, q and p are incompatible if there is no possible world where q and p both are true. Uniform substitution of “Lp” for “p” gives us the starting point for a chain of equivalent formulae leading to our alternative definition of necessity: 4 5 6
q incompatible with Lp ⇔ ¬∃W (q true in W ∧ Lp true in w) ⇔ ∀w (q true in W ⇒ Lp false in W)
q is incompatible with necessarily p. There is no world where both q and necessarily p are true. In all possible worlds the truth of q implies the falsity of necessarily p.
140
7
8
9
Benedikt Paul Göcke, Martin Pleitz, Hanno von Wulfen
⇔ ∀W (q false in W) ∨ ∀W (q true in W ⇒ Lp
q is false in all possible worlds or the truth of q implies the falsity of necessarily p in all possible worlds.
false in W)) q is false in all possible worlds or for every ⇔ ∀W (q false in W) ∨ ∀W (q true in W ⇒ ∃W’ q-world there is an accessible non-p-world. (W’ accessible from W ∧ p false in W’))) ⇔ q self-incompatible ∨ ∀W1 (W1 maximally coherent ∧ W1⎥= q ⇒ ∃W2 (W2 maximally coherent ∧ W2 is accessible from W1 ∧ W2⎟≠ p))
The last line in this chain of equivalent formulae gives us our alternative definition of necessity: Brandom’s definition of necessity: q is incompatible with Lp :⇔ q self-incompatible ∨ ∃Y ({q}∪Y coherent ∧ Y⎟≠ p) Alternative definition of necessity: q is incompatible with Lp :⇔ q self-incompatible
∨ ∀W1 (W1 maximally coherent ∧ W1⎥= q ⇒ ∃W2 (W2 maximally coherent ∧ W2 is accessible from W1 ∧ W2⎟≠ p))
It says that for q to be incompatible with necessarily p, q has to be selfincompatible or q has to be such that for every maximally coherent set of sentences entailing q there is another maximally coherent set of sentences accessible (as defined by us) from the first, such that the second does not entail p. Our alternative definition has some similarity in structure to Brandom’s definition of necessity: The first adjunct is the same and the second requires the existence of a set of sentences somehow connected to q that does not entail p.
How To Kripke Brandom’s Notion of Necessity
141
But there are differences, most notably those resulting from the relation of accessibility. Although the alternative definition is still recognizably Kripkean, it only contains concepts defined in incompatibility semantics. Therefore it is on a par with Brandom’s definition of necessity: Both definitions are candidates for the formalisation of the intuitive notion (or notions) of necessity. Brandom has already shown that his definition of necessity validates all axioms of S5. In the remaining part of this paper we shall show that our alternative notion of necessity validates the axioms of K, T, and B, but not of S4 and therefore not of S5.
3
PROPERTIES OF ALTERNATIVE NECESSITY
Let us say in advance that as we do not have enough time for a proper formal treatment, we are presenting the results of the formal work in an informal way. We have prepared an appendix containing all relevant definitions and proofs. The starting point for our formal investigation is Brandom’s incompatibility semantics and his treatment of the propositional calculus. In this framework we have already defined the concepts of possible world, truth in a world, accessibility and our alternative notion of necessity. We now want to know which axioms of modal logic are validated by our definition of necessity. Although all our proofs are accomplished in the set-theoretical metalanguage of incompatibility semantics, we are, in most cases, guided by Kripkean intuitions. The idea behind proving the validity of the T-axiom is that in Kripke semantics the T-axiom is validated by a reflexive accessibility relation between possible worlds. Analogously, the B-axiom is connected to the symmetry of accessibility and the S4-axiom to the transitivity of accessibility. Therefore the main point in our proofs is to show whether the accessibility relation between possible worlds, as defined in incompatibility semantics, satisfies the relevant property of reflexivity, symmetry or transitivity.
142
Benedikt Paul Göcke, Martin Pleitz, Hanno von Wulfen
3.1 The T-axiom Lp → p is valid We start by showing that accessibility, as we defined it, is reflexive. Remember that, given two maximally coherent sets (i.e. possible worlds) W1 and W2, W2 is accessible from W1 just in case every p entailed by W2 is compatible with some sentences entailed by W1. Now let W be a maximally coherent set of sentences. Then every sentence entailed by W is self-compatible, as otherwise W could not be coherent (given the definition of incompatibility-entailment). So for every p entailed by W we find some sentence entailed by W with which it is compatible, namely p itself. Therefore W is accessible from W. The relation of accessibility is reflexive. To show the validity of the T-axiom Lp → p it suffices to show that necessarily p incompatibility-entails p, i.e. Lp⎥= p. According to Brandom’s definition of incompatibility-entailment, we therefore have to show that every q incompatible with p is incompatible with necessarily p. Let some arbitrarily chosen q be incompatible with p. Then either q is self-incompatible or q is compatible with itself but incompatible with p. Let us take a quick look at the first case in which q is self-incompatible. Then, by the first adjunct of the alternative definition of necessity, q is incompatible with necessarily p. The second case, in which q is compatible with itself but incompatible with p, is more interesting. According to the alternative definition of necessity we now have to show that q satisfies the second adjunct of the alternative definition of necessity (from now on called “A2”): A2: ∀W1 (W1 maximally coherent ∧ W1⎥= q ⇒ ∃W2 (W2 maximally coherent ∧ W2 is accessible from W1 ∧ W2⎟≠ p)) The best way to prove this point is by reductio. Thus, suppose A2 is false. Predicate logic, then, tells us that the following is true: non A2: ∃W1 (W1 maximally coherent ∧ W1⎥= q ∧ ∀W2 (W2 maximally coherent ∧ W2 is accessible from W1 ⇒ W2⎟= p))
How To Kripke Brandom’s Notion of Necessity
143
In other words: The second adjunct is false if and only if there is a q-world such that all accessible worlds are p-worlds. As we have already shown that the accessibility relation is reflexive, we can conclude that W1 is one of the worlds accessible from W1. We obtain:
∃W1 (W1 maximally coherent ∧ W1⎥= q ∧ W1⎟= p) That is, p and q are both true in the possible world W1. But as possible worlds are maximally coherent sets of sentences, q must be compatible with p, which (of course) contradicts q’s being incompatible with p. Thus, we have shown that every q which is incompatible with p is also incompatible with necessarily p: We have shown the T-axiom to be valid. 3.2
The K-axiom L(p → q) → (Lp → Lq) is valid
On the basis of the T-axiom, we can show that incompatibility semantics validates the K-axiom, i.e. the theorem that is an axiom of every modal logic (cf. the appendix). 3.3 The B-axiom p → LMp is valid The B-axiom says that if something is the case then it is necessary that it is possible. In possible worlds semantics, the B-axiom is valid in every frame where accessibility between possible worlds is symmetric. The B-axiom can be shown to be valid in incompatibility semantics, as well. But let us for now stick to a remark about why accessibility is symmetrical. Possible worlds are maximally coherent sets of sentences. World 2 is accessible by world 1 just in case every proposition entailed by world 2 is compatible with some proposition entailed by world 1. In particular, the proposition that gives a complete description of world 2 is surely entailed by world 2.2 Therefore, if world 2 is accessible by world 1, there must be a proposition entailed by world 1 that is compatible with the proposition completely de2
As we are operating with a language closed under the operations of propositional calculus, we will, for every set of sentences, have a proposition incompatibilityequivalent to this set.
144
Benedikt Paul Göcke, Martin Pleitz, Hanno von Wulfen
scribing world 2. As the proposition completely describing world 2 is maximally consistent, every proposition compatible with it must be a conjunct of it. Therefore, if world 2 is accessible by world 1, then there must be a proposition that is entailed by both. Speaking figuratively, accessibility is only given between possible worlds that overlap. But overlapping—the sharing of at least one element—is a symmetrical situation. In incompatibility semantics, accessibility between possible worlds is a symmetrical relation. 3.4
The S4-axiom Lp → LLp is not valid
Let us now turn to the S4 axiom, which says that what is necessary is necessarily necessary. To prove that this axiom is not valid we have to find a counter-model. We take our clue from Brandom’s blackberry example, which describes a situation of pairwise compatible but in toto incompatible sentences. There are three sentences—B is ripe, B is a blackberry, B is red—which are pairwise compatible but altogether incompatible, as blackberries are ripe when they are black and not when they are red. We try to find a similar model contradicting the S4-axiom. Thus, consider a language consisting only of four atomic sentences (p, q, r and s) which are pairwise compatible (apart from the “diagonal” pairs p–r and q–s) but every set containing three or four of these sentences is not. By our definition of ‘possible world’ we thus obtain a model with four possible worlds: world 1 = {p, q}, world 2 = {q, r}, world 3 = {r, s}, world 4 = {p, s}. All other sets of sentences are either incoherent or not maximally coherent. According to our definition of accessibility, all possible worlds sharing at least one element are accessible to each other. We obtain, for instance, the following accessibility relations: From world 1 we reach world 2, from world 2 we reach world 3, but from world 1 we do not reach world 3. Hence, the accessibility relation is not transitive, as the following diagramm shows: v, Lv, not: LLv
v world 1: {p, q} ⏐ world 4: {p, s}
v
⎯ ⎯
world 2: {q, r} ⏐ world 3: {r, s} not: v
How To Kripke Brandom’s Notion of Necessity
145
This is the possible worlds situation we were looking for. In order to prove that the S4-axiom is not valid, we have to show that in this model some possible world, for instance world 1, entails the necessity of a certain proposition, but does not entail the necessary necessity of that proposition. Hence, we need to find a proposition entailed by world 1, by world 2 and by world 4 (i.e. by every world accessible from world 1) but not entailed by world 3 (i.e. not entailed by the only world in our model not accessible from world 1). This proposition, let us call it v, turns out to be the proposition p∨q. p∨q is true in every world in which one of the two adjuncts is true, but false in world 3, i.e. in the possible world {r, s}. (All this can be shown formally by considering the incompatibility sets of all these propositions.) {p, q}⎥= {v}, hence v is true in world 1 {q, r} ⎥= {v}, hence v is true in world 2 {r, s} ⎥≠ {v}, hence v is false in world 3 {p, s} ⎥= {v}, hence v is true in world 4 If this were standard possible worlds semantics, in world 1 we would have the necessity of v, but not the necessary necessity of v. But as we have to conduct our proof in incompatiblity semantics, we now have to show that world 1 is compatible with Lv but incompatible with LLv. Thus, the first thing we have to show is that the proposition world 1 is compatible with the proposition Lv. Remember the alternative definition of necessity: Alternative definition of necessity: q is incompatible with Lp :⇔ q self-incompatible
∨ ∀W1 (W1 maximally coherent ∧ W1⎥= q ⇒ ∃W2 (W2 maximally coherent ∧ W2 is accessible from W1 ∧ W2⎟≠ p))
As we want to show that world 1 is compatible with Lv, we want the following not to be the case:
146
Benedikt Paul Göcke, Martin Pleitz, Hanno von Wulfen
world 1 self-incompatible ∨ ∀W (W maximally coherent ∧ W⎥= world 1 ⇒ ∃W’ (W’ maximally coherent ∧ W’ is accessible from world 1 ∧ W’⎟≠ v)) As world 1 is coherent, we only have to check the second adjunct. As world 1 is maximally coherent, it is only entailed by itself. Therefore the second adjunct in this case is equivalent to
∃W’ (W’ maximally coherent ∧ W’ is accessible from world 1 ∧ W’⎟≠ v) call this (*) Beside world 1 itself, only world 2 and world 4, i.e. {q, r} and {p, s}, are maximally coherent and accessible from world 1, but both entail v. There is no maximally coherent set accessible from world 1 that does not entail v. Therefore (*) is false and world 1 is compatible with Lv according to the alternative definition of necessity. In a similar way we can show that the proposition world 1 is incompatible with LLv (cf. the appendix). Hence, our model of a language consisting only of four pairwise compatible sentences proves that the S4-axiom Lp → LLp is not valid. As this formula is also a theorem of the modal logic S5, this in turn shows that alternative necessity does not lead to S5, either.
4
CONCLUSION
Let us sum up what we hope to have shown. Our first step was to represent the Kripkean concept of possible worlds in Brandom’s incompatibility semantics. This led to an alternative definition of necessity in Brandom’s own terms of incompatibility. In a second step, we have shown that—in contradistinction to Brandom’s definition of necessity—this alternative definition does not automatically generate the modal system of S5. The reason is that although K, T, and B are valid, S4 is not valid and therefore S5 cannot be valid, either.
How To Kripke Brandom’s Notion of Necessity
147
We want to close with some remarks about the philosophical assessment of these formal results. What are the implications for Brandom’s project of inferentialist semantics and its formalisation in incompatibility semantics? We want to emphasize that we do not see our results as a problem for Brandom’s general account. Rather, the opposite is the case: Deploying our alternative definition of necessity, incompatibility semantics is not doomed to result in the “plain vanilla” system S5. The alternative definition of necessity enables Brandom to choose between the modal logic of S5 and that of B. As in the case of standard possible worlds semantics, incompatibility semantics provides more than one system. This should be a pleasant result for Brandom. Of course, Brandom’s choice is still quite restricted: Incompatibility semantics validates only those modal logical systems that contain the theorems of B. But we do not think that this restriction of modal logic is a drawback to incompatibility semantics. Brandom’s overall project of a pragmatist and inferentialist theory of meaning gives good reasons to prefer propositions to possible worlds as the basic entities of semantics. As a consequence, the semantics of modality can only rely on possible worlds that are not basic entities, but constructed from propositions. Our formal result was that the natural way to do this—representing possible worlds by maximally coherent sets of propositions—leads to the restriction of modal logic to systems containing B. Our philosophical conjecture about this result is that as reasons to prefer propositions to possible worlds in the theory of meaning commit us to this particular system of modal logic, this theory of meaning gives us valuable insight into the notion of necessity.
REFERENCES Robert Brandom 1985. “Varieties of Understanding”, in: Reason and Rationality in Natural Science, ed. by Nicholas Rescher, Lanham/London: University Press of America, pp. 27-51. Brandom, Robert 2006. Between Saying and Doing: Towards an Analytic Pragmatism, Locke Lectures held at Oxford University, Trinity Term 2006, to appear 2008 from Oxford University Press. Unpublished Manuscript at http://www.pitt.edu/~brandom/ locke/index.html, (BSD).
148
Benedikt Paul Göcke, Martin Pleitz, Hanno von Wulfen
Cresswell, M.J./Hughes, G.E. 1996. A New Introduction to Modal Logic, London/New York: Routledge. Plantinga, Alvin 1974. The Nature of Necessity, Oxford: Clarendon. Stalnaker, Robert C. 1979. “Possible Worlds”, in: The Possible and the Actual, ed. by Michael J. Loux, Ithaca/London: Cornell University Press.
APPENDIX The appendix presents the formal work our paper is based on. First we present those definitions, axioms and theorems of Brandom’s incompatibility semantics that we will need for our investigation of the notion of necessity. Secondly, we motivate our alternative definition of necessity in the formal framework of incompatibility semantics. Thirdly, we characterize the properties of alternative necessity by showing that it validates the axioms of T, K, and B, but not of S4 (and hence not of S5).
1
Incompatibility Semantics
Our formal investigation is an investigation within incompatibility semantics as developed by Brandom. We want to present an alternative definition of the concept of necessity in terms of incompatibility. Therefore we start with those definitions, axioms and theorems of incompatibility semantics that do not presuppose Brandom’s definition of necessity. What we need is the following: The concepts of incoherence and incompatibility, the definition of incompatibility entailment, the definition of the content of a sentence, the introduction of the operators of the propositional calculus and some theorems regarding entailment and implication, which can be found in the appendix of the fifth John Locke Lecture.
2
An Alternative Definition of Necessity
We first give definitions of the concepts of a possible world and accessibility motivated in the paper, show some basic properties of possible worlds and then move on to an alternative definition of necessity.
2.1
Definitions
Possible World The set of sentences W⊆L is maximally coherent (in other words: it represents a possible world) iff W is coherent and for any set of sentences X⊆L: W⎥= X or W∪X is incoherent.
How To Kripke Brandom’s Notion of Necessity
149
Truth-in-a-world The proposition p is true in the possible world W iff W⎥= p. The proposition p is false in the possible world W iff W⎥≠ p. Possibility-in-a-world The proposition p is possible in the possible world W iff there is a non-empty X with W⎥= X so that X∪{p} is coherent. Preliminary Definition of Accessibility The possible world W2 is accessible by the possible world W1 iff for every proposition p true in W2 p is possible in W1. Spelling out the preliminary definition of accessibility, we now get the following characterization of the accessibility relation: Accessibility The possible world W2 is accessible by the possible world W1 iff for every p satisfying W2⎟= p there is an X with W1⎥= X so that X∪{p} is coherent.
2.2
Some Basic Properties of Possible Worlds
Coherence Lemma Every set of sentences entailed by a possible world is coherent: ∀W ∀X (W max. coh. ∧ W⎥= X ⇒ X coh.) Proof: Suppose not. Then we have a possible world W and a set of sentences X with W ⎥= X and X incoherent. In this situation X is incompatible with every set of sentences. As W is incompatible with everything X is incompatible with, it has to be incoherent, too. This contradicts W’s being a possible world. Possibility Lemma If a proposition is true in a possible world then it is possible in this world: ∀W ∀p (W max. coh. ∧ p true in W ⇒ p possible in W) Proof: As everything entailed by a possible world is coherent, every proposition true in a possible world is self-compatible. Therefore there is for every proposition p entailed by a world a set of sentences entailed by this world namely the set {p} that is compatible with p. Either-Or Lemma Given a possible world, each set of sentences is either entailed by it or incompatible with it: ∀W ∀X (W max. coh. ⇒ (W⎥= X ⇔ W∪X coh.))
150
Benedikt Paul Göcke, Martin Pleitz, Hanno von Wulfen
Proof: Suppose not. Then we have a possible world W and an X such that W⎥= X and W∪X is incoherent. (The definition of possible world does not prima facie forbid this situation.) We also have W⎥= W, as every proposition incompatibility-entails itself. From W⎥= X and W⎥= W we get W⎥= W∪X (as we are operating in the language closed under the operations of propositional logic). As W∪X is supposed to be incoherent this entailment implies that W is incoherent, which contradicts its being a possible world. Membership Lemma Each set of sentences X entails all of its members: ∀X⊆L ∀p (p∈X ⇒ X⎥= p) Proof: Suppose not. Then there is a set of sentences X and a proposition p ∈ X with X ⎥≠ p. Therefore (by the definition of entailment) there is a q incompatible with p but compatible with X, in other words: {p, q} is incoherent and X∪{q} is coherent. But (by set theory) we have {p, q} ⊆ X∪{q}, and so by persistence we know that X∪{q} is incoherent, a contradiction. Inclusion Lemma Each possible world entails all and only its subsets: W max. coh. ⇒ (W⎥= X ⇔ X ⊆ W) Proof: Let W be max. coh. Then (by the definition of possible world) W coh. ∧ ∀X⊆L: W⎥= X ∨ W∪X inc. (*). “⇐”: Let X ⊆ W. Therefore W∪X = W (by set theory). As W is coh. this gives us W∪X coh. Therefore (by (*)) W⎥= X. “⇒”: Let W⎥= X. By the Either-Or lemma this implies W∪X coh. As W is a possible world and W∪W∪X = W∪X is coh., we now have W⎥= W∪X. On the other hand, persistence gives us W∪X ⎥= W. Together we have W∪X =⎢⎥= W. If we look at W and X not as sets of sentences, but as sets of propositions (i.e. sets of entities individuated solely in virtue of their incompatibility relations), then this gives us W∪X = W. From there, set theory gives us X⊆W.3 3
To illustrate the difference between sentences and propositions, let us take a look at the Three-in-a-Line Model. It consists of a language of three sentences r, s, and t, such that all three sentences are self-compatible, r is compatible with s, s is compatible with t, and all other sets are incoherent. The Three-in-a-Line Model shows that though set-inclusion may imply entailment (cf. the membership lemma), entailment does not imply set-inclusion. We have {r}⎟= {r, s}, but obviously not {r} ⊇ {r, s}.—Now we have to realize the difference between sentences and propositions. As propositions are individuated by their incompatibility sets, different sets of sentences can express the same proposition, as is in fact the case in the Three-in-aLine Model: {r} and {r, s} are incompatibility-equivalent and thus must be seen as sets containing the same proposition(s).
How To Kripke Brandom’s Notion of Necessity
151
Completion Lemma Every coherent set of sentences is entailed by some possible world: X coh. ⇒ ∃W (W max. coh. ∧ W⎟= X) Instead of proving this, we direct your attention to Hughes/Cresswell 1996, 115ff, where a similar result is proved in the framework of possible worlds semantics. Incompatibility Lemma A proposition is self-incompatible just in case it is false in all possible worlds: ∀W (W max. coh. ⇒ q false in W) ⇔ {q} inc. Proof: “⇐”: The definition of possible worlds immediately implies that no incoherent set of sentences is implied by any possible world. “⇒”: Suppose not. Then there is a q such that ∀W (W max. coh. ⇒ q false in W) but {q} coherent. But the completion lemma gives us a W max. coh. with {q}⊆W. Therefore W⎥= q, which contradicts ∀W (W max. coh. ⇒ q false in W).
2.3
An Alternative Definition of Necessity
To motivate the alternative definition of necessity, we will analyse “q is incompatible with necessarily p” in standard Kripke semantics. Let us start with formulating the plain incompatibility of q and p in Kripke semantics. As Brandom himself notes, the incompatibility of q and p is the impossibility of the conjunction q ∧ p (“non-compossiblity”) (cf. BSD 5.4, p. 13). Therefore: 1 q incompatible p 2 ⇔ ¬M(q ∧ p) incompatibility as non-compossibility 3 ⇔ ¬∃w (q true in w ∧ p true in w) Kripke-semantics for modality We use w and w’ as variables ranging over Kripkean possible worlds. Uniform substitution of Lp for p gives us: 4 q incompatible with Lp 5 ⇔ ¬∃w (q true in w ∧ Lp true in w) 6 ⇔ ∀w (q true in w ⇒ Lp false in w) 7 8
⇔ ∀w (q false in w) ∨ ∀w (q true in w ⇒ Lp false in w) ⇔ ∀w (q false in w) ∨ ∀w (q true in w ⇒ ∃w’ (w’ accessible from w ∧ p false in w’))
1 – 4; substitution Lp/p 5; propositional calculus, predicate calculus and Kripke semantics 6; predicate calculus: ∀x (Fx → Gx) ⇔ ∀x (¬Fx) ∨ ∀x (Fx → Gx) 7; Kripke-semantics: definition of necessity
152
Benedikt Paul Göcke, Martin Pleitz, Hanno von Wulfen
9
8; definitions of possible world, truth ⇔ q self-incompatible in a world; incompatibility lemma ∨ ∀W1 (W1 max. coh. ∧ W1⎥= q ⇒ ∃W2 (W2 max. coh. ∧ W2 accessible from W1 ∧ W2⎟≠ p)) The motivation of step 6-7 is to get a structure similar to that of Brandom’s definition (all definitions of necessity in incompatibility semantics have to treat self-incompatible sentences). Compare the alternative definition to Brandom’s own definition: Brandom’s definition of necessity: q is incompatible :⇔ q self-incompatible ∨ ∃Y ({q}∪Y coh. ∧ Y⎟≠ p) with Lp Alternative definition of necessity: q is incompatible :⇔ q self-incompatible with Lp ∨ ∀W1 (W1 max. coh. ∧ W1⎥= q ⇒ ∃W2 (W2 max. coh. ∧ W2 accessible from W1 ∧ W2⎟≠ p))
3
Properties of Alternative Necessity
We investigate the properties of the alternative notion of necessity by locating it in the well-studied hierarchy of modal logical systems. We start with an overview we have collected from Hughes/Cresswell 1996 and then show that the axioms of T, K, and B are valid, whereas the S4-axiom is not. Overview over some Modal Systems System Syntax: Axioms
Kripke Semantics: Properties of Accessibility
K
(K) L(p → q) → (Lp → Lq)
none
T
(K) + (T) Lp → p
reflexivity
B
(K) + (T) + (B) p → LMp
reflexivity and symmetry
S4
(K) + (T) + (4) Lp → LLp
reflexivity and transitivity
How To Kripke Brandom’s Notion of Necessity S5
(K) + (T) + (E) Mp → LMp
153
reflexivity, transitivity and symmetry
3.1 The T-axiom Lp → p Reflexivity Lemma Every possible world is accessible from itself: ∀W (W max. coh. ⇒ W accessible from W) Proof: Let W be max. coh.. Then for every p∈W there is an X⊆W, namely X={p}, such that X∪{p} is coh., as X∪{p} = {p} and {p} coh. (because of persistence W could not be coherent if there were a p∈W with {p} inc.). The T-axiom The definition of necessity validates the T-axiom. What is necessarily the case is actually the case: ⎥= Lp → p Proof: In view of ⎥= (p → q) ⇔ (p⎥= q) it suffices to show Lp⎥= p. Let q be incompatible with p. Case 1: {q} incoherent. Then (by the first adjunct of the alternative definition of necessity) q incompatible with Lp. Case 2: {q} coherent. Then {p, q} incoherent, because q incompatible p. We have to show ∀W1 (W1 max. coh. ∧ W1⎥= q → ∃W2 (W2 max. coh. ∧ W2 accessible from W1 ∧ W2⎟≠ p)). Suppose not. Then ∃W1 (W1 max. coh. ∧ W1⎥= q ∧ ∀W2 (W2 max. coh. ∧ W2 accessible from W1 → W2⎟= p)) (i.e. there is a q-world such that all accessible worlds are p-worlds). Together with the reflexivity lemma, this implies ∃W1 (W1 max. coh. ∧ W1⎥= q ∧ W1⎟= p)), which implies {p, q} coh., which contradicts q being incompatible with p.
3.2
The K-axiom L(p → q) → (Lp → Lq)
The definition of necessity validates the K-axiom: ⎥= L(p → q) → (Lp → Lq). Proof: We generally have ⎥=(p → q) ⇔ (p⎥= q). We also have (p⎥= q) ⇔ (⎥=p ⇒⎥=q).4 So it suffices to show that⎥=L(p → q) ⇒ (Lp⎥= Lq). 4
Proof of (p⎥= q) ⇔ (⎥=p ⇒⎥=q): “⇒” is just modus ponens, which holds for incompatibility entailment. “⇐”: We start from ⎥=p ⇒⎥=q. Interpreting the metalanguage conditional in the usual truth-functional way, there are three cases to be considered: Case 1) ⎥=p and ⎥=q: The incompatibility sets of p and q are identical, as they both include just the incoherent sets. Therefore everything incompatible
154
Benedikt Paul Göcke, Martin Pleitz, Hanno von Wulfen
Now let⎥=L(p → q) and let X be incompatible with Lq. Then X inc. ∨ ∀W1 (W1 max. coh. ∧ W1⎥= X ⇒ ∃W2 (W2 max. coh. ∧ W2 accessible from W1 ∧ W2⎟≠ q)). (To show: X incompatible with Lp.) Case 1: X inc. Then X incompatible with Lp. Case 2: X coh. Then ∀W1 (W1 max. coh. ∧ W1⎥= X ⇒ ∃W2 (W2 max. coh. ∧ W2 accessible from W1 ∧ W2⎟≠ q)). For this W2 we have W2⎟≠ p. (Proof: Suppose not. Then W2⎟= p. We have ⎥=L(p → q), and therefore by the T-axiom ⎥=(p → q), which is equivalent to (p⎥= q). The transitivity of incompatibility entailment gives us W2⎟= q, which contradicts W2⎟≠ q.) W2⎟≠ p leads (by the definition of necessity) to X’s being incompatible with Lp.
3.3 The B-axiom p → LMp Before we prove the validity of the B-axiom in incompatibility semantics, we show that accessibility is symmetrical. Symmetry Lemma The accessibility of one world by another is given just in case the two worlds overlap, i.e. their intersection is not empty, which is a symmetrical situation. Put formally: ∀W1 ∀W2 (W1, W2 max. coh. ⇒ (W2 accessible from W1 ⇔ W1 accessible from W2)) Proof: Let W1, W2 be max. coh. and let W2 be accessible from W1. Then (by the definition of accessibility) ∀X (W2⎥= X ⇒ ∃Y (W1⎥= Y ∧ X∪Y coh.)). As W2⎥= W2, this implies ∃Y (W1⎥= Y ∧ W2∪Y coh.). As W1 and W2 are max. coh., we have W1⎥= Y ⇔ W1∪Y coh. and W2⎥= Y ⇔ W2∪Y coh. Therefore (by uniform substitution of equivalents) ∃Y (W2⎥= Y ∧ W1∪Y coh.). Now let Z be such that W1⎥= Z. As W1 is max. coh., this implies W1∪Z coh. The inclusion lemma gives us Y⊆W1 and Z⊆W1, therefore Y∪Z⊆W1, therefore (again by the inclusion lemma) W1⎥= Y∪Z, therefore (by the definition of maximal coherence) Y∪Z coh. Put together we have ∀Z (W1⎥= Z ⇒ ∃Y (W2⎥= Y ∧ Y∪Z coh.)), or equivalently: W1 accessible from W2. Before we can prove the B-axiom, we have to show the following intuitive Kripkean equivalency:
with q is incompatible with p and we have p⎥= q. Case 2) ⎥≠p and ⎥=q: The incompatibility set of q includes all and only the incoherent sets and the incompatibility set of p includes all sets. Therefore I(q) ⊂ I(p) and so p⎥= q. Case 3) ⎥≠p and ⎥≠q: The incompatibility sets of p and q are identical, as they both include all sets. Therefore everything incompatible with q is incompatible with p and we have p⎥= q.
How To Kripke Brandom’s Notion of Necessity
155
Kripkean Equivalency The necessity of a proposition in a world is equivalent to the truth of this proposition in all accessible worlds: W1 max. coh. ⇒ (W1⎥= Lp ⇔ ∀W2 (W2 max. coh. ∧ W2 accessible from W1 ⇒ W2⎥= p)) Proof: “⇒”: Let W1 be max. coh. and let W1⎥= Lp. Then the definition of maximal coherence gives us W1 compatible with Lp. By the definition of necessity this implies that the following is the case: ¬(W1 inc. ∨ ∀W3 (W3 max. coh. ∧ W3⎥= W1 ⇒ ∃W4 (W4 max. coh. ∧ W4 accessible from W3 ∧ W4⎟≠ p))). De Morgan: W1 coh. ∧ ¬∀W3 (W3 max. coh. ∧ W3⎥= W1 ⇒ ∃W4 (W4 max. coh. ∧ W4 accessible from W3 ∧ W4⎟≠ p)). Since W1 is supposed to be coherent, this by predicate logic implies: ∃W3 (W3 max. coh. ∧ W3⎥= W1 ∧ ∀W4 (W4 max. coh. ∧ W4 accessible from W3 ⇒ W4⎟= p)) Since W1 is itself maximally coherent, the only maximally coherent set it is entailed by is itself. So: ∀W4 (W4 max. coh. ∧ W4 accessible from W1 ⇒ W4⎟= p) “⇐”: Let W1 max. coh. and let ∀W2 (W2 max. coh. ∧ W2 accessible from W1 ⇒ W2⎥= p). Let X be incompatible with Lp. Then X inc. ∨ ∀W1 (W1 max. coh. ∧ W1⎥= X ⇒ ∃W2 (W2 max. coh. ∧ W2 accessible from W1 ∧ W2⎟≠ p)). (To show: X incompatible W1.) Case 1: X inc. Then X incompatible with W1. Case 2: X coh. Then ∀W1 (W1 max. coh. ∧ W1⎥= X ⇒ ∃W2 (W2 max. coh. ∧ W2 accessible from W1 ∧ W2⎟≠ p)). Suppose X is compatible with W1. Then (by the definition of possible world) W1⎥= X and therefore ∃W2 (W2 max. coh. ∧ W2 accessible from W1 ∧ W2⎟≠ p). This contradicts ∀W2 (W2 max. coh. ∧ W2 accessible from W1 ⇒ W2⎥= p). Therefore X is incompatible with W1. The B-axiom The definition of necessity validates the B-axiom. What actually is the case is necessarily possible: ⎥= p → LMp Proof: It suffices to show p⎥= LMp. So let X be incompatible with LMp. As Mp ≡ NLNp, we have X incompatible LNLNp. Therefore X inc. ∨ ∀W1 (W1 max. coh. ∧ W1⎥= X ⇒ ∃W2 (W2 max. coh. ∧ W2 accessible from W1 ∧ W2⎟≠ NLNp)). Case 1: X inc. Then X incompatible p. Case 2: X coh. Then ∀W1 (W1 max. coh. ∧ W1⎥= X ⇒ ∃W2 (W2 max. coh. ∧ W2 is accessible from W1 ∧ W2⎟≠ NLNp)). Because of the closure with respect to the operations of propositional logic, we have W2⎟≠ NLNp ⇔ W2⎟= LNp. We therefore have ∀W1 (W1 max. coh. ∧ W1⎥= X ⇒ ∃W2 (W2 max. coh. ∧ W1 acc. W2 ∧ W2⎥= LNp)) (i.e. for every X-World there is an accessible LNp-world). The Kripkean Equivalency gives us: W2⎥= LNp ⇔ ∀W3 (W3 max. coh. ∧ W3 accessible from W2 ⇒ W3⎥= Np) Therefore ∀W1 (W1 max. coh. ∧ W1⎥= X ⇒ ∃W2 (W2 max. coh. ∧ W2 accessible from W1 ∧ ∀W3 (W3 max. coh. ∧ W3 accessible from W2 ⇒ W3⎥=
156
Benedikt Paul Göcke, Martin Pleitz, Hanno von Wulfen
Np))). Now the symmetry of accessibility implies that ∀W1 (W1 max. coh. ∧ W1⎥= X ⇒ W1⎥= Np). This shows that X is incompatible with p in case 2. Taking case 1 and 2 together we have p⎥= LMp.
3.4
The S4-axiom Lp → LLp is not valid
To show that the S4-axiom Lp → LLp is not valid, the guiding intuition is this: In Kripke semantics, we only find counter-examples to the S4-axiom if the accessibility relation between possible worlds is not transitive. We need three worlds w1, w2, w3, such that w2 is accessible from w1, w3 is accessible from w2 but w3 is not accessible from w1, and a proposition p true in w1 and w2 but false in w3. Then Lp → LLp is false in w1.—But how do we get from incompatibility partitions to the desired possible worlds situation? Constructing Countermodels in Incompatibility Semantics Let us try our method first in a simpler case: What about ⎥= p → Lp? We ask whether p⎟= Lp, i.e. whether every q incompatible with Lp is incompatible with p. After dealing with self-incompatible qs as usual, we have to check: ∀W1 (W1 max. coh. ∧ W1⎥= q ⇒ ∃W2 (W2 max. coh. ∧ W2 accessible from W1 ∧ W2⎟≠ p)) ⇒ {p, q} inc. We want this to be false, as we want p → Lp not to be valid. (p → Lp together with the T-axiom Lp → p would make modal logic collapse into propositional calculus.) We therefore have to find a model that shows p → Lp to be not valid. The Blackberry Model: A Countermodel to p → Lp Consider a language containing only the three sentences r, s, t with the following incompatibility partition: All three sentences r, s, and t are self-compatible, all three pairs of sentences {r, s}, {s, t}, and {r, t} are coherent, but the set containing all three sentences {r, s, t} is incoherent. A way of representing the incompatibility partition, which already gives all incompatibility sets, is the following (read “+” as “coherent” and “-” as “incoherent”): 1 2 3 4 5 6 7 α {r} {s} {t} {r, s} {s, t} {r, t} {r, s, t} β + + + + + + γ 1 {r} + + + + + + 2 {s} + + + + + + 3 {t} + + + + + + 4 {r, s} + + + + 5 {s, t} + + + + 6 {r, t} + + + + 7 {r, s, t} The Blackberry Model satisfies symmetry and persistence.
How To Kripke Brandom’s Notion of Necessity
157
Some general remarks concerning this way of representing incompatibility partitions: • The two rows β and γ (set bold) already contain all the information (by Brandom’s Axiom of Partition). • The main diagonal is identical to row γ and column γ. • “-” in row γ gives us “-” for the whole row (by Brandom’s Axiom of Persistence). • “-” in colomn γ gives us “-” for the whole column (by Brandom’s Axiom of Persistence). • The diagram is symmetrical at the main diagonal (by Brandom’s Axiom of Symmetry). This particular diagram gives us the following incompatibiltiy sets: Incompatibility Set {r} {{s, t}, {r, s, t}} {s} {{r, t}, {r, s, t}} {t} {{r, s}, {r, s, t}} {r, s} {{t}, {s, t}, {r, t}, {r, s, t}} {s, t} {{r}, {r, s}, {r, t}, {r, s, t}} {r, t} {{s}, {r, s}, {s, t}, {r, s, t}} {r, s, t} {{r}, {s}, {t}, {r, s}, {s, t}, {r, t}, {r, s, t}} We get the following entailments: {r, s}⎟= {r} {r, s}⎟= {s} {s, t}⎟= {s} {s, t}⎟= {t} {r, t}⎟= {r} {r, t}⎟= {t} There are no non-trivial entailments on one level, that is between {r}, {s}, {t} or between {r, s}, {s, t}, {r, t}. {r, s, t} entails everything (as any incoherent set should). {r, s}, {s, t}, and {r, t} are possible worlds, for every set is either entailed by them or incompatible with them. Every world is accessible by every world. For example: {r, s} is accessible by {s, t}. Proof: {r, s} has three subsets: {r}, {s}, and {r, s}. For each of these there is a subset of {s, t}, namely {s}, such that their union is coherent. What is the incompatibility set of Ls in the blackberry model? Remember: X incompatible Ls iff X self-incompatible ∨ ∀W1 (W1 max. coh. ∧ W1⎥= X ⇒ ∃W2 (W2 max. coh. ∧ W2 accessible from W1 ∧ W2⎟≠ s)) (call this “Φ”) s in the blackberry model is not entailed by the possible world {r, t}, and this possible world is accessible by all three possible worlds there are in this case. Therefore every co-
158
Benedikt Paul Göcke, Martin Pleitz, Hanno von Wulfen
herent set X satisfies Φ: There is a non-s-world accessible by every X-world. Thus every set is incompatible with Ls. Therefore s is incompatible with Ls. But as {s} is coherent, s is compatible with s. So we have a set of sentences—namely {s}—that is incompatible with Ls but compatible with s. Therefore s⎥≠ Ls. So the Blackberry Model shows that p → Lp is not valid.
{u}
{r, s}
6
7
8
9
10 11 12 13 14 15 {r, t, u}
{r, s, u}
{s, t, u}
{r, s, t}
{r, t}
β
{r, s, t, u}
5
{s, u}
4
{r, u}
3
{t, u}
2
{s, t}
1
{t}
0
{s}
α
{r}
The Square Model: A Countermodel to the S4-axiom Lp → LLp The Square Model is inspired by the Blackberry Model, but we now start with a language that contains the four sentences r, s, t, and u. Of the six pairs of sentences four are coherent ({r, s}, {s, t}, {t, u}, and {r, u}), but the two “diagonal” pairs ({s, u} and {r, t}) are incoherent. All sets of three or four sentences are incoherent. All other sets are coherent. We get the following diagram:
γ + + + + + + + + 1 {r} + + + - + + - + 2 {s} + + + + - + + 3 {t} + - + + + - + + 4 {u} + + - + + - + + 5 {r, s} + + + - + 6 {s, t} + - + + - + 7 {t, u} + - + + - + 8 {r, u} + + - + - + 9 {s, u} - 10 {r, t} - 11 {r, s, t} - 12 {s, t, u} - 13 {r, s, u} - 14 {r, t, u} - 15 {r, s, t, u} - The Square Model satisfies symmetry and persistence. Only the rows and columns 1-8 are of interest here. We get the following incompatibility sets and entailments: 1 2
set {r} {s}
incompatibility set 3, 6, 7, 9-15 4, 7, 8, 9-15
all non-trivial entailments by the set none none
How To Kripke Brandom’s Notion of Necessity
159
3 {t} 1, 5, 8, 9-15 none 4 {u} 2, 5, 6, 9-15 none 5 {r, s} 3, 4, 6, 7, 8, 9-15 {r, s}⎥= {r}, {r, s}⎥= {s} 6 {s, t} 1, 4, 5, 7, 8, 9-15 {s, t}⎥= {s}, {s, t}⎥= {t} 7 {t, u} 1, 2, 5, 6, 8, 9-15 {t, u}⎥= {t}, {t, u}⎥= {u} 8 {r, u} 2, 3, 5, 6, 7, 9-15 {r, u}⎥= {r}, {r, u}⎥= {u} 9 {s, u} 1-15 {s, u}⎥= ∅ 10 {r, t} 1-15 {r, t}⎥= ∅ 11 {r, s, t} 1-15 {r, s, t}⎥= ∅ 12 {s, t, u} 1-15 {s, t, u}⎥= ∅ 13 {r, s, u} 1-15 {r, s, u}⎥= ∅ 14 {r, t, u} 1-15 {r, t, u}⎥= ∅ 15 {r, s, t, u} 1-15 {r, s, t, u}⎥= ∅ (For reasons of readability, we use the numbers from 1 to 15 as names of the sets of sentences.) By the definition of possible world, we have four possible worlds: w1 := {r, s}, w2 := {s, t}, w3 := {t, u}, and w4 := {r, u}. By the definition of accessibility, all and only worlds sharing one element are accessible from each other. Thus accessibility is not transitive in this case. In particular, we get from w1 to w2 and from w2 to w3 but not from w1 to w3. This is the possible worlds situation we were looking for. Now (guided by Kripkean intuitions) we need a proposition true in w1, w2, and w4 but false in w3. This proposition should be v :≡ r∨s. The proposition v :≡ r∨s should be necessary in w1, as it is true in in all accessible worlds (w1, w2, w4). Thus Lv should be necessary in w1. But LLv should be false in w1, as v is false in w3. Put directly in terms of incompatibility, our goal at this point is to show that w1 is compatible with Lv, but incompatible with LLv and that thus Lp → LLp is not valid. First, however, we have to get clearer about v. What is the incompatibility set of v :≡ r∨s? Intuitively, only what is incompatible both with r and with s should be incompatible with r or s, thus I(v) should be I(r)∩I(s). Here I(v) = {{t, u}, *}, where * abbreviates the list of the incoherent sets 9-15. This implies that v is entailed by all sets but {t}, {u}, and {t, u}. Note that {t, u} = w3. In possible worlds terms we get (from the inclusions among incompatibility sets): {r, s}⎥= {v}, hence v true in w1, {s, t}⎥= {v}, hence v true in w2, {t, u}⎥≠ {v}, hence v false in w3, {r, u} ⎥= {v}, hence v true in w4. (So we should have w1⎥= Lv and w1⎥≠ LLv. But let us continue our argument in direct incompatibility semantics.)
160
Benedikt Paul Göcke, Martin Pleitz, Hanno von Wulfen
To show: w1 compatible with Lv Remember w1 incompatible Lv iff w1 self-incompatible ∨ ∀W1 (W1 max. coh. ∧ W1⎥= w1 ⇒ ∃W2 (W2 max. coh. ∧ W2 accessible from W1 ∧ W2⎟≠ v)). As w1 is coherent we have to check whether ∀W1 (W1 max. coh. ∧ W1⎥= w1 ⇒ ∃W2 (W2 max. coh. ∧ W2 accessible from W1 ∧ W2⎟≠ v)) which (as w1 is the only maximally coherent set entailing w1) is equivalent to ∃W2 (W2 max. coh. ∧ W2 accessible from w1 ∧ W2⎟≠ v). Only w1, w2, and w4, i.e. {r, s}, {s, t}, and {r, u}, satisfy W2 max. coh. ∧ W2 accessible from w1, but all three entail v. Thus w1 is compatible with Lv. To show: w1 incompatible with LLv As w1 is coherent we have to show that ∀W1 (W1 max. coh. ∧ W1⎥= w1 ⇒ ∃W2 (W2 max. coh. ∧ W2 accessible from W1 ∧ W2⎟≠ Lv)). As w1 max. coh., this is equivalent to ∃W (W max. coh. ∧ W accessible from w1 ∧ W⎟≠ Lv). As w2 max. coh. and w2 accessible from w1, it suffices to show w2⎟≠ Lv. As w2 is compatible with w2, one way to show this would be to show that w2 is incompatible with Lv. As w2 is coherent we have to show that ∀W1 (W1 max. coh. ∧ W1⎥= w2 ⇒ ∃W2 (W2 max. coh. ∧ W2 accessible from W1 ∧ W2⎟≠ v)) which is equivalent to ∃W (W max. coh. ∧ W accessible from w2 ∧ W⎟≠ v). This W is given by w3, as we already know that w3 max. coh. and w3 accessible from w2 and w3⎟≠ v. Hence the Square Model shows that the S4-axiom Lp → LLp is not valid.
3.5
Comparison of Brandom’s and the Alternative Definition of Necessity
We now want to compare: Brandom’s definition of necessity: q is incompatible :⇔ q self-incompatible ∨ ∃Y ({q}∪Y coherent ∧ Y⎟≠ p) with LBp Alternative definition of necessity: q is incompatible :⇔ q self-incompatible with LAp ∨ ∀W1 (W1 max. coh. ∧ W1⎥= q ⇒ ∃W2 (W2 max. coh. ∧ W2 is accessible from W1 ∧ W2⎟≠ p)) The alternative definition of necessity has the (desirable) consequence that Brandom’s necessity entails alternative necessity: LBp⎥= LAp. Proof: We have to show that everything incompatible with LAp is incompatible with LBp. Let q be incompatible with LAp. Then (by the alternative definition): q self-incompatible
How To Kripke Brandom’s Notion of Necessity
161
∨ ∀W1 (W1 max. coh. ∧ W1⎥= q ⇒ ∃W2 (W2 max. coh. ∧ W2 accessible from W1 ∧ W2⎟≠ p). Case 1: q is self-incompatible. Then (by Brandom’s definition) q is incompatible with LBp. Case 2: q is not self-incompatible. Then ∀W1 (W1 max. coh. ∧ W1⎥= q ⇒ ∃W2 (W2 max. coh. ∧ W2 accessible from W1 ∧ W2⎟≠ p)) (call this “Φ”). As {q} is coherent, the completion lemma and the inclusion lemma imply that there is a W1 such that W1 max. coh. ∧ W1⎥= q. For this W1, Φ gives us ∃W2 (W2 max. coh. ∧ W2 accessible from W1 ∧ W2⎟≠ p). The closure of the language L with respect to propositional calculus (together with the maximality of possible worlds) now gives us W2⎟= Np. We know (from the symmetry lemma) that accessibility is symmetrical. Therefore W1 is accessible from W2. Therefore, for every proposition true in in W1 there is a set Y of propositions true in W2 such that they are compatible. In particular, there is a Y⊆W2 such that Y∪{q} is coherent. On the other hand, Y⎥≠ p. (Proof: Suppose Y⎥= p. We know that W2⎥= Y. The transitivity of entailment then gives us W2⎥= p, a contradiction.) In sum, we have for our q: ∃Y (Y∪{q} ∧ Y⎥≠ p). So we have shown that (by Brandom’s definition of necessity), q is incompatible with LBp.
REPLIES
Robert Brandom
REPLY TO “BRANDOM ON KNOWLEDGE AND ENTITLEMENT” SCHMORANZER AND SEIDE This is a rich and challenging paper and I am grateful for it. I think there may be three large issues worth thinking about here. One of them is the question of whether the notion of justification that is in play makes the dialectical tool of default entitlement epistemically useless. That is, does the notion of all things considered entitlement collapse into that of default entitlement? Here is how I’m thinking about it working. If somebody who is prima facie a reliable reporter of, say, vultures makes a claim, she has prima facie or default entitlement to that claim. Now how can that default be challenged? The challenge is always in the eyes of a scorekeeper. As suggested in the first part of the paper, it’s really knowledge attributions that I care about. It’s keeping score on someone as knowing or not. Well, the barn façade cases indicate, again as you acknowledge, that assessments of reliability depend on what the reference class is. A challenger can challenge a claim that somebody made by suggesting that the right reference class for assessing reliability is a different one from the reference class that had been in play before. So, somebody says, “That’s a vulture.” You, acting as a challenger, ask him, “How do you know?” He says, “I know vultures when I see one.” (That is in effect invoking his reliability as default entitlement to that claim.) But now you say, “Look, apparently you don’t realize that Disney has been testing their animatronic vultures
164
Robert Brandom
in the vicinity. There are lots of these replicas flying around. Are you really sure you can tell the difference between the one and the other?” Now, he may or may not think that he can tell the difference. But I, as the scorekeeper, have been keeping score on him, and if I believe the challenger that there are all these robot vultures flying around, then the reference class is for me as if he were in vulture-façade county now. If that is the right reference class, then I’m going to say, “Well, now that I’ve come to believe that the situation is different and that there are these robot things, I’m going to say that though he was prima facie entitled to it (he had a default entitlement), he’s not all things considered entitled to it.” I’m not going to take him to be entitled to his claim— I’m not going to take him to know that that’s a vulture—even if, in fact, I happen to know that there is a real vulture there. So even if the only sort of entitlement in the vicinity is externalist, in the sense that is a matter of entitlement of reliability, still, just because collateral premises can change, in the eyes of the scorekeeper, what the relevant reference class is, the giving and asking for reasons can still interfere with the attribution of reliability and hence of knowledge. So it does not actually seem to me as though the two notions (of default justification and all things considered justification) collapse in on one another. Now, the second most basic challenge offered was about second-order skepticism. This, I think, comes out of the argument in section 2 that ultimately all justification is external justification in this account. It seems to me a very positive and useful suggestion that we need to think harder than I do in the book about the difference between inferentialism and non-inferentialism about justification, on the one hand, and internal and non-internal theories of justification, on the other. I’m not sure how we should come out there. But the argument presented for the justification being external is ultimately epistemological. It is concerned with such questions as whether free moves are justified because they’re accepted in the speech community. And it is then asked: “Does the fact that something is accepted in the speech community actually give us a good reason to consider it justified? Is that a reliable belief forming mechanism?” And so on. And here I think arguing that way betrays an epistemological rather than a semantic orientation. So the framing consideration—“Look this is a book about semantic questions, but we can ask epistemological questions about it”—comes to the fore. Making It Explicit is a
Replies
165
book about semantics, and the epistemology comes downstream from that. And that’s not the way classical epistemology thinks about its subject. I think the soft underbelly of classical epistemology is its implicit semantics. Someone like Alvin Goldman takes it for granted that the beliefs and claims he’s assessing epistemologically come with their semantics fixed. We know what they mean, and the question is: “Under what circumstances would something that means this count as justified?” But I have a more Quinean perspective. It seems to me that apart from questions about what would and even does justify a claim, one can’t treat its meaning as being settled. The elements in the inferential articulation of a claim—the articulation of a claim with respect to what would entitle you to it and how it’s related to other claims that are entitled—are also fundamental elements in the semantic articulation of the claim. So, when Quine says that a claim like “There have been black dogs” is a free move (that is, a move everyone will let you get away with), it’s not that we already know, when Quine is making that remark, what the claim “There have been black dogs” means, and we can then argue that, since everybody we know accepts this, we have good reason to accept it. The web of belief for him is partly held in place by what the free moves are. That such and such claims are free moves makes an essential contribution to the meaning of the various terms that are used in expressing those claims. So when we think about meaning and justification from a semantic point of view, we’re looking at the relative justificatory relations of the different claims as part of what we’re going to use to decide what these things mean, and, at the very same time, as part of what we’re going to use to decide which of them we’re justified in believing. We cannot, in a Carnapian way, separate the enterprise: first fix the meanings, that is, fix the language that you are working in, and only then worry about which of the claims with those meanings are justified. So it is not that we say in general that whatever the community accepts, we have good reason to accept. Rather we have to think more globally and holistically about the constraint on the meaning of all the terms, which is given by which of the claims are free moves. So I would reject the question, “In general, what justifies free moves? What justifies noninferential reports?” The fact that we treat the free moves (where the default entitlement derives from the linguistic type of the expression) and the non-inferential reports (where the entitlement derives from the circumstances of the particular
166
Robert Brandom
tokening) as justified provides some of the global constraints on what we mean. When we think about the epistemological questions from that semantic perspective, the challenge, it seems to me (and the reason I delved into any of these epistemological issues to begin with), is to say how an inferentialist in semantics—someone who thinks of meaning in terms of what follows from what, what is a reason for and against what, and so on—can see features of the world transposed into a conceptual key. How can an inferentialist take facts about what events reliably covary with what other events and see them reflected in the inferential structure that articulates the contents of the concepts? How do you get causal covariation in the world to be reflected in the inferential structure that confers content on your concepts? And the answer it seemed to me was: the reliability inference. If there’s a reliable covariation, then the scorekeeper can take it that the one event gives one good reason to infer the existence of the other event. In some sense, the basis of all our empirical justification (and of our even being in touch with the world at all) is that reliably covarying occurrences have been incorporated into inferences that are accepted within the community. In that sense, it is reliability all the way down. But it is absolutely crucial in this picture that that reliability get transposed into a conceptual, that is, an inferential key—that it take the form of giving and asking for reasons. In what I take to be a purely externalist epistemology, someone’s entitlement never turns on whether they can actually give a justification or a reason for what they’re doing. But in my setting, it absolutely is incumbent on people, at least sometimes, to be able to give reasons. Indeed, in the typical case, if they can’t cite their own reliability, we are not going to attribute knowledge to them. The pure reliabilist cases are outliers, where we have to be convinced that even though the person doesn’t know that he is reliable, he still is. Typically, when people are reliable, they know they are, and if they do not take themselves to be reliable, it is because they know something more about the reference class than we know. They say, “Look, I know I seem to have been lucky and gotten this right a couple of times, but you have to know that my overall track record on this is terrible. The reference class is not what you think.” In exceptional cases, we can have good reason to believe that even though they do not think they’re reliable, in fact they are. But typically, if
Replies
167
someone cannot give any reason for the claim she makes, we’re not going to attribute knowledge to her. So, it is critical that our tracking the causal structure of the world with our inferential structure have the form of giving and asking for reasons. Now, is that enough to count as splitting the difference between internalism and externalism in epistemology? The internalism and externalism are really functioning at different levels. Overall, since tracking the world with our inferences is what being in touch with the world is, there isn’t on this picture a global skeptical worry about our being out of touch with the world. But in the actual practice of giving and asking for reasons, not only do free moves and noninferential reports play a role, but so does the actual giving of reasons. So, it seems to me that it’s fair to describe it as splitting the difference between internalism and externalism.
REPLY TO “OF MU-MESONS AND ORANGES” APEL, BAHRENBERG, KÖHNE, PRIEN, SUHM This is a challenging paper. I do accept that there are epistemically significant differences between observable and unobservable entities. The question is whether I can explain those differences in the methodological terms that I claim are appropriate to that distinction, or whether I have to think about them in ontological terms. That is, do I have to think of them as fundamentally different kinds of entities in order to explain these epistemic differences, or are the merely methodological differences in our mode of access to them sufficient? I want to acknowledge that the differences that you point to exist, and to try to give explanations of them. Of course I do acknowledge that if someone is challenged in their observation of things that people do not normally think of as observable, like mumesons, they will retreat to what I sometimes call, following Quine, “more observable” entities. And by that I mean entities where the capacity reliably non-inferentially differentially to respond to and report their presence is either wider spread in the community, so that more people have that non-
168
Robert Brandom
inferential reporting capacity (for instance, there are more people who can report the difference between the shapes of vapor trails than there are who can report the presence of mu-mesons), or the conditions under which they are reliable reporters of them are much wider. So the physicists can only report the presence of these things if there is a bubble chamber there, but not otherwise. And that is a very specialized set of circumstances. There are good methodological reasons for someone who is asked to justify a non-inferential report to retreat to an inferential justification of it from things that are more widely or more reliably reportable. That justification is going to be shared by people who either do not have the bubble chamber in front of them or don’t have the specialized training to observe it. But it does not follow from the fact that people would under such circumstances retreat to an inferential justification of that claim, that the original report was actually a result of an inference, rather than of a non-inferential reporting capacity. That is the inference I reject. I do not think that is the right way to decide whether an inference was performed or not. That really in some sense is the crux of the question, I think: how do you tell whether something was an exercise of a non-inferential, observational reporting capacity, on the one hand, or whether it was the result of a process of inference, on the other? It used to be that people just thought, well, ask them whether they made an inference; let them introspect and see. But the trouble is that it is possible to make all sorts of unconscious inferences. Any attorney who is cross-examining an eyewitness will find that the eyewitness herself is not very good at distinguishing between what she actually saw and what she inferred from what she saw, perhaps given common background knowledge. If you give up the idea that people can just introspect and tell whether they have performed an inference or not, how do you decide whether an inference really went on? From the fact that our retinal images are all two-dimensional, does it follow that we actually made inferences to the three-dimensional objects? We’re not aware of it, but if there are unconscious inferences, how do we decide? Now, people have suggested various ways of resolving this question: when are there inferences involved? A traditional and popular one says, “Whenever it’s possible to make a mistake, there must have been an inference that was made: whatever you can just see you can’t be wrong about. So we
Replies
169
can sort things by the reliability of the report you make.” The trouble is that on this line it is always going to turn out that you can only report the current contents of your own mind. For you can be wrong about everything else. Quine, in his essay, “Epistemology Naturalized,” looks at some reliably covariant chain of events: it is a ringing of a bell, which causes the perturbation of the air between it and me; that changes the pressure inside of my ear and that moves my eardrum back and forth; that changes the electrical potentials in my auditory nerve, and so on. And he points out that the farther out you go from your sensory surfaces the more different ways there are that you could be fooled. You think that you’re just hearing the bell ringing, but there are other ways that the air could be made to move that way. There are other ways besides the air moving that way that the pressure could be changed in your ear canal, and there are other ways that your ear drum could be moved. He suggests that what we really report is whatever is most proximal. We are most reliable about what’s closest to us in that chain. But again the result is going to be that all we can literally observe is the states of our own sensory surfaces. I claim that there’s nothing intrinsic to any element of this chain of more or less reliably covarying events that makes one or another of them in principle observable, and another one not. What settles what you are observing is not the chain of stimuli that leads you to respond, but rather what concept you apply at the end of the chain. If I apply the concept bell, then what I’m reporting is the bell; if I apply the concept changes in pressure in my auditory canal, then what I’m reporting are changes in pressure in my auditory canal. These two concepts are distinguished not by the circumstances of application of them, so much as by the inferential consequences of them, what I have committed myself to when I say that. When I say “that’s a bell,” it commits me to very different things and would be confirmed or refuted by going out and seeing whether there is a bell there, rather than by measuring what is going on in my ear. So these are different concepts that I apply. And it is those different inferential connections of the concepts that I apply that determine how far out—how distally in this chain of reliable, covarying chain of events the thing that I am reporting is. Though that report gets less and less reliable— the reliable differential responsive disposition is less and less reliable as we go out distally—there is no in principle end point that sets a limit to what is in
170
Robert Brandom
principle observable. So this is a proposal about what to mean by “observation”: observation is the noninferential application of a concept, as part of the exercise of a more or less reliable differential responsive disposition, a disposition to respond differentially to those things. “Noninferential” in this point means that I did not actually perform an inference myself; that is, that the application of the word or the concept was the final event in this chain of reliably covarying events. If the application of the concept was the final event in the chain, and that chain of reliably covarying events didn’t go through any other assertions, then I say that the application of that concept was observational. Now the fact that what I actually observed was farther out in the chain is compatible with my pointing out, when questions are raised about the reliability of different stages in this chain, that you can infer the fact that I came to non-inferentially from the occurrence of more proximal stages, or from features to which other people are in as good of a position as I am reliably differentially to respond. So, I deny that it follows from the fact that we would retreat to a safer observation and then infer from it to justify an observation, that it was an inferential move in the first place. I should say that one of my favorite alternative responses, besides the Quinean proximal/distal test for what one is noninferentially observing, and apart from mine, is Gil Harman’s in his 1974 book Thought. Harman, impressed by the idea of the possibility of unconscious inferences and sensitive to the invitation to skepticism that’s involved in the Quinean answer, suggests using whether some claim is subject to Gettier counterexamples (where one has a justified true belief, but doesn’t have knowledge, because intuitively the process by which it’s justified involves some false claims or some bad inferences) as the test of whether actually an unconscious inference was involved in it. I just mention that to say that it’s possible to go another way. Now the two sorts of cases that were forwarded as difficulties for this account of observation show, I think, genuine differences between observation of things where the reliable differential responsive disposition that’s appealed to is not widely spread and observation of things where it is widely spread, for example, when telling oranges. (Though remember that 10% of males are red/green colorblind and so have to make inferences where the rest of us do not.) Why should one on this account not be surprised that we
Replies
171
change our mind or find out we are wrong, a lot more about our entities that are in fact only de facto inferentially available to us (which on this methodological account is the only difference between theoretical entities and observable entities), than about entities that we actually can make observation reports about? Well, because, as a matter of contingent methodological fact (though not as a reflection of any ontological or metaphysical differences between the entities), we can only make claims about certain things (like phlogiston or, for that matter, oxygen) as the result of processes of inference. Now, inferences are much more sensitive to the truth of collateral beliefs than reliable differential responsive dispositions are. And the inferences outside of formal logic, material inferences, are always seriously multi-premise inferences; they are inferences that can be infirmed if various collateral premises turn out not to be true. So, when you inferentially justify something, your warrant for the conclusion is held hostage to the correctness of a whole host of collateral beliefs, whereas my capacity to respond differentially to the presence of red things, or indeed of mu mesons, is much less sensitive to those collateral beliefs. Now, because the mu mesons are so distal in the chain of reliably covarying events, collateral beliefs do affect the reliability of my reporting. If the bubble chamber is out of calibration or something like that, that is going to affect the reliability of my reporting. The longer that chain is, the more the truth of my beliefs about what is going on, and hence my reliability, are hostage to it. But that doesn’t depend on any ontological difference between theoretical entities and observational entities. It is explicable entirely in terms of the methodological difference that, as a matter of fact, we only have an inferential mode of access to certain entities.
REPLY TO “SUCCESSFUL ACTION AND TRUE BELIEFS” KOMPA, MÜLLAN, PRIEN, SCHWEIKARD I do not find very much to object to in this essay. What I objected to about the Whyte proposal is principle R, that you could define truth in terms of suc-
172
Robert Brandom
cess. And since this paper gives that up along the way, I do not have anything invested in objecting to the construction that is developed here. I guess I want to start thinking about it with a question as to exactly what remains at issue. I think this is put in different ways. Early on, we are told: “In our view, the main idea of Success Semantics is the claim that contact between our discursive practice and the world is established by the practical syllogism.” Now there is a quantifier missing in that claim: does it mean that some contact is established that way, or that all the contact is established that way? And the same ambiguity reappears in the concluding remark. The penultimate sentence says “the world can ‘force’ us to revise our beliefs.” So there is one point of contact or friction there. But two sentences before it we read, “the ultimate point of contact of our discursive practices with the world around us is established by practical inferences and the success of the actions thus arrived at.” So, is the claim that the only point of friction or contact is on the practical side or is it that that is just one of them? If the claim is that that is just one way in which the world can force itself in on us, I do not object to it at all. I think it can happen theoretically, as in the acid case, and I think it can happen practically. I do not think one can define truth, and so truth conditions, and so meaning, exclusively from the practical case—for that you would need the offending principle R—but the paper does not claim that that is so. I see the processes that lead us to find ourselves with incompatible commitments (that is, commitments that are incompatible by our own lights), which is when the world is telling us you cannot have the concepts and commitments that you have got, occurring both on the doxastic or theoretical side, and on the practical side. So if that is the claim being made, then there is nothing at issue between us. But that is much weaker than saying “the ultimate point of contact between our assertional practices and the world around us is established by practical inferences.” On the other hand, I don’t see that the paper actually argues for the stronger claim either. It seems to me that what is argued for is that there can be constraints from the practical side. So I invite clarification on this point. But it looks to me as though there is not any disagreement between us on the large point. I do have a question about specifics in the final section of this paper. We are looking at possible defeasors of which we are ignorant. And the strategy is to look at implicit commitment to these. What I worry about is that the nor-
Replies
173
mality conditions in the set N are not necessarily ones that we know. We have expanded the definition of N so that it is true that anything that could defeat the inference—any p that could defeat the inference—has to entail the negation of one of these sentences. Now, it is clear that that set is going to be infinite at least. My worry is that it is indefinite—that there just is no fact of the matter about how many different ways there are that the inference could go wrong. The list of things that could go wrong with the cookie case was supposed to show that it looked as if you could just keep adding these things: yes, maybe something has shrunk them to microscopic size; maybe something has encased them in glass; maybe something has gravitationally connected them to something else, and so on. If it is really an indefinite class, and anyway an infinite one, then what are we performing when we carry out the factoring that the authors want to do? For each of these at least infinite things we have got to try to get it separated from all of these other infinite and indefinite things, where there is going to be a whole other infinite set of defeasors for the inference when we pick some one element of it. And we have got to test it in a couple of different circumstances at least, each of them with another infinite set of possible defeasors for it. It looks to me as though this is an infinite task, and anyway one that spreads out in ripples throughout the whole web of belief. I question whether it is intelligible that one could do the factoring that is required here. I do not see why one has to be committed to doing that factoring if one is not trying to define truth that way. But the task of assigning responsibility to any one of these things seems hopeless. It looks to me as though Quine was right: you can move the bump in the rug around to various places. I do not see that there is a procedure that could in principle pin it down to any one of these things, to the exclusion of others.
REPLY TO “ARE FUNDAMENTAL DISCURSIVE NORMS OBJECTIVE?” LAUKÖTTER, PRIEN, SCHEPELMANN, THEIN This is a very clearly argued paper and I think it raises a deep and fundamental issue. It is an issue that I’ve struggled with, and perhaps a manifestation of
174
Robert Brandom
that is that the issue keeps coming up, not just in the context in which the paper originally raised it, in Making It Explicit, but also as an explicit topic in some of my writings on Hegel, and again in a different form in some of the Locke Lectures. So this is something that is deep and important, and I by no means think I have seen my way to the bottom of it. What you find in these different places is me trying out different lines of thought about this. I think the paper is dead right that there are two different distinctions: one distinction between the inferential norms and these fundamental discursive norms that form something like a framework in which the inferential ones make sense and get some sort of a grip; and, at least within the inferential norms, a second distinction between objective norms and the norms about which the community has dispositive authority. (Small quibble: I’d rather not call the latter norms ‘conventional,’ since it suggests the existence, somewhere in the background, of an explicit convention that’s been agreed to. And I would just rather not raise that distracting possibility.) I say that the distinction between objective and social norms is clear in the first instance for the inferential norms because the notion of objectivity that is articulated in the second part of Making It Explicit really has three parts. One of them is the attitude-transcendence of both doxastic and inferential commitments; that is the one that the paper carries over and addresses for the fundamental discursive norms. But to call these norms “objective” instead of simply “attitude transcendent” is meant to invoke two other dimensions. First, the commitments they govern are about something, and that notion of “aboutness” is explained in terms of the difference in social perspective between acknowledging or undertaking commitments on the one hand and attributing them on the other. That in turn gets made explicit in the form of de re ascriptions of propositional attitudes, which is where intentionality (and so words like ‘of’ or ‘about’) get their grip. So we need that story. And second, they are about objects, which are given authority over the correctness of these things. Not only is the authority not constituted by the attitude of the practitioners, it’s invested in the objects. And so we need the account of the role of substitution inferences in getting some semantic touch with objects, and then an account of the interaction of these inferences with the aboutness that’s made explicit in de re ascriptions of propositional attitude.
Replies
175
The objectivity that is really the topic of the second part of Making It Explicit has all three of those aspects. And it seems to me that only the first aspect, the attitude transcendence, carries over as an issue about the fundamental norms. They are not supposed to be about anything, and there is no granting of authority to anything else in their case. Nonetheless, it is fair enough to ask about the attitude-transcendence of the fundamental norms that constitute or articulate the constitution of the game of giving and asking for reasons—the fundamental discursive practice. (Again, parenthetically, I actually think that the third of the three examples offered has a different status from the first two. I think of ascription of beliefs in the de re mode as just a special application of substitution inferences—substitution inferences involving terms in commitments I attribute and terms in the expression of commitments that I undertake. But still, it is just substitution-inferential, I would say. But this is clearly not true of the other two sorts of norms discussed—the obligation to justify, and the obligation to remove incompatibilities. One could probably mention other constitutive features of the deontic score keeping model: that each practitioner is supposed to keep score in a way that removes entitlements wherever incompatibilities are attributed, that extends entitlements whenever good inferences hold between commitments that are attributed and those that might be attributed, and so on. So each feature of the rules for keeping score is going to correspond to a fundamental discursive commitment that belongs in a box, I think, with these others.) A couple of remarks in the paper suggest that the obligation to do these things is being thought of in terms of the motivations of the speakers. How does the norm get a grip on somebody so that he feels obliged to justify his claims, to act so as to rectify incompatibilities? And so on. I am not sure whether that is true of the paper or not. If it is, I think it is an irrelevance and a distraction. All the norms are supposed to be norms of assessment. The sense in which they have a grip is that one is liable to be assessed having correctly or incorrectly done things. What practitioners actually do or are motivated by is really neither here nor there. So the question for these norms is when it becomes liable in the eyes of a scorekeeper for assessment as having acted appropriately or inappropriately.
176
Robert Brandom
All of that is by way of stage-setting for addressing the fundamental question: What is the status of these fundamental discursive norms? Are they objective in the sense of attitude transcendence? (Again, I would not say that they are reflective of features of the world. I would deny an “objective pragmatism” about these norms—that you could read them off of the world.) Or are they, subject to the caveats mentioned above, conventional or social norms: a matter of what the community does? And here I do want to have things both ways. I want to say they are socially instituted and attitudetranscendent—but attitude transcendent in a very different sense from the way the inferential and assertional norms within the game are. One could think of them as conventional, but because they are constitutive of playing the game of giving and asking for reasons, to take someone to be a discursive practitioner is to take her to be subject to these norms. If you do not take someone to be subject to assessment as having acted inappropriately if they are unmoved by finding themselves with incompatible commitments, then you are simply taking him not to be playing the game of giving and asking for reasons. Now there is no objective reason why one has to play that game. One of Sellars’s bon mots is his observation that one could always simply not speak—but only at the cost of having nothing to say. If you are not in the game, then you cannot think or whistle those thoughts either. So I think of those norms as in one sense only conditionally applicable. You do not have to be in the community. There is nothing about the world that makes you subject to these norms. Non-discursive creatures are not subject to them, and the world does not oblige us to be discursive creatures (though our biology may all but do that). But to be one is to be subject to these norms. These norms are unconditional given that one is playing the game of giving and asking for reasons at all. So if one can have deontic attitudes at all, then one is subject to these constraints. In that sense they are attitude-transcendent. But in another sense they are optional. Now what language we use to describe that seems to me a nice question. It is perfectly true that I do not discuss that issue in Making It Explicit. And, as I say, I have really tried out different ideas about it in other places. Let me just mention one reason I think it is important how we talk about the status of these discourse-constitutive norms. A prominent strategy for understanding moral normativity (a strategy that we certainly see in Kant and that we see in a
Replies
177
different form in Hegel) says that what makes moral norms unconditionally binding in the sense distinctive of moral normativity is that they are, as I would put it, commitments that are implicit in being a discursive creature at all. Merely in having thoughts, having intentions, even selfish, merely instrumental heteronomous intentions, desires, and thoughts, one has—whether one realizes it or not—implicitly committed oneself to act in certain ways. That is, one has subjected oneself to assessment according to these fundamental discursive norms. And if that is right, then there is a way to vindicate the bindingness of these norms by exhibiting that one is always already implicitly committed to act in these ways. I am not sure that that is the right story to tell about moral normativity, but the fact that there is this way of understanding that important sort of normativity in terms of the fundamental discursive norms shows what the stakes are for understanding properly the status of these things.
REPLY TO “REALIST AND IDEALIST INTERPRETATIONS OF BRANDOM’S ACCOUNT OF OBJECTIVITY” POHL, ROSENHAGEN, WEBER For thirty years, anyway, people have tried to give counterfactual theories of causation. I take it that all of them have run aground on the difficulty of overdetermination: cases of causation where the appropriate counterfactuals do not seem to be true because, even though this event did cause the later event, it is not true that if it had not, the later event would not have occurred. There is a very sophisticated literature about this, most of it downstream from David Lewis’s discussion. But I certainly do not know how to fill the gap between making sense of counterfactuals and making sense of causation. (Some people—I guess the first one I know who argued this way was Bertrand Russell—point out that the notion of causation does not play any role in fundamental physics: we just talk about the evolution of various systems according to equations. And some people have been tempted to take that as a measure of the maturity of our scientific understanding of various domains, and sug-
178
Robert Brandom
gested that in the end we will have a causation-free understanding of whatever it is we understand. I have no idea whether that is true or not.) The official boundaries to Making It Explicit—to what is already a long and complex book—are set by the desire for expressive completeness. That is, I want to give an account of the inferential roles of all of the expressions you need to express the theory of inferential roles and normativity in the book. And here the key place where notions in the vicinity of causation come up is indeed in the discussion of the RDRDs, the reliable differential responsive dispositions. Now I think that if one helps oneself to counterfactuals, one can get everything one needs to discuss the RDRDs. As I understand it in Making It Explicit, reliability is explicitly an inferential notion. The reliability inference is what we endorse when we take a reporter to be reliable. To take somebody to be reliable is to endorse the inference from his saying, for instance “that’s red,” to its being red: the inference, in other words, from attributing the responsively acquired commitment to the scorekeeper’s acknowledging it herself. There remains the issue of the responsive disposition: that is, the application of the concept, or the use of the word, is supposed to be a response to a stimulus. Well, response and stimulus talk is causal talk. I think we can probably understand that in terms of counterfactuals and second-order counterfactuals. Here, we’d have counterfactuals connecting the occurrence of the stimulus and the occurrence of the response. That is a first-order counterfactual, or counterfactual-supporting connection. Why do we need the secondorder ones? Again looking to the wisdom that’s been gained in the discussion of counterfactuals and causation, we need to look to second-order counterfactuals in order to get the asymmetry within the relation between the length of the flagpole and the length of its shadow. It is the length of the flagpole that is causing the length of the shadow. So I would hope to understand the reliability in terms of first-order counterfactual-supporting claims, and understand the responsiveness in terms of first- and second-order counterfactuals. But I do not have a worked-out theory of that. This is a topic that my hero Sellars agonized about in his 1957 paper, “Counterfactuals, Dispositions, and Causal Modalities,” which I discussed in a recent paper (not yet published, but available on my website), called “Pragmatism, Inferentialism and Modality in Sellars’s Arguments Against Empiri-
Replies
179
cism.” Of the stuff that was accessible to the authors of this paper, the treatment of modality at the end of the Locke lectures (which is in exactly the sort of idealist terms that you discussed in Section 3 of the paper under discussion here) is as far as I have gotten in understanding the counterfactual robustness that is involved there. So all of that is only to second the things that are said in the paper, both the diagnosis of the situation and the recommended therapy—a worked out inferentialist account of causation. But I do not have one in my pocket.
REPLY TO “COMPARING BRANDOM’S CRITICAL READING AND HEGEL’S SYSTEMATIC ENTERPRISE” FALKENROTH, KARAKUŞ, SCHWEIKARD I appreciate this careful and thoughtful assessment of one program I have for reading Hegel. I suppose the most general thing to say is that I am not opposed to making something of the ontological dimension in Hegel. I just do not know how to do it. I cannot myself find a philosophical way forward from the ontology, where that aspect of the system is thought of as distinct from the semantics. I think that whatever there is of philosophical interest in ontology typically is to be found in semantics. This was true of Aristotle when he invented both semantics and ontology, and it is still true. What seems right to me about the third Kripkean phase of the modal revolution, the revolution of Naming and Necessity, is once again that we think about metaphysics in terms of semantics—that we start with the use of expressions, and think about the meanings of them. But, that is all a confession of incapacity on my part rather than an expression of confidence that there is nothing to be found in the ontology that cannot be found in the semantics. That having been said, I found two more specific issues in the paper: one about how one thinks of the relation between the project of the Phenomenology and the project of the Science of Logic, and the other about what one says about the logical concepts as articulating the structure both of thought and of the world. Let me address those in order.
180
Robert Brandom
Probably it was injudicious to say it was the same system that is at issue both in the Phenomenology and in the Logic. Of course, they are both parts of the same system, and in that sense the same system is at issue in these two works. But the thought I was clumsily trying to express is that there is a common view about how things work—a view about how concepts work, what makes them contentful and how we ought to understand that—that is arrived at under the heading of “Absolute Knowing” in the Phenomenology and under the heading of “the Idea” at the end of the Science of Logic. It is the conclusion of the books that I think is common to them. I see them as taking disparate roots to an exposition of that same final understanding of us, of the world we inhabit, and of the concepts that articulate the distinction and relation between those. That is perhaps not happily characterized as saying they expound the same system. Now the difference between them is apropos of the second of the theses cited from my paper “Sketch of a Program for a Critical Reading of Hegel.” From p. 159: “Hegel thinks that logical concepts are like empirical concepts in that in spite of the difference between their characteristic expressive tasks, the only way in principle to understand, specify, or convey the contents of both sorts of concepts is by a rational reconstruction of a history of their development”—what at the end of the Phenomenology he calls an “Erinnerung” of them. As the paper points out, I disagree with him that that is the only way to understand the logical concepts (though I think he is right on this score about the empirical concepts). But because he thought that, in trying to convey the nature of the logical concepts (and hence I want to say—and this is my second point—the structure both of thought and of the world that thought is about, that is articulated by those logical concepts), he gives us an Erinnerung: a rational reconstruction of a course of development through less adequate, partly misleading, versions of those concepts that culminates in what he takes to be the final adequate understanding of them. In the Phenomenology it is a literally historical development that tracks the actual achievement of this conception—but in a stylized and selective way because it is emphasizing just the progressive movements of thought. In the Science of Logic, I take Hegel to be doing the same sort of thing, but instead of starting from, as it were, the most concrete understanding possible (“sense-certainty” in the Phenomenology), he starts with something like the most abstract understanding one can have (in
Replies
181
“pure being”), and does not pretend in any way to track the actual history of our coming to understand this. This difference is enough to explain the books’ different roles in the system. They are different paths to the same conclusions, but we learn different things along the way. I am claiming that they are two different sorts of genealogical reconstructions of the same picture of how things actually work. My alternative way of presenting the common endpoint of those two books is to look at the story of how ordinary determinate empirical and practical concepts actually do work, and how they serve to relate thought to the world. That is, to look at what they make explicit, rather than coming to understand them in terms of a culmination in a sequence of ever less inadequate conceptions thereof. That is the view that I was trying to convey. It seems to me that the distinctions as they are put in the paper—that there is an epistemology in the Phenomenology, but there is a metaphysics, an objective ontology, on the one hand, and a subjectivity, on the other hand, in the Logic—seem to be pointing to the same aspects of these books. I still find it to be less happy: there seems to me to be every bit as much of a discussion of the metaphysics of subjectivity in the Phenomenology as there is in the Logic. It is not that epistemology is done to the exclusion of thinking about what subjectivity consists in—for instance, in the “Self Consciousness” chapter, in “Reason,” and, indeed, in the “Spirit” chapter. There is an epistemological starting point for the Phenomenology, and that is a different starting point than we get in the Science of Logic: that is as far as we need to go there. Now, as to the second point, the paper distinguishes between, on the one hand, talking about the concepts that will make explicit the structure of determinate ground level empirical and practical concepts, on the one hand, and making explicit the structures of thought and objective being in virtue of which it’s possible for thought to know about how things are, on the other hand. It had better be a feature of the reading I give of the structure of those determinate empirical concepts as conveyed in the metavocabulary of the logical, philosophical, or speculative, concepts, that in order to understand them, one must at the same time be understanding both halves and the relation between the subjective and the objective. I cannot explain here how I think that works without going more deeply into the nature of that final picture of them than is appropriate, but I absolutely accept that as a criterion of
182
Robert Brandom
adequacy for having conveyed the content of those logical concepts. That is clear, I think, in the nature of absolute knowing that I take to be the final form of self-consciousness—of what we have been doing all along in knowing and acting in the world, and so in understanding the relation between these two poles. And, if anything, it is even more explicit in the discussion of the Idea at the end of the Science of Logic. So, it seems to me that I can tell all of that story without doing anything I would characterize as ontology. And if so, so much the better. If there are more good things to be done with the ontology, I will be happy to hear them. What sets off alarm bells for me in the ontological version is the notion of the true or ultimate structure of reality. I can understand that in semantic terms: Hegel thinks that there is one final set of logical concepts that will let us describe how it is possible to say everything that can be said. And this I deny. I think the set of concepts that make explicit different features of that is in principle open-ended and, indeed, is as open-ended as the determinate ground level empirical concepts are. But to talk about the ultimate nature of reality seems to me to be privileging one vocabulary for making explicit the applicability of the empirical concepts in a way I see no basis for. Each metaconcept in this sense genuinely makes explicit some features of the use of ordinary ground level concepts, and I do not see any use for the idea that some of those are deeper or more important than others. And that idea seems to me to be built into the idea of ontology. That is the sense in which one might with justice see my Locke Lectures as ontologically quietist: they are precisely not privileging one vocabulary over another as cutting reality at the joints or getting things right in a sense that the other vocabulary does not. The question always is: which features of the way things are are made explicit by this vocabulary? Which by this other sort of vocabulary? What sort of relations can we understand between these? But it is not a matter of priority. A reflection of this, where I think things may be clearer, is in logic in a narrower sense than the one I attribute to (and in a sense share with) Hegel. It has been a mainstay of 20th century philosophy of logic to ask: which is the correct logic? Is it intuitionism? Is it classical logic? Is it relevance logic? Is it some modal logic? Which is the right notion of negation? Is it intuitionistic negation? classical negation? Who has got “if…then__” right? Is it the two-
Replies
183
valued conditional? Is it the relevance arrow? Is it the intuitionistic arrow? On my generally expressivist view of logic, which is in no small part inspired by Hegel, these are questions without sense. The question about each bit of logical vocabulary is rather “Which feature of the game of giving and asking for reasons does it make explicit?” The two-valued so-called “material” conditional makes explicit the dimension along which we can assess the goodness of an inference, according to which it’s a bad thing about an inference that it have true premises and a false conclusion. It is subject to criticism in that sense, and so if that is not true of an inference, then that is a good-making property. And that is a feature of good inference that is made explicit even by the poor benighted two-value conditional. The intuitionistic conditional makes explicit a sense of “if…then___”—a dimension of appraisal of inferences—according to which it is a good thing for an inference if there is a recipe for turning an argument for the premise into an argument for the conclusion. That is not the only dimension along which one can assess an inference, but it is a significant one. C. I. Lewis’s strict conditional makes explicit a dimension of assessment of inferences along which it is a good thing if it is impossible that the premises be true and the conclusion not be true. How many conditionals are there that make explicit important features of the normative assessment of inferences, in the way these do? Well, how many different dimensions of assessment of the goodness of inferences are there? I don’t know. But a lot more than this handful. And it seems to me that our job as philosophical logicians is to make those explicit. What is the right logic? Well, intuitionism is the right logic for the dimensions of inference that are made explicit by its vocabulary, und so weiter. And I think the same thing is true of logical vocabulary read more broadly, so as to include the other kinds of things that Hegel (and, in a different tone of voice, I) would like to include. And I would say the same thing about the sorts of vocabularies—normative vocabularies, modal vocabularies, intentional vocabularies, and so on—that I’m concerned with in my Locke Lectures. And naturalistic vocabulary, or the vocabulary of observation reports, has no global privilege over any of the others. The question is just “What can you use it to say?” So it is the privileging of one vocabulary over another that makes me suspicious of ontological idioms.
184
Robert Brandom
REPLY TO “BRANDOM AND FREGE” KLEEMEIER, WEIDEMANN I actually think there’s less difference between me and the authors than maybe it seemed from their paper. Let me go through the points section by section. They say that the context principle should not be identified with the Kantian idea of the primacy of judgment. I agree that the principle should not be identified with that idea, but the notion of the primacy of judgment as I am using it follows from the context principle about sense. That is, the sense of an expression consists in its contribution to the determination of the senses of sentences in which it may occur. If one has to think of the senses of everything in terms of the contributions they make to the senses of sentences they occur in, that is the primacy of judgment as I am using it. We should not identify these two things, since there is nothing corresponding to the sense/reference distinction in Kant, and as is pointed out, following Dummett, the context principle involves both sense and reference. So I agree that we should not identify these two claims, but the primacy of judgment that I am interested in follows from the sense version of the context principle. So I don’t see a substantial disagreement there. The second section addresses the relation of Frege to Kant. I agree that Sluga goes overboard in thinking about Frege as a Kantian. So I would not assimilate Frege to Kant. I see him as making different contributions. But there are three respects in which, it seems to me, Frege follows Kant in innovations that Kant has made. One of them is in the primacy of judgment, in appreciating how special what is expressed by judgments is. So for Frege sentences are the only things that pragmatic force can attach to, and you have to understand the senses of all things in terms of the contributions they make to the senses of sentences they occur in. And for Kant, judgments are the smallest units of experience. The second thing is that the two of them share a nonclassificatory view of judgment, where the whole tradition had thought about judging as predicating one thing of another, typically something general of
Replies
185
something particular. And Kant, like Frege, sees that that will not do, because you cannot account for things like hypothetical judgments that way. Kant builds his table of judgments so as to include complex judgments such as hypothetical, negative, and modal forms of judgment. He sees that he needs to have a more general notion of what judging is than just predication or classification. And that is a very important insight in Frege as well. There is no trace of a predicative theory of judgment in him. And, finally, and maybe most centrally for Frege, Kant wants us to think about objects as whatever are represented by a certain kind of representation. So he’s going to pick out particular representations (for him, intuitions) and say that for something to be an object just is for it to be something that can be represented by a representation like that. Now, of course, Frege doesn’t accept the identification of singular terms with sensuous intuitions, but it is absolutely central to his thinking about numbers as objects that we can argue from number words being singular terms, being proper names, to numbers being objects. And that is a form of argument—if they’re represented by the right kind of representation, then they’re objects—that Kant already made. So, while I certainly would not want to assimilate Frege to Kant, these three moves—the primacy of judgment, the non-classificatory theory of judging based on the appreciation of the variety of complex forms of judgments there are, and thinking about objects as whatever is picked out by a certain kind of representation—are the points of contact I would see between them. Sluga sees a lot more: for instance, about the function of reason. Sluga is worried in a way I am not about definitions of analyticity, and so on. Really it is just those three points that I would see as points of contact, and so I would entirely agree about the other differences the paper points out. Now for the Caesar problem. We also agree that Frege didn’t solve it. What we seem to disagree about is how general a problem that is: Does it extend also to notions of sense, reference, and so on? Here I am not convinced by the defense of Frege. The authors say, “Look, we know that Caesar isn’t the sense of any expression because senses are eternal and immutable.” Well, if we knew that senses were eternal and immutable, that would help us with Caesar. And Frege certainly says that they are eternal and immutable. But is he entitled to say that by his own lights? The only context in which we get a discussion of senses is as the result of applying the function sense of to some
186
Robert Brandom
expression. And we know how to tell whether the sense of one expression is identical to that of another. So we have “G(a) = G(b) iff __” form of explanation. How does that entitle him to say anything about these objects—these senses—for instance, that they are eternal and immutable? That is just another manifestation of the problem that he has not settled all of the identities involving expressions of that form. Furthermore, even if he were entitled to say, by his own lights, that senses are eternal and immutable, so are the directions of lines, which are introduced by definitions of the same form. He might get out of the Caesar problem, but now he would have the “direction” problem for senses: how could he be entitled to say that the sense of the expression was not the direction of any line? Those are eternal and immutable as well, but introducing them by the definitions of the form “G(a) = G(b) iff ___” only tells us whether the direction of one line is identical with the direction of another line and whether the sense of one expression is identical with the sense of another expression—not whether the direction of a line is identical with the sense of an expression. And, similarly, is the truth-value of a sentence identical to the sense of any expression? It better not be for him. But the definitions don’t settle it. So maybe it would be better to think of the “Direction of the Line” problem rather than the Caesar problem. But the corrosive effect of the failure to settle all of those identities does seem to extend to his notion of senses, reference, truth, thought, value—all of which are introduced by definitions of that form. The issue in the fourth section of the paper is of course nearest and dearest to my heart: do I get to claim Frege as an inferentialist, as having introduced the idea of thinking of meanings in terms of inferential roles? Now, though it is not exactly the one you mention, my favorite passage for Frege’s inferentialism is indeed from section 3 of [Begriffsschrift] (p. 53 in Michael Beaney, The Frege Reader, Oxford: Blackwell Press, 1997; p. 2-3 in the original 1879 edition). Here Frege introduces the notion of begriffliche Inhalt by saying there are two ways in which the contents of expressions may differ: it may be the case that the one expression, perhaps when combined with other judgments, has all the same inferential consequences [Schlussfolgerungen] as the other, or it may not. The expressions “At Plataea the Greeks defeated the Persians” and “At Plataea the Persians were defeated by the Greeks,” to use
Replies
187
Frege’s example there, differ only in the former way—they have all the same Schlussfolgerungen. Frege calls what’s common to [the two expressions] their “begriffliche Inhalt.” There I think he is identifying the conceptual content with the inferential role. At least he is saying (as I paraphrase it) that two claims have the same conceptual content if and only if a good inference is never turned into a bad one by substituting the one for the other. This is his criterion for identity and individuation for the begriffliche Inhalt: two things have the same conceptual content just in case, when combined with other judgments, they have the same inferential consequences. (Of course, his notion of begriffliche Inhalt is still undifferentiated in Begriffsschrift.) The authors point out properly that the interest of arithmetic for Frege is nontrivially a matter of expressions like “2 + 3” and “5” having the same reference and expressing different senses. He is still pretty confused about that in Begriffsschrift. If you look at the notion of identity in Section 8 of Begriffsschrift, he desperately needs the sense/reference distinction that he is going to use later in order precisely to be able to talk about non-trivial identity claims. I agree that the later Frege, once he has got a firm grip on the the sense/reference distinction, is not an inferentialist anymore. And it may be that the confusions that are still in play in the Begriffsschrift are essential to his inferentialism—so as soon as he stopped being confused about that he stopped being an inferentialist. That wouldn’t make things look too good for inferentialism. That might be true for him. I hope I am not similarly obliged to be confused. But it seems clear to me that an idea he is trying out in Begriffsschrift is to identify two things as having the same conceptual content if they play the same inferential role. It is just that he has got other ideas in play there too—the ideas that will become sense and reference—and he doesn’t yet know how to put them together. When he gets clear about that, the one he decides to give up is identifying the conceptual content with inferential role. I think it is still possible to read the later work as understanding senses as individuated by inferential roles. But the difficulties that are so clearly pointed out involved in mapping the things he says about when thoughts are and are not the same onto inferential roles means that you have to tell a very fancy story about the later work, if you want to do that. At any rate, though I am tempted by that enterprise, I do not claim in Making It Explicit or in the two Frege es-
188
Robert Brandom
says in Tales of the Mighty Dead that you can identify his later notion of thought with inferential role.
REPLY TO “BETWEEN NORMATIVE AND MODAL VOCABULARY” PRIEN I want to talk about a very small point just at the end and then a great big point that I think Bernd Prien is making in his paper. The small point at the end: both the normative and modal vocabularies are supposed to be LX (that is, elaborated from, and explicative of, the practices). So we could draw (and possibly I should, given that there was this remark, have drawn) the arrows coming back down showing the PV sufficiency as well. But I intended to claim both directions—both the PV and the VP sufficiencies. The large question that Bernd is asking here is really important: how could it be that the same practices can both be sufficient to introduce normative vocabulary and sufficient to introduce modal vocabulary? Is this magic or what? This is a big, important question. Let me say how I am thinking about this, not in the form of an argument, but just laying out the ideas that are in the background. The subjective notion of incompatibility is a matter of incompatibility of commitments. That is a normative notion: these are commitments that you ought not have both of at the same time. If you find yourself having a commitment to p and a commitment to q, you have done something wrong, and so you have got to give up one or other, or you have got to do something else. Because commitment to one precludes entitlement to the other, you cannot be entitled to either of these if you are committed to both of them. Rational rectification of one’s beliefs is eliminating incompatibilities: when you find yourself with incompatible commitments, you have to acknowledge an obligation to do something about that. Now when I treat two commitments as incompatible, one of the things I am doing is treating them as being about some common object. Think of the sentences as being simple predications. The claim that A is a cat is not incompatible with the claim that
Replies
189
B is a dog. It’s incompatible with the claim that A is a dog. So taking the cat claim and the dog claim to be incompatible is taking them to be about the same object. A similar claim could be made about an inference from a dog claim to a mammal claim. It does not follow from A being a dog that B is a mammal; from A being a dog it only follows that A is a mammal. So if you take the mammal claim to follow from the dog claim, you are taking them to be about the same object. What is wrong with claiming that A is both a dog and a cat? What is wrong is that it is impossible for one and the same object to be both a dog and a cat. Those are incompatible properties. That incompatibility is objective incompatibility. It is impossible for one and the same object to have two incompatible properties. That is a different notion of incompatibility from the subjective-normative notion of incompatibility that applies to commitments, because it is not impossible for me to be committed to A being a dog and to A being a cat. It is all too possible: we find ourselves with incompatible commitments all the time. It is just inappropriate, rather than impossible. It is also not, in the same way anyway, inappropriate for me to take A to be a dog and for Bernd to take A to be a cat. We disagree, but neither of us is obliged to do something about that in the way that we would be if one and the same subject had those incompatible commitments. So you can think (and this is now the way that Hegel talks about this very point, I think) of a subject as what repels incompatible commitments in the sense that what is inappropriate is for one and the same subject to have incompatible commitments. That is parallel to the way in which we can think of objects: for there to be one and the same object is for it to be impossible for that object to have incompatible properties. How are these things related? Well, the very same practical activity that is treating two claims as incompatible (that is, acknowledging the obligation to get rid of one of them) is taking two properties to be incompatible. When I take it that I or you or any one subject should not have a commitment both to A being a cat and to A being a dog, doing that is taking the properties of being a cat and being a dog to be incompatible. So, it is the very same activity, which, thought of one way, is treating two commitments as subjectively incompatible, incompatible in the way that matters for defining subjects, and taking two properties to be modally or objectively incompatible, incompatible
190
Robert Brandom
in the way that matters for defining objects. It is the very same activity that is doing both of those things. Objects, Hegel says, repel incompatible properties, in the sense that it is impossible, in the objective modal sense, for one and the same object to have incompatible properties. Subjects repel incompatible commitments in that it is inappropriate for one and the same subject to have incompatible commitments. So there is this one activity, which you can think of either as the enactment of, the acknowledgement of, the subjective incompatibility of commitments, or as the acknowledgement of the objective incompatibility of two properties. And the idea in my sixth Locke Lecture is that that is a way of understanding the intentional nexus: subjective and objective incompatibility, what is made explicit by normative vocabulary on the side of the subject and what’s made explicit by modal vocabulary on the side of the object, are each an expression of the very same activity. I am taking these thoughts over from and explicating the two senses in which Hegel is an idealist. On the one hand, he sees that in order to understand what you are saying when you say being a dog and being a cat are objectively incompatible properties (that is, no one object can have both of them), you have to understand what you are doing when you treat a dog commitment as subjectively incompatible with a cat commitment. So he sees the relations of objective incompatibility (one aspect of his notion of determinate negation) and the relations of subjective incompatibility (the other aspect of his notion of determinate negation)—the relations that articulate the conceptual structure of the objective world, on the one hand, and the relations that articulate the conceptual structure of subjects, on the other hand—as two sides of the same coin. You can only understand what you are saying when you say that two things are objectively incompatible by understanding what you have to do in order to treat them as incompatible, namely, acknowledge the requirement that you rationally rectify your beliefs. And we understand both of those in terms of the activity of the subject. This is what I take to be the essence of his idealist insight. So all of that is a long-winded way of saying that the question Bernd is asking is “How could idealism work? How could idealism be true? How could it be that there is one practical activity that is both treating two commitments as subjectively incompatible and treating two properties as objectively incompatible?” Now, Bernd at this point might want to say, “All right, I see that
Replies
191
there is an aspiration to understand the intentional nexus in the terms of this picture of objects as what objectively repel incompatible properties and of subjects as what subjectively repel incompatible commitments. Still how does it actually work, that you get normative vocabulary over here making explicit the subjective notion of incompatibility, and modal vocabulary over here? How do you get something that makes explicit the activity, what you are doing in talking or thinking, on the one hand, and that makes explicit the content you are asserting or thinking on the other hand (for instance, the claim that it is impossible that something both be a cat and a dog)?” In my fifth Locke Lecture, I show how to move from treating two claims as incompatible (having this relation R introduced in just the way that Bernd, filling in between the lines of what I actually say in the lectures, has so thoughtfully explained) to elaborate a full blown modal logic—which you can prove is S5, the standard, plain-vanilla, modal logic. The operator that gets introduced there, the necessity operator, interacts with the other logical operators, conjunction and negation, in exactly the way the necessity operator does. I claim that the fact that you get standard modal logic out of it by itself shows that the operator that’s introduced is necessity. It plays the inferential role of the S5 necessity operator. It is an interpretation to say that that means that the operator is articulating the content side rather than the pragmatic force side, the semantic content rather than the pragmatic force. But we have a way of telling whether something is making explicit a matter of force or making explicit a matter of content. That is the embedding test, usually now called the Geach-Frege test. It comes up in the following way. There have been various philosophical theories of morality that said that what’s really expressed by a word like “good” is not something that has to do with the content. The word “good” does not allow you to attribute a property to something; it only expresses the force. So to say that something is good is not actually to describe it; it is to endorse it or to approve of it. It is to express a certain attitude toward it. Geach asks “How do you tell whether a bit of vocabulary is expressing the force or expressing the content?” I hope this sounds like a version of the question that Bernd was asking. We want to know about this relation R whether it is telling, as a matter of the content, that these two properties are objectively incompatible, or whether it’s just expressing our normative attitude to the commitments. And
192
Robert Brandom
Geach says, “Look, can’t we trade the one off for the other?” He finds the lovely old English word “macarize,” which means to characterize someone as happy. Couldn’t I now say that “happy” is not used to describe things, but when I call someone happy I’m really doing this other thing, pragmatically I’m macarizing him? That is, couldn’t we get rid of that content or descriptive expression, “happy,” and say that there is only this other speech act, macarizing, the way that emotivists want to say that uses of “good” do not express a content, a description of something, but express the force, the attitude toward it? He says, look, we could do this with anything: we could do it with “red,” we could do it with “mass.” What are the rules of the game here? And looking toward Frege, he says actually we have a way of telling. If you look, not at the free-standing use of the sentence, but at the sentence when it is embedded, say as the antecedent of a conditional, then only the content of it matters for what follows from it. For instance, because when I say “Shut the door,” I’m doing something, requesting or commanding someone to shut the door—that’s the force of it—, you can’t say, “If shut the door!, then …” Embedding it as the antecedent of the conditional would strip the force off. I haven’t asserted p when I assert “if p, then q.” So, if it were the case in saying that something is good that I have not described it (that is, the “good” didn’t go with the content, but it was just a matter of force), you ought not to be able to say things like “if that’s good, then I have reason to do it.” But we can say that. And that means that “good” is part of the content. It is not just a matter of force; “good” expresses a descriptive property that is being attributed to something. So Geach proposed that the way you tell whether some expression is an expression of force (your attitude towards the claim, your attitude towards the content, what you’re doing with it) or whether it is part of the content itself, is to look at its behavior when you embed it. And to finish off my long story here, the operator that I introduce in terms of incompatibility and call “the necessity operator” embeds till the cows come home. To show that it is an S4 operator, you have to show that if it’s necessary that p, then it’s necessary that it’s necessary that p. To show that it is an S5 operator, you have to show that if it is possible that it is necessary that p, then it is necessary that p. That is, you have got to be able to embed that operator as the antecedent of a conditional. You have got to be able to show if it is necessary that p, then it is necessary that it is necessary that p. So,
Replies
193
long story short, the operator embeds as a content-expressing operator. It does not act as something that merely expresses the attitude toward these claims. So, it looks like magic that the very same activities can both be taking two claims to be subjectively incompatible and taking two properties to be objectively incompatible, but if you get the thing to embed in the right way, then the bit you are making explicit is the objective content notion. That’s why the long technical appendix to Lecture 5, that shows that you get proper S5 modal logic from this, is important. It is because in the end it is the key move in responding to Bernd’s important question.
REPLY TO “HOW TO KRIPKE BRANDOM’S NOTION OF NECESSITY” GÖCKE, PLEITZ, VON WULFEN I cannot say how happy this work makes me. This paper alone would have made the trip to Münster worthwhile. No one appreciates more than I do how difficult it is to think these things through. And this is just a very interesting approach and way of using the semantic apparatus of incompatibility. What of course I would most have liked from the incompatibility semantics for modal logic is a result, like Kripke’s result, that had a parameter, his accessibility relation, by varying the algebraic properties of which you could generate all sorts of modal logical systems. And I could not figure out how to do that. You seem to have gotten, as you say, B as well as S5 out of it. If there were some way to jigger it to get S4 out of it, that would satisfy all of my aspirations. However, there may be a serpent in the garden here. What I would really like to know, to be sure that you have done what you’ve set out to do here, is to know for sure whether the converse of the B axiom is not validated in your system. If it were, that would be a terrible thing, because we know that if you get the converse of the B axiom, then the system is modally degenerate. That is, there cannot be any contingent truths in it. A reason to worry that that might be true is that I spent a good deal of time working with a variant of the definition of necessity that I have here. According to my definition, to be in-
194
Robert Brandom
compatible with “necessarily p” is to be self-incompatible or compatible with something that does not entail p. Now, if instead of that definition you have “… is to be self incompatible or compatible with something that is incompatible with p,” instead of “does not entail p,” then you don’t get S4 or S5, but you get B and also the converse of B—and so you get a modally degenerate system. So, another question—is: “Is the definition of necessity offered here equivalent to that definition?” It is not equivalent to mine, but is it equivalent to that other one? If so, then it leads to this degenerate system. That is, if this definition were equivalent to that variant of my definition, then it would be true that it validates the converse of B and is degenerate. Of course, it might anyway; that is, it might not be equivalent, but it might still give one the toxic converse of the B axiom. I have no reason to believe that either of these is so, except that the monstrous version of the modal incompatibility semantics is sitting right around the corner. I basically wasted a year of my life on that system. Indeed, I presented it at a meeting, Tim Williamson noticed the problem, and I ended up having to go back and think again. But the one I have now works. So those are the questions I have, and I really hope that the answer to both of those is “no.” But that is what I worry about.