Anthropocentrism in Philosophy: Realism, Antirealism, Semirealism 9781614518495, 9781614517924

Citation preview

Panayot Butchvarov Anthropocentrism in Philosophy

Eide

Foundations of Ontology Edited by Javier Cumpa Arteseros, Jorge J. E. Gracia, E. Jonathan Lowe, Peter Simons and Erwin Tegtmeier

Volume 8

Panayot Butchvarov

Anthropocentrism in Philosophy

Realism, Antirealism, Semirealism

ISBN 978-1-61451-792-4 e-ISBN (PDF) 978-1-61451-849-5 e-ISBN (EPUB) 978-1-61451-947-8 ISSN 2198-1841 Library of Congress Cataloging-in-Publication Data A CIP catalog record for this book has been applied for at the Library of Congress. Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available on the Internet at http://dnb.dnb.de. © 2015 Walter de Gruyter Inc., Boston/Berlin Printing: CPI books GmbH, Leck ♾ Printed on acid-free paper Printed in Germany www.degruyter.com

For Karter, Kole, and Norah

Contents Chapter One: Introduction 1  Anthropocentrism 1  A Glance at History 9  Antirealism and its Varieties 15  Logical Antirealism and Semirealism

20

Part One: Epistemology and Ethics Dehumanized Chapter Two: Three Varieties of Epistemology  Naturalistic Epistemology 29 32  Subjective Epistemology  Epistemology-as-Logic 42

29

Chapter Three: The Property Good 52  Anthropocentrism and Conceptual Analysis in Ethics  The Good and the World 59  The Relevance of the Property Good 62 Chapter Four: Saying and Showing The Good  The distinction explained 66 71  Logic and the World  The World and the Good 74

52

66

Part Two: Metaphysics Humanized Chapter Five: The Role of Language in Cognition  The empirical and the a priori question 85  Philosophical opinions 91  Scientific opinions 94  Language in logical cognition 101

85

Chapter Six: Metaphysical Realism and Logical Antirealism 104  The Logic of Realism 104 111  Antirealism: ontological, cosmological, and logical  The logical structure of the world 116  Frege and Russell on Negation and Generality 122

VIII

Contents

Chapter Seven: Logical Semirealism 130 130  Ineffability  Wittgenstein on Generality 133  Bergmann on Generality 141 151 Chapter Eight: Generic Statements  The Ubiquity of Generic Statements 151 155  Facts, Generic Facts, and Realism  The Irreducibility of Generic Statements 159  Logical experiences 165 Chapter Nine: Facts and Truth 169  Realism and antirealism regarding facts 178  Semirealism regarding facts  Truth 180

169

Part Three: Metaphysics Dehumanized Chapter Ten: I and the World 189 189  The paradox of antirealism  First-person singular pronouns 192  The Self 198 Chapter Eleven: We and the World 203  Consciousness 203  The I that is We and the We that is I  Idealism 212

206

Chapter Twelve: Mind and the World 218  Concepts, Properties, and Universals 218  Solipsism and Pure Realism 225  Philosophical method 232  Philosophy without anthropocentrism 235 Bibliography Index

244

238

Chapter One: Introduction 1 Anthropocentrism Anthropocentrism, the belief that humans enjoy special, central, even cosmic significance, is present in everyday thought as an attitude toward other animals and the environment generally, and in religion as the Biblical teaching that humans alone were made in the image of God. “I am unable to believe that, in the world as known, there is anything that I can value outside human beings, and, to a much lesser extent, animals,” wrote Bertrand Russell.¹ Many think that such anthropocentrism mars our relationship to other animals and the environment, just as egocentrism mars our relationship to other humans. Speciesism, they would say, is no more acceptable than is egoism, androcentrism, or ethnocentrism. Many also think that the anthropocentrism in religion mars our conception of God. They would agree with Spinoza that, contrary to standard religious doctrine, “neither intellect nor will pertain to the nature of God,”² and that “God is free from passions, nor is He affected with any emotion of joy or sorrow.”³ To attribute to God human characteristics such as intellect, will, joy, or sorrow, they would say, is to think of God as a sort of superhuman. These instances of anthropocentrism are well-known and have been amply discussed for centuries. They are not the topic of this book. Its topic lies deeper: the anthropocentrism present, though seldom discussed or even acknowledged, in philosophy, the discipline charged with our most fundamental thinking – about knowledge (in epistemology), goodness (in ethics), and the world itself (in metaphysics). Ethics is commonly understood as concerned with human well-being, even happiness, and epistemology with human knowledge, especially perception. But these are empirical matters, investigated today in psychology and neuroscience, philosophers generally lacking the qualifications or even inclination for empirical research. Ethics and epistemology remain anthropocentric even when concerned only with language, because the language in question is surely human and investigated properly in linguistics and lexicography. In metaphysics, anthropocentrism takes the form of antirealism, the orientation that has dominated philosophy since Berkeley and especially Kant. Broadly understood, it claims that the world depends, at least insofar as it is knowable, on our cognitive ca-

 Paul Arthur Schilpp, The Philosophy of Bertrand Russell (La Salle, IL: Open Court, 1989), 19 –20.  Benedict de Spinoza, Ethics (New York: Hafner, 1963), Part One, Proposition VII, Note.  Ethics, Part Five, Proposition XVII.

2

Chapter One: Introduction

pacities. The claim seems absurd if taken to mean, as it often is, that we, humans, “make the world.” I shall argue that, if properly understood, epistemology is not about human knowledge and ethics is not about the human good despite the fact that we all desire the human good and respect human knowledge, and that metaphysics is not about “us,” despite the tautology that we can know the world only as it can be known by us. My argument will rest not on abstract and often enigmatic philosophical premises but on specific and readily understandable truths. Whatever the nature of the world may be, humans are only inhabitants of it. The world can hardly depend on them. And knowledge of humans, like knowledge of its other inhabitants, is credibly sought only by empirical, evidencebased methods. But philosophy is not an empirical discipline, and its claims are seldom supported with empirical evidence. Philosophers perform no experiments, maintain no labs, use neither telescopes nor microscopes, embark on no field trips. The moral to be drawn, however, is not that philosophers are experts on nonempirical things or facts. If numbers are such things, it is mathematicians, not philosophers, who specialize in them. Concern with human beings, of course, is natural and morally expected of us all. It is a professional concern, however, only for some: neuroscientists, psychologists, psychiatrists, economists, sociologists, anthropologists, demographers, linguists, lexicographers, physicians. Aristotle did engage in biological investigations, but at the time biology was hardly a science. Today it is. Philosophers’ willingness to assume authoritative stands on human beings became especially incongruous when the experimental sciences devoted to the study of humans emerged. For most of the history of philosophy and science, if a topic did not belong in mathematics, astronomy, medicine, or theology, it was dispatched to philosophy – there seemed to be no other place to put it. Even today, in some institutions psychology is called “mental philosophy” and physics “natural philosophy.” But neither is considered part of philosophy, and few philosophers today can claim expertise in psychology or in physics. I shall be concerned here with the ways anthropocentrism has affected epistemology, ethics, and metaphysics, arguing that it has no place in them, that all three should be radically refocused. The same reasoning would apply, directly or indirectly, to the other branches of philosophy – from the philosophy of art and of science to political philosophy and the philosophy of education – but they will not be discussed here. They all depend in part on theories developed in metaphysics, epistemology, or ethics, but also on developments in fields like history, economics, or psychology. Logic is an exception, for reasons to be explained shortly. Suffice it to note here that, insofar as it belongs in philosophy rather than mathematics, it is a part of metaphysics.

1 Anthropocentrism

3

Some may say that not all of human nature is empirical, that humans also have immortal nonphysical souls. But this is a matter of faith, not investigation, empirical or not. Others may say that even if humans have no immortal souls they have nonphysical minds, entirely distinct from both their brains and their behavior. But there has been an empirical science investigating such minds: the introspective psychology of James, Wundt, Titchener, and many others. To be sure, it was largely unsuccessful, though not because its subject matter called for nonempirical investigation – the introspective psychologists explicitly relied on experience, often in collaboration with others, sometimes in “laboratories.” Much the same can be said about continental phenomenology in its early stages, which was a close relative of introspective psychology and was summed up in Husserl’s slogan “We must go back to the things themselves,” back to what we actually find before us, rather than what philosophical or scientific theory, or even common sense, says is there. In Husserl’s later works, and especially Heidegger’s and Sartre’s, phenomenology evolved into a kind of metaphysics, similar to Kant’s transcendental idealism, Hegel’s absolute idealism, or even Nelson Goodman’s “irrealism.” Its chief tenet became that the empirical world itself is in some sense human, “made by us,” as Goodman put it. This was essentially the thesis of antirealism, in a very broad sense of the term that would apply to Berkeley’s “immaterialism” as well as to Kant’s, Hegel’s, and Goodman’s views. Much of this book will be devoted to that thesis. More likely today is to be told that in fact philosophical inquiries are not about human beings, that they really are conceptual or linguistic. They are about concepts or words, not about the things or facts, human or nonhuman, those concepts or words stand for. For example, it would be said, in ethics philosophers investigate the concept of happiness or the use of the word “happiness,” not any facts about happiness, which indeed are usefully investigated today by psychiatrists and pharmacologists, and in epistemology they investigate the concept of perception or the use of “perceive,” not any facts about perception, which for centuries have been investigated by psychologists and in recent years also by neuroscientists. But surely the concepts and words in question are themselves human, not platonic or divine, and thus are part of an empirical subject matter. The investigation of them calls for observation and sometimes experiment – as in psychology, neuroscience, linguistics, and lexicography – not philosophical speculations, intuitions, a priori arguments, analyses, or definitions. That this is so is hardly news. It was powerfully argued more than half a century ago by W.V. Quine in “Two Dogmas of Empiricism,” where he attacked philosophical appeals to meanings. At about the same time Wittgenstein’s Philosophical Investiga-

4

Chapter One: Introduction

tions was published, arguing in part that words in ordinary language are not used in accordance with necessary and sufficient conditions, and therefore that their use cannot be captured in definitions. In the same decade Gilbert Ryle castigated “the confusion between a ‘use’, i.e., a way of operating with something, and a ‘usage’…. A usage is a custom, practice, fashion or vogue…. The method of discovering linguistic usages are the methods of philologists.”⁴ Also in that decade, Chomsky began publishing articles and books that stressed the biological, largely inherited, core of linguistic competence, and urged that the study of language employ the standard methods of scientific research. The traditional claim of philosophy to a distinctive place among the cognitive disciplines has rested on its absolute fundamentality, supreme abstraction, and unlimited scope. In these respects it surpasses even mathematics: one of its topics is the subject matter of mathematics itself. Its scope includes that of physics and astronomy – space, time, and whatever is in them – but philosophy is also concerned with anything that is not or might not be in space and time. Philosophy presupposes nothing and conceals nothing. This is why philosophers court paradox when preoccupied with things as concrete, literally “down to earth,” as humans. The paradox is no less glaring than it would be if they were preoccupied with cetaceans. If some do not see the paradox, the reason presumably is that they are human. Had they been cetacean, they might have been preoccupied with cetaceans. The concern in philosophy with humans is not a trivial consequence of its unlimited scope, of its interest in “all time and existence.” It is not the trivial application to humans of general philosophical propositions, like the application to humans of arithmetic by the Census Bureau or of physics by a pilot monitoring takeoff weight. It is supposed to be a substantive concern. It may be woefully misguided, but it is natural. The reason is obvious. Plumbers or philosophers, we all are humans. We are deeply interested in ourselves and other humans. We see ourselves as the center of the universe even when we know that we are at its periphery. To suggest that philosophy should not be about humans, that it ought to be in this sense “dehumanized,” may seem even offensive. “Dehumanized” is an ugly word, but it does capture literally and succinctly the aim of the drastic change needed in philosophy – freedom from anthropocentrism – just as “humanized” captures much of the current state of philosophy.

 Gilbert Ryle, “Ordinary Language,” Philosophical Review LXII (1951), 173 – 74. See also Gilbert Ryle, “Use, Usage, and Meaning,” Proceedings of the Aristotelian Society Supplementary Volume XXXV (1961).

1 Anthropocentrism

5

A familiar thesis in ethics is that one ought to treat others “humanely,” as ends, not just as means, perhaps even love them. The claims of kings and barons to special dignity were rejected in the past by declaring the dignity of all humans. “Inhuman” and “inhumanity” are standard terms of condemnation, “humane” and “humanitarian” of approbation. Politicians’ handlers try to “humanize” their clients in order to get them elected. The “human condition” is a perennial object of despair but sometimes of marvel. Works of art are often praised for their “human quality.” A favorable review of a recent novel emphasizes its being “deeply human.” Many demand that space exploration be funded only if it leads to cures for human diseases. A recent television commercial announces, “The human element, nothing is more fundamental, nothing more elemental,” and advocates adding it to the periodic table. A distinguished contemporary philosopher writes of the “heart-breaking specialness” of the human.⁵ Indeed, we all think humans are special, and even feel their specialness. But the reason is not that we think we are angelic and thus special in a way that, say, cetaceans are not, or that we are intellectually, aesthetically, morally, etc., superior at least to all other terrestrial life. The plain reason is that we happen to be human. We all desire and seek pleasure, happiness, well-being – for ourselves, for those we love, often for strangers. So, ethics has devoted itself to investigating the human good, even happiness and pleasures, and the habits, actions, and institutions conducive to it, rather than to the good of cetaceans, extraterrestrials, or angels. We all treasure our ability to see and hear, and to remember and think. So, epistemology has devoted itself to investigating the nature and sources of human knowledge, not cetacean, extraterrestrial, or angelic knowledge. That human happiness and human knowledge are empirical matters, belonging today in the subject matter of developed empirical sciences, has been pointed out repeatedly and eloquently by the proponents of naturalism, the dominant orientation in current philosophy. Its most prominent defender, W.V.O. Quine, called it “the recognition that it is within science itself, and not in some prior philosophy, that reality is to be identified and described.”⁶ Quine limited the substantive claims in his “epistemology naturalized” to mentioning the role in cognition of what he called “surface irritations,” but wisely left the investigation of these irritations to biology.⁷ Other naturalists, however, seem to lack the courage of their convictions and continue to engage in “analyz-

 Cora Diamond, The Realistic Spirit: Wittgenstein, Philosophy and the Mind (Cambridge, MA: MIT Press, 1991), 35.  W.V.O. Quine, Theories and Things (Cambridge, MA: Harvard University Press, 1981), 21.  Willard Quine, “Two Dogmas of Empiricism,” From a Logical Point of View (Cambridge, MA: Harvard University Press, 1994).

6

Chapter One: Introduction

ing concepts” or describing the “workings of our language,” as if concepts and language were not themselves parts of nature and thus belonging in the province of empirical science, not philosophy. Anthropocentrism has been fueled by various assumptions, some true and some false. We assume that – with the possible exception of gods, angels, or extraterrestrials – humans alone are “rational,” capable of reasoning. We assume also that only humans are capable of moral and aesthetic judgment, and perhaps that only humans enjoy the moral and political status, “dignity,” of possessing “rights.” The monotheistic religions, which were the home of medieval philosophy and also profoundly influenced modern philosophy, assure us of humans’ unique origin and special place in nature. Some of these assumptions and assurances are matters of faith, e. g., that, though all things were created by God, only man was created in God’s image. Some are essentially scientific but now abandoned, e. g., geocentrism, the Ptolemaic view that the Earth is the center of the Universe. Other assumptions are also essentially scientific but of unknown and perhaps unknowable truth-value, e. g., that there is no intelligent life elsewhere in the Universe. But many do seem to rest on scientific fact, e. g., that no other terrestrial animals equal humans in intelligence. These assumptions not only encourage our special interest in ourselves but also seem to justify it. They help explain the presence of anthropocentrism in epistemology and ethics. But antirealism, the form anthropocentrism assumes in metaphysics, is a purely technical philosophical orientation, exemplified by positions ranging from Berkeley’s “immaterialism” to Kant’s “transcendental” idealism, Hegel’s “absolute” idealism, and recent positions such as Michael Dummett’s “antirealism” and Nelson Goodman’s “irrealism.” The term “antirealism” may be too strong to apply to all these positions, but there is no convenient alternative to it. “Nonrealism” might be one, but it is too indefinite. According to the metaphysical antirealist, reality depends, at least insofar as it is knowable, on our ways of knowing it, our cognitive capacities – sense perception, introspection, intellectual intuition, imagination, memory, recognition, conceptualization, inductive and deductive reasoning, use of language and of other symbolism. The rather stilted word “cognition,” rather than “knowledge,” is needed to refer to the employment of these capacities, which leads to knowledge when successful but to error when unsuccessful. So understood, antirealism allows for the possibility of an unknowable reality, e. g., Kant’s “things-inthemselves,” which is independent of our cognitive capacities, even if, as Goodman claimed, it might not be “worth fighting for or against.”⁸

 Nelson Goodman, Ways of Worldmaking (Indianapolis: Hackett, 1978), 6.

1 Anthropocentrism

7

The essentially Kantian claim that the world is “shaped,” “sculpted,” by our cognitive faculties, and thus that it depends on us, who of course are humans, amounts to the humanization of metaphysics. It rests on the tautology that we have no cognitive “access” to the world except through our cognition of it, that whatever we know, perceive, understand, believe, imagine, or say about the world depends on our cognitive capacities. The world, of course, is cognized also by nonhuman animals, as well as, perhaps, by extraterrestrials and angels, but in order to know or understand and say even this we must rely on our cognitive capacities, if only our imagination and language. Anthropocentrism is paradoxical in all branches of philosophy, though for different reasons. Epistemology and ethics claim expertise about what they must and usually do regard as certain animals. But animals are part of the subject matter of the empirical sciences, not philosophy. Metaphysical antirealism holds, however tacitly, that reality, the whole world – at least insofar as it is knowable – depends on the cognitive faculties of those animals. But common sense – the mature and thoughtful judgment we all share and all theorizing, scientific or philosophical, begins with and must respect even if not accept – finds such cosmological humanism, human creationism, bizarre.⁹ How could the whole world, it asks, depend on some members of one of its planets’ fauna? In the case of epistemology and ethics, anthropocentrism faces only the paradox of implying that philosophers, supposedly the spectators of all time and existence, engage really in zoological investigations. In the case of metaphysics, it faces the paradox of implying that the world itself is zoological. To avoid the former paradox, we need only to redirect our efforts in epistemology and ethics. But to avoid the latter paradox, we must do much more than redirection. We must understand the first-person pronouns used in the formulation of antirealism as impersonal and thus as not referring to humans. This would require a radical rethinking of their role. When used in philosophical contexts like Cartesian doubt or the realism/antirealism debate, which question the existence of the world itself, consistency requires that “I” and “we” are not taken to refer to parts of that world. The rethinking of the role of these pronouns would require also a radical rethinking of the distinction between oneself and the world. To redirect epistemology, ethics, and metaphysics away from anthropocentrism is not, of course, to abandon them. Metaphysics understood as ontology, the listing and description of the most general kinds (“categories”) of entities and the relations among them, would be unaffected. The realism/antirealism

 For a detailed and incisive discussion of common sense, see Noah Lemos, Common Sense (Cambridge: Cambridge University Press, 2004).

8

Chapter One: Introduction

issue is irrelevant to many metaphysical inquiries, especially those in pre-Kantian philosophy. Also unaffected would be epistemology as tasked with the appraisal of certain fundamental but nonformal inferences, for example, from what has been the case to what will be the case,¹⁰ and ethics as the home of nonanthropocentric theories of goodness like Plato’s, Aquinas’s, and Moore’s.¹¹ Although in epistemology and ethics anthropocentrism is natural and understandable, though indefensible, in metaphysics it is unnatural and almost incomprehensible, but at least as defensible as Kant’s transcendental idealism and 20th century antirealism. In epistemology and ethics, we can reject anthropocentrism unqualifiedly, however painful this might be. In metaphysics, such rejection would be a blunder. The reason is that it is present there as antirealism, which is made plausible even if not entailed by a tautology and therefore truth. The tautology is that whatever we know, perceive, understand, believe, imagine, or say about the world depends on our cognitive capacities. So, in metaphysics we face the challenge of finding a way to reject anthropocentrism without rejecting this tautology. For we ought not to return to pre-Kantian metaphysics just in order to avoid anthropocentrism, much as we ought not to return to pre-Socratic philosophy just in order to avoid Platonism. Yet human creationism is hardly acceptable. We must therefore interpret antirealism as making no reference to humans. Hence, the unusual dialectical structure of this book. Part One is devoted to defense of the dehumanization of epistemology and ethics, Part Two to explanation and provisional defense of the antirealist humanization of metaphysics, and Part Three to the dehumanization of antirealist metaphysics. In Part One anthropocentrism in epistemology and ethics is rejected, in Part Two it is defended in metaphysics, and in Part Three it is rejected in metaphysics. There is no need for an explanation of the humanization of epistemology and ethics. The anthropocentrism present in them is easily understood, and so is what motivates it. Not so in the case of metaphysics. Its humanization was strictly a philosophical event, obscure to virtually everyone outside philosophy and even to many professional philosophers.

 See my “Epistemology Dehumanized,” in Quentin Smith, ed., Epistemology: New Essays (Oxford University Press, 2008).  See my “Ethics Dehumanized,” in Mark Timmons and Terry Horgan, eds., Metaethics After Moore (Oxford University Press, 2006).

2 A Glance at History

9

2 A Glance at History The history of anthropocentrism in philosophy is illuminating, though we can devote to it here no more than a passing glance. It was preceded and encouraged by the anthropocentrism already present in religion. According to Genesis, after creating the heaven and the earth, including the earth’s flora and fauna, God created man in the image of God. But Genesis attributed to God actions such as resting from work, speaking, and inflicting punishment. Thus it also appeared to depict God in the image of man. Its anthropocentrism with respect to Creation seemed accompanied by anthropomorphism with respect to the Creator. Cosmological anthropocentrism seemed to lead to theological anthropomorphism. This personification of God became essential to Judaism, Christianity, and Islam. Pre-Socratic philosophy focused on the cosmos – the heavens and the earth, the four elements (earth, air, fire, and water), ensouled animals. The Socratic revolution shifted the focus dramatically. It moved philosophy from concern with all things to concern with just one. Its slogan became “Know thyself!” It did not go unchallenged. The stand most distinctive of Plato’s philosophy was both antipreSocratic and anti-Socratic. The chief concern of philosophy, he taught, is neither the cosmos nor oneself, but rather what lies beyond both: the abstract entities he called Forms (“Ideas,” eidos). Almost immediately, however, Aristotle urged a return to the cosmos, away from both Socrates’ preoccupation with the innermost and Plato’s preoccupation with the uttermost. Medieval philosophers – Christian, Jewish, and Muslim – sometimes followed Plato, sometimes Aristotle, but all took for granted the declaration in Genesis that humans were made in the image of God. Following the scientific revolution in the 16th and 17th century, when Copernicus, Galileo, Boyle, and Newton became the authorities on space, time, and matter, philosophers came to hold, in sharp contrast to their ancient and medieval predecessors, that their concern was only with “minds” and “ideas.” The new physics compelled them to adopt the “new way of ideas.” Anthropocentrism thus became firmly established in philosophy, since it was not cetacean, extraterrestrial, or angelic minds and ideas that attracted the attention of the early modern philosophers. But their anthropocentrism consisted in preoccupation with just one part or aspect of human beings: their minds and their ideas. It eventually led many to idealism, the view that minds and ideas are all that there is. Early in the 17th century, Descartes began his Meditations by arguing that “I exist” was the only truth he could not doubt, presumably referring by “I” to himself, Descartes, a certain human. Such reference, of course, would have been inconsistent, since humans have bodies, the existence of which he still doubted. But after his propaedeutic dalliance with epistemology, and thus with human matters like

10

Chapter One: Introduction

the possibility and extent of human knowledge, Descartes focused on the existence of God and the nature of mind and matter. This focus was metaphysical and nonanthropocentric. It was shared later in the century by Spinoza and Leibniz. But it was vigorously opposed by the British empiricists, who returned to Descartes’ epistemology and thus to anthropocentrism. The titles of their chief works spoke volumes: An Essay Concerning the Human Understanding (Locke), The Principles of Human Knowledge (Berkeley), A Treatise of Human Nature (Hume). In 1781, less than half-a-century after the publication of Hume’s Treatise, Kant rejected in his Critique of Pure Reason both continental metaphysics, which he called dogmatism, and British epistemology, which he called empirical idealism. He described his own philosophy as transcendental idealism, arguing, for example, that space and time are not “actual entities” because they belong only to “the subjective constitution” of the mind as “forms of appearances.”¹² But Kant allowed for what he called “things-in-themselves,” entities independent of that subjective constitution of our minds, i. e., of our cognitive faculties. Transcendental idealism is often described as part of the humanism characteristic of the Enlightenment. But this is misleading, and so is the vague term “humanism.” Like any antirealist position, transcendental idealism does rest on the tautology that our knowledge of the world depends on our cognitive faculties. On the basis of this tautology, Kant went on to assert also that the world as knowable by us depends on our cognitive faculties. But whether there is a world that is not knowable was a further question. So was also the question whether there are parts or aspects of the knowable world that are not knowable.¹³ The latter question makes Kant’s distinction between things-in-themselves and things-for-us appear less unreasonable, and his antirealism more plausible, by allowing that, even if only the things we can know exist, these things may have parts or aspects that we cannot know. Neither science nor common sense need disagree. When Kant wrote that space and time belong only to the subjective constitution of the mind, he was explicitly referring to humans. “We can … speak of space, extended things, and so on, only from the human standpoint,” he wrote, and explained that this is why space is “transcendentally ideal,” though also empirically real in the sense that it is not an illusion, or mere fancy.¹⁴ But when he wrote

 Immanuel Kant, Critique of Pure Reason, trans. Paul Guyer and Allen W. Wood (Cambridge: Cambridge University Press, 1998), B 37 ff.  E.g., Critique of Pure Reason, B xxvii-xxviii.  Critique of Pure Reason, B 42– 44. For the general distinction between transcendental and empirical idealism, see A 369 – 370. Heidegger remarked that Kant’s “transcendental conception” was “possible only on the basis of the subjectivity of man’s essence.” (“On the Essence of Truth,” in D. F. Krell, Martin Heidegger: Basic Writings (New York: Harper & Row, 1977, 118.)

2 A Glance at History

11

that space and time belong only to the subjective constitution of the mind, he could not have meant Kant’s or any other human’s mind. For humans themselves, including Kant, would be inhabitants of the transcendentally ideal space and thus transcendentally ideal. To be sure, they also are empirically real: although Kant held that our knowledge is limited to appearances, he made a sharp distinction between appearance (Erscheinung) and illusion (Schein).¹⁵ Transcendental idealism was thus anthropocentric insofar as humans are empirically real, but it was not anthropocentric insofar as humans are transcendentally ideal. Nonetheless, it has been understood by many as unqualifiedly anthropocentric. So have its immediate successors: Fichte’s absolute idealism, as defended in The Vocation of Man, and Hegel’s absolute idealism, according to which the Absolute achieves self-knowledge through the human mind. We shall find that a nonanthropocentric reading of all three – Kant, Fichte, and Hegel – is not only possible but plausible. To explain this reading would be part of the challenge of dehumanizing antirealist metaphysics, which will be attempted in Part Three. Standard ethics, of course, is not concerned with humans – their well-being, happiness, pleasure, etc. – as transcendentally ideal. It is concerned with them only as empirically real. Kant called such ethics practical anthropology, which he described as an empirical discipline, and reserved the term “metaphysics of morals” for what he considered the properly philosophical inquiry into morality. Standard epistemology also regards humans – their cognitive capacities – as empirically real. Kant did not have a special name for the empirical study of humans’ cognitive capacities, but presumably would have agreed today that it is the task of what we call cognitive psychology and neuroscience. The properly philosophical inquiry into our cognitive faculties, he thought, was exemplified by his transcendental aesthetic and analytic. Nevertheless, Kant certainly thought, as everyone does, that all disciplines – the transcendental aesthetic and analytic as well as the physics of space and time, the metaphysics of morals as well as practical anthropology – are anthropocentric, indeed literally human, in the straightforward sense that they employ human cognitive powers and thus are constrained by the demands and limitations of those powers. Whether we contemplate all time and existence or just snails and tomatoes, we cannot transcend our sense organs and brains, just as we cannot get out of our skins. Understood as the tautology, on which antirealism rests, that our knowledge of the world depends on our cognitive faculties, this truth is not grounded in facts about human beings, just as the truth of

 Critique of Pure Reason, B 69.

12

Chapter One: Introduction

the tautology “All humans are human” is not grounded in facts about humans. Both enjoy the certainty of logic. Antirealism became dominant after Kant in the form of straightforward idealism. Early in the 19th century, Hegel declared that Spirit (mind, Geist) develops or “unfolds” logically (“dialectically”) from “subjective spirit” (individual mental states like sensations) to “objective spirit” (society as exemplified by the family, customs and traditions (Sittlichkeit), the state, and institutions such as corporations and guilds, among which would be today’s academic disciplines), and reaches its fulfillment in “absolute spirit,” the three stages of which are art, religion, and philosophy (in Hegel’s sense of philosophy as the perfect system of knowledge). Thus, seen superficially, Hegel appeared to endorse not only metaphysical anthropocentrism but also a sort of metaphysical anthropomorphism. But his anthropocentrism involved a major epistemological innovation: a dramatic move from earlier philosophers’ cognitive individualism, the view of knowledge as a personal achievement, to what, for the lack of a better term, I shall call cognitive collectivism: the view of knowledge as a social, often literally collaborative, achievement. It is especially evident today that the cognitive disciplines are inherently social and, at least to users of Wikipedia, that so is virtually all cognition beyond the infantile stage. Cognitive collectivism, of course, need have no political implications. The state is not the only “collective.” In the 20th century, Wittgenstein declared in Tractatus Logico-Philosophicus, “I am my world,” seemingly referring, though if so inconsistently, to a human being, namely himself.¹⁶ Heidegger focused his Being and Time on human being, “the inquirer,” using Dasein as a technical term for it,¹⁷ though two decades later he insisted that “[e]very kind of anthropology and all subjectivity” should be “left behind.”¹⁸ Sartre referred in Being and Nothingness to beingfor-itself as realité humaine, which he contrasted with being-in-itself ¹⁹, and wrote the famous article “Existentialism is a Humanism,” to which Heidegger responded in the not less famous paper titled “Letter on Humanism.” While the scientific revolution of the 16th and 177h century motivated philosophers to confine their subject matter to minds and ideas, in the 20th century

 Ludwig Wittgenstein, Tractatus Logico-Philosophicus, trans. D. F. Pears and B. F. McGuinness (London: Routledge, 1972), 5.63. In references to the text, I shall use Wittgenstein’s numerical designations of sentences. All italics, upper-case letters, and parentheses in the quotations will be Wittgenstein’s.  Martin Heidegger, Being and Time (New York: HarperOne, Revised edition, 1962), 27.  Martin Heidegger, “On the Essence of Truth,” section 9, Note, in Basic Writings, 138.  Jean-Paul Sartre, Being and Nothingness, trans. Hazel E. Barnes (New York: Philosophical Library, 1956).

2 A Glance at History

13

minds and ideas themselves became the subject matter of flourishing empirical sciences – first psychology, then neuroscience – and so philosophers retreated further. Instead of the way of ideas, they took this time the way of words, “the linguistic turn,” claiming that their subject matter was language. Sometimes they described their inquiries as conceptual, not factual, but by “concepts” they did not mean the ideas or mental images of early modern philosophy. They meant meanings or uses of words and syntactic structures, the “workings of our language.” But language and words are matters no less empirical than space and time, or human minds and ideas, and they are investigated today by linguists, lexicographers, even computer scientists. Research in them requires meticulous empirical description and fruitful, empirically verifiable hypotheses, not “analyses” and “definitions,” as in philosophical writings. Philosophers have no more special insight into the workings of language than into the workings of matter or the workings of the human mind. Instead of investigating space, time, and matter, the 17th and 18th century philosophers settled for investigating mind. Instead of investigating mind, the 20th century philosophers settled for investigating talk. They remained open to the charge of inquiring into empirical matters. To be sure, philosophers have a good ear for the nuances of some segments of speech, as J. L. Austin famously did, but Austin insisted that such an ear was not a substitute for empirical research.²⁰ In ethics and epistemology some have eschewed the need for empirical investigation by taking “the deontic way,” concerning themselves with what ought to be the case, or is right, justified, or valid, rather than with what is the case. They have thus avoided encroaching on the empirical sciences, which have no interest in deontic matters except to record and explain people’s beliefs about them. The deontic way seems natural in ethics, given the latter’s emphasis on imperatives. But a deontological ethics must ground its imperatives. Kant held that “the ground of obligation … must not be sought in the nature of man or in the circumstances in which he is placed,” urging that “it is a matter of the utmost necessity to work out for once a pure moral philosophy completely cleansed of everything that can only be empirical and appropriate to anthropology.”²¹ He appealed to the nonsensible world of noumena as the ground of the “commands of practical reason.” Most contemporary deontologists do not follow Kant, but they are aware that a ground is needed. A theory in eth-

 J. L. Austin, How to Do Things with Words (New York: Oxford University Press,1976).  Immanuel Kant, Groundwork of the Metaphysics of Morals, trans. H. J. Paton (New York: Harper and Row, 1964), 57.

14

Chapter One: Introduction

ics, of course, may be essentially an elaborate proposal of ways of acting, especially in politics, but to that extent it is not a cognitive contribution. The deontic way is also natural in epistemology insofar as its concerns have to do with the validity of certain inferences and thus are essentially logical. The inferences in question are seldom deductive, and for this reason the need for grounding them is especially evident. Much of epistemology has consisted in the search for such a ground. In contemporary ethics and epistemology the ground is often what philosophers call “intuitions,” meaning usually their felt inclinations about what they would or would not say is morally right or epistemically justified in various, usually imagined, circumstances, perhaps attempting to render them “coherent.” Philosophers who shun the deontic way also often resort to such intuitions, and for the same reason – to secure a ground for their claims. In both cases success depends on the worth of the intuitions. If the appeal to them is merely a way to avoid empirical investigation of the linguistic facts, it is likely to be minimal. In his Foreword to the 2013 edition of Robert Nozick’s Anarchy, State, and Utopia, Thomas Nagel writes of “a belief in the reality of the moral domain, as an area in which there are real questions with right and wrong answers,” but adds that “progress…toward discovering the right answers” could be made by “formulating hypotheses…and subjecting them to confirmation or disconfirmation by the intuitive moral credibility of their various substantive consequences…The method depends on taking seriously the evidential value of strong moral intuitions about particular cases, including imaginary cases.”²² Neither Hume nor Kant, neither Plato nor Aristotle, neither Hegel nor Mill, would have allowed that such intuitions enjoy evidential value. Nagel does not explain how and why they might, though he does say that “[T]hese convictions form the deep common element in…Rawls’ A Theory of Justice, Dworkin’s Taking Rights Seriously, Walzer’s Just and Unjust Wars, Thomson’s “A Defense of Abortion,” and Nozick’s Anarchy, State, and Utopia.“ When combined with antirealism, as it often was, the linguistic turn had a startling implication: If the world is sculpted by our cognition of it and our cognition is sculpted by our language, then the world is sculpted by our language. Indeed, we do find Wittgenstein saying in Tractatus Logico-Philosophicus, “the limits of my language mean the limits of my world.”²³ We also find Heidegger writing in “Letter on Humanism” that “language is the house of being,”²⁴ and

 Robert Nozick, Anarchy, State, and Utopia (New York: Basic Books, 2013), pp. x-xiii.  Tractatus, 5.6.  Martin Heidegger, “Letter on Humanism,” in Basic Writings, 193.

3 Antirealism and its Varieties

15

in “The Origin of the Work of Art” that language is “the happening in which for man beings first disclose themselves to him each time as beings.”²⁵ These assertions, of course, require detailed discussion, some of which will be attempted later. Suffice it here to note that the language to which Wittgenstein and Heidegger were referring presumably was not cetacean, extraterrestrial, or angelic.

3 Antirealism and its Varieties Antirealism is not limited to metaphysics. There is also antirealism in ethics, aesthetics, philosophy of science, philosophy of mathematics, and so on. Even in the case of metaphysical antirealism, numerous qualifications, distinctions, and explanations are needed. No metaphysical antirealist denies the reality of everything, just as no metaphysical realist asserts the reality of everything, including, say, the Easter Bunny. The solipsist says, “Only I exist,” not “Nothing exists.” Berkeley denied that there are material objects, he called them “stupid material substances,” but he insisted on the existence of minds and their ideas. According to Kant, as we saw earlier, material objects are “transcendentally ideal,” dependent on our cognitive faculties, but they are nonetheless “empirically real,” not mere fancy. Bertrand Russell distinguished between existence and subsistence: some things do not exist, yet they are not nothing – they subsist; for example, material objects exist but universals only subsist. According to Wittgenstein’s Tractatus, some things cannot be “said,” i. e., represented in language, but they “show” themselves in what can be said. Among them, he held, are those that matter most in logic, ethics, and religion. Metaphysical antirealism rests on the proposition that our knowledge of the world depends on our cognitive faculties – perception, conception, language. This proposition and its antirealist implications are not unfamiliar or even wholly unacceptable to science and common sense. We know that dogs can hear sounds we cannot and that eagles can see things when we could not. We take for granted that things look very different to a fish or a fly. If there are intelligent extraterrestrial life-forms, their sense organs, concepts, and modes of reasoning presumably would be fundamentally different from ours, and so would be the world as cognized by them; space scientists puzzle over what pictures or symbols to engrave on space probes that might reach such life-forms. Russell tried, somewhat naively, to explain Kant’s transcendental idealism by saying: “If you always wore blue

 Martin Heidegger, “The Origin of the Work of Art,” in Poetry, Language, Thought (New York: Harper&Row, 1971), 74, trans. Albert Hofstadter.

16

Chapter One: Introduction

spectacles, you could be sure of seeing everything blue.”²⁶ Neuroscience unambiguously holds that our brain is not a tabula rasa on which the senses just write information – it processes that information. Before the insurance payout to the leaseholder of the two World Trade Center Towers, it was necessary to decide whether their destruction on 9/11/2001 counted as one or two events, a decision obviously not dictated by any facts in the world.²⁷ Our beliefs about the history of the United States depend on what history teachers have told us and what history books we have read. The difficulty, sometimes impossibility, of faithful translation from one language to another is notorious, especially in the case of poetry. In science, “the observer effect” is a familiar phrase, meaning the changes that the act of observing produces in what is observed. What is expressed in mathematical symbolism may be inexpressible in a natural language, and vice versa. We often find a picture of a happy child’s face or of a city devastated by a hurricane worth more than a thousand words, and a caricature sometimes tells us what no text could. Educated people do find the idea that the physical world is totally unlike what physics and astronomy tell us idle, if not empty. What the world might be “in itself,” i. e., as neither known nor knowable, does seem to many, as it did to Goodman, not worth fighting for or against. Of course, as a philosophical doctrine, metaphysical antirealism deserves our interest only if argued. But the standard arguments have been densely abstract, often enigmatic, and seldom persuasive. Even if, as Berkeley held, we cannot know the objects we perceive unless we perceive them, it hardly follows that they cannot exist unperceived. Kant argued that we can know things only as they are “for us,” not as they are “in themselves,” but his premise – that for knowledge to be possible, the objects of knowledge must conform to knowledge, rather than knowledge to its objects – is hardly less obscure or more plausible than his conclusion. Perhaps, as Michael Dummett argued two centuries later, a realist interpretation of a sentence requires understanding what would count as its conclusive verification or its conclusive falsification, and such understanding is possible in the case of few if any sentences. But this is comprehensible only to a few professional philosophers, and even they seldom find it persuasive. Hilary Putnam argued for one of his several versions of antirealism by saying that it “does not require us to find mysterious and supersensible objects behind our language games that we actually play when language is working.”²⁸ But the objects behind our language “games” need not be supersensible, and supersen Bertrand Russell, A History of Western Philosophy (New York: Simon and Schuster, 1972), 707– 708.  Steven Pinker, The Stuff of Thought (New York: Viking, 2007), 1.  Hilary Putnam, Ethics without Ontology (Cambridge: Harvard University Press, 2004), 21– 22.

3 Antirealism and its Varieties

17

sible objects need not be mysterious (love and hatred are familiar but they are not objects of the senses, whether as events or as dispositions). A clear and plausible defense of antirealism, I suggest, must bypass the standard arguments. It must start afresh, from specific and readily understandable truths, not abstract and vague philosophical assumptions. The reasoning underlying the standard arguments for antirealism, in Berkeley’s, Kant’s, as well as its recent versions, may be sketched as follows: (1) We cognize only what we can, i. e., have the capacity to, cognize. This is a tautology. But (2) that there is no reality, no world, independent of our cognitive capacities does not follow from (1). What follows is another tautology: (3) that we cannot cognize reality independently of our cognitive capacities. Antirealists often argue on the basis of (1) for (2) probably because the negation of (2), namely, Kant’s view (4) that there is a reality, “things-in-themselves,” which is independent of our cognitive capacities, seems to them idle. But there is at least one very good reason for accepting (4), namely, that (2) seems to imply human creationism, the proposition, presumably held by no one, that the whole world – from the page you are reading now to the outermost known galaxies, and since the Big Bang until the farthest conceivable future – depends for its existence and nature on the minds, cognitive capacities, of humans, a certain species of animals on one of its planets. Because of their forbidding level of abstraction, the standard arguments leave unclear both what they claim and what motivates them. I shall not ignore these arguments, but my focus will be on certain specific and readily understandable truths that lead to antirealism. Arguments for antirealism from such truths have the following form: (1) we cognize (perceive, understand, describe) the world as necessarily having a certain uncontroversial and familiar specific feature. But it is obvious that (2) the world does not, perhaps cannot, have that feature. Therefore, (3) the world as we cognize it, as it is “for us,” is not as it is “in itself.” The major defenders of antirealism, from Kant to Goodman and Putnam, offered also arguments of this second kind. In defense of his obscure thesis of the ideality of space, Kant pointed out that we can imagine only one space, and that we can imagine it as empty but not as absent. Regarding the ideality of time, he noted that all objects of sense, outer and inner, are necessarily in time, and that time is necessarily one-dimensional. Regarding the ideality of causality, Kant argued that we necessarily conceive of the objects of sense perception as causally related even though we do not perceive causal relations. Goodman dazzled his readers with examples of features of the world that are best understood as “made” by us, as how we perceive, conceive, or represent the world in language

18

Chapter One: Introduction

or in art, not as how it is in itself.²⁹ We see the sun rising in the east, majestically moving overhead, and setting in the west, but if educated we know that it is we, not the sun, that is moving. The “fairness” of samples is a sacrosanct requirement both in science and business, but there are no objective criteria for it. We often see the world, at least briefly, as radically different after we watch a film, read a novel, or hear a symphony. We conclude at time t that all emeralds are green because we have observed only green emeralds, but the same observations support with equal logical legitimacy also the conclusion that all emeralds are grue, if “grue” applies to all things observed before t just in case they are green, and to other things just in case they are blue. We reach the former conclusion because “green,” not “grue,” is “entrenched in our linguistic practice.” Putnam pointed out that we can “count the objects in a room (a lamp, a table, a chair, a ballpoint pen, and notebook) and come up with the answer ‘five’,” but that if we also count their mereological sums and ignore the null object then we come up with the answer ‘31.’³⁰ Such examples were often their most persuasive arguments for antirealism. The specific and readily understandable truths on which I shall focus are quite different and less impressionistic. The first is that, as Wittgenstein pointed out in the Tractatus, the world, if there is one, is the totality not of things but of facts, in the robust Russellian sense of “fact.”³¹ (Its ordinary sense, in which many speak even of “false facts,” is too vague to be of philosophical value.) Most philosophers today would deny that there are such entities as Russellian facts. So would common sense, since they are supposed to be entities categorially different from the individual objects, properties, or relations familiar to common sense. Nevertheless, the category of fact is essential to realism regarding the world, which following etymology I shall call cosmological realism, even if not to realism regarding individual things, which, again following etymology, I shall call ontological realism. The reason is simple and obvious. If Jack admires Jill but Jill does not admire Jack, what would distinguish the world in which this is so from the world in which Jill admires Jack but Jack does not admire Jill, the world in which they admire each other, and the world in which neither ad-

 Ways of Worldmaking, especially chapters I and V.  Hilary Putnam and James Conant, Words and Life (Cambridge: Cambridge University Press, 1995), 308.  The phrase “state of affairs” is often used as a substitute for “fact.” A distinction between actual and possible states of affairs is made, true statements supposed to correspond to the former and false statements to the latter. “Fact” then, if used at all, is applied only to actual states of affairs. For the sake of terminological simplicity, I will use only “fact” and ignore the question whether false statements correspond to anything.

3 Antirealism and its Varieties

19

mires the other, if these worlds differed in no other respect? There would be no answer if we supposed that there are only individuals, properties, and relations in the world. Only the fact that Jack admires Jill but Jill does not admire Jack, not their mere presence in the world, including even the presence of a relation of admiration, would distinguish that world from the other three. Hence the conclusion that the world is the totality of facts. There may be no such entities as facts, but then there would be also no world, and cosmological antirealism would win by default. Realists cannot consistently hold both that there is a world and that there are no facts – robust, brute facts. The second truth on which I shall focus is the nonexistence of “logical objects,” in Wittgenstein’s sense, to be explained in the next section.³² We shall find that facts also should be counted as logical objects. Common sense would readily agree that there are no such entities in the world as ifness and allness, though it would be surprised that anyone should think otherwise. Indeed, very few philosophers would disagree. The third truth on which I shall focus is the obvious absence from the world of facts that might correspond to what linguists call generic statements, usually of the form “Fs are G,” as contrasted with universal statements, usually of the form “All Fs are G.” Missourians believe that Iowa winters are severe, but not that all are. Physicians believe that patients with prior strokes benefit from taking Lipitor, but they do not believe that all patients with prior strokes benefit from taking Lipitor, nor do they confuse what they believe with their evidence, say, that in a clinical trial 265 or 11.2 % of the patients who took Lipitor suffered a stroke over five years, while 795 or 37 % of those who took a placebo did. Much if not most of what we think we know about the world is properly expressed only in generic statements. But we do not need abstruse argument to see that no distinctive facts correspond to them, that there are no generic facts, even if there were facts that correspond to the related singular and universal statements.

 Frege had used the phrase “logical objects” for the objects of arithmetic in the context of his project of reducing arithmetic to logic, a project continued later by Russell and Whitehead. Bertrand Russell used it in his posthumously published Theory of Knowledge: The 1913 Manuscript (London & New York: Routledge, 1984), 97, claiming that we must be “acquainted” with logical objects in order to understand “logical terms” such as “particulars, universals, relations, dual complexes, predicates.” Wittgenstein read Russell’s manuscript and criticized it severely. Several years later he asserted in Tractatus Logico-Philosophicus, “There are no logical objects” (4.441).

20

Chapter One: Introduction

4 Logical Antirealism and Semirealism Stated briefly and crudely, the antirealist thesis – that the world, at least insofar as it is knowable, depends on us – would convince no one. It gains plausibility, however, if we agree with Wittgenstein that “there are no logical objects”³³ even though “logic pervades the world,”³⁴ because the world must have a logical, not just spatiotemporal, physical, or causal, structure. Wittgenstein meant that logical expressions, in particular, the sentential connectives (“not,” “and,” “or,” “if… then”) and the quantifiers (“all,” “some”), stand for nothing in the world. Yet, they are indispensable for any cognition beyond that of babes. I shall call all cognition that requires their use logical. It would include but should not be confused with the cognition pursued by logic. Statements expressing logical cognition usually contain also nonlogical expressions, while those in logic do not. “All men are mortal” includes the logical expression “all,” but it is not a statement of logic. Few, if any, statements do not include logical expressions. Hence the power of what I shall call logical antirealism.³⁵ It is a far-reaching version of metaphysical antirealism, yet almost everyone would find it plausible. If a realist interpretation of a true statement is one that pairs the statement with a fact, in the Russellian sense of “fact,” i. e., an entity that “makes” the statement true but is categorially different from anything mentioned in the statement, then almost everyone would find plausible an antirealist interpretation of general statements, whether universal (e. g., “All men are mortal”), particular (e. g., “Some men are mortal,” “At least one man is mortal”), or generic (e. g., “Men are mortal”), and of compound or molecular statements, whether negative (“This page is not blue”), conditional (“If this page is white then the next page is white”), disjunctive (“Either this page is white or it is blue”), or conjunctive (“This page is both white and rectangular”). Few believe that there are in the world universal, particular, generic, negative, conditional, disjunctive, or conjunctive facts, even if they believe that there are “atomic” facts, e. g., the fact to which the statement “This page is white” might correspond. Even fewer believe that there are in the world entities such as allness, notness, ifness, orness, andness, which might be the defining constituents of those facts. Yet, as Sartre eloquently argued, negation plays an essential role in cognition; his striking example was seeing that the person you expected to meet in a café is not there. And Gustav Bergmann pointed out that there would be no laws  Tractatus, 4.441.  Tractatus, 5.61.  See my “Metaphysical Realism and Logical Antirealism,” in Richard Gale, ed., Guide to Metaphysics (Oxford: Blackwell, 2002).

4 Logical Antirealism and Semirealism

21

of nature if generality, meaning universal facts and something the quantifier all stands for that makes them universal, were not in the world, for most laws of nature are universal statements.³⁶ Indeed, facts themselves may be counted as “logical objects,” in the sense that, like the sentential connectives and the quantifiers, they are required by logical, not empirical, considerations. If so, then the declarative sentences expressing them would also be logical expressions. Indeed, we do not learn sentences as we learn names of things, we learn to make sentences. This is why Wittgenstein counted the concept of fact as formal, like the concept of object, and thus unsuited for literal application.³⁷ This is why he denied that there can be representation in language of the logic of facts. If there are no logical objects in the world, then the world has no logical structure and therefore there is no world. Spatiotemporal/physical structures and even individual things may involve logical structure, but whether this so is far from evident. They are not facts and thus involve neither propositional connectives nor quantifiers in the way facts, and the sentences expressing them, do. They can be subjected to chemical analysis but not to logical analysis. Logical antirealism does not deny the reality of spatiotemporal/physical structures or of individual things. It denies only the reality of logical structure, of “logical objects.” This is why it is more moderate and far more plausible than, say, Kant’s or Goodman’s antirealism. It has perhaps the same metaphysical bite, but it does so in a principled fashion. Logical antirealism is not the only species of metaphysical antirealism. There is one corresponding to each species of cognition. The ancient and most familiar version is perceptual antirealism. It denies that there “really” are objects corresponding to what we seem to perceive. Berkeley’s “to be is to be perceived” was its brief but memorable slogan. But perceptual antirealism need not be so drastic. It can be limited to only some features of perceived objects, say, their “secondary qualities,” such as color, and then it would represent the view held by contemporary science and most educated people: we seem to see colors but in the world there is only the light that initiates vision. More innovative is the species of metaphysical antirealism we may call conceptual –objects, perceived or not, depend on our understanding of them, on the concepts we possess. Conceptual antirealism was Kant’s major contribution to philosophy. It can be limited to just some concepts and thus gain greater plau See Gustav Bergmann, New Foundations of Ontology (Madison: University of Wisconsin Press, 1992, ed. by William Heald), 173. Heald’s introduction is obligatory reading for all interested in Bergmann’s philosophy. He has also included an invaluable glossary.  Tractatus, 4.1272.

22

Chapter One: Introduction

sibility. It can be limited to what Kant called the pure concepts of the understanding, such as causality. It can be limited especially plausibly to the concepts expressed by logical expressions, and then it would coincide with logical antirealism. I have repeatedly argued elsewhere that at least the concept of identity cannot be given a realist interpretation (we hardly perceive a relation of identity or suppose that its existence is somehow hidden), but virtually no cognition, not even simple recognition, is possible unless the concept of identity is applicable, even if not explicitly applied. The version of conceptual antirealism that became characteristic of 20th century philosophy after it took the linguistic turn is linguistic antirealism. If cognition necessarily involves language, then insofar as it is cognizable the world is dependent on language, if not on the particular characteristics of the language we speak, then on the characteristics that all languages share. Conceptual antirealism would entail linguistic antirealism if concepts are words or syntactic structures, rather than “ideas in the mind.” And if limited to logical expressions and syntactic structures, it would be also a version of logical antirealism. But not all versions of logical antirealism need be linguistic. We may allow, as Kant did, for the possibility of logical concepts that are purely psychological. Logical antirealism is the most plausible version of metaphysical antirealism. Few would disagree with Wittgenstein that there are no logical objects – no items, no fragments of the world, that correspond to the propositional connectives and the quantifiers, and thus no distinctive facts that correspond to compound and general statements. No such items and facts belong in what can be “said,” and so they would not count as ordinary denizens of reality. But, Wittgenstein also held, they can be “shown.” Wittgenstein’s distinction between saying and showing thus introduced a position that is neither unqualified logical realism, like that of Frege and Russell, nor unqualified logical antirealism. We may call it semirealism.³⁸ In chapter 4 I shall argue that the distinction between saying and showing has a clear and noncontroversial application to ordinary, not just “logical,” pictures, and that Wittgenstein’s picture theory of meaning on which the distinction rested was no more than a sophisticated version of the traditional view that meaning and thought involve mental representations, i. e., ideas, perhaps mental images. Before Kant, the tautology on which antirealism rests – that our knowledge of the world depends on our cognitive faculties – may have led to skepticism. But antirealism should not be confused with skepticism, though they are similar in some

 Semirealism should not be confused with what Simon Blackburn calls “quasi-realism” in Essays in Quasi-Realism (Oxford: Oxford University Press, 1993).

4 Logical Antirealism and Semirealism

23

respects. Both are rooted in the commonsense belief that “things may not really be what they seem to be.” Both appeal to the proposition that “we can never discover what things really are because any discovery would still be a discovery only of what some things seem to be.” The skeptic concludes that we can never know what things really are. The antirealist concludes that there is no distinction between what things really are and what they seem to be. This is why, left unexplained, antirealism is paradoxical, while skepticism is at most outrageous.³⁹ Antirealism is a metaphysical view, skepticism belongs in epistemology. We shall see in chapter 2, however, that much of traditional epistemology can be understood as logic, insofar as its chief concern has been the validity of certain inferences. If the home of skepticism is such epistemology-as-logic, it is no more anthropocentric than is logic. But in its usual formulations skepticism is overtly and unabashedly anthropocentric. It concerns the limits of human perception and understanding. And today the proper study of these limits belongs in psychology and especially neuroscience. Whether certain brain states are the outcome of external stimulation, and if they are to what extent and in what way they represent anything external, is hardly to be determined by armchair speculation. Moreover, if what I called cognitive collectivism is accepted, the epistemological question “What do I know?” would be replaced by the question “What do we know?” and traditional skepticism would become less implausible. We may still hold that there are things that, say, physics does not know, perhaps cannot know, but our reasons would rest on certain facts about the scientific discipline of physics and bear little resemblance to Cartesian skeptical concerns. Metaphysical antirealism would also become less implausible. Surely the nature of the spatiotemporal world insofar as it is cognized is what the disciplines of physics, astronomy, and biology say it is. Metaphysical antirealism enjoys little public celebration, but its indirect and usually unnoticed influence on nonphilosophical thought has been enormous. A noteworthy example is Kuhn’s important account of the history of science as involving relativity to shifting “paradigms.” Less admirable are the fashionable but careless and unphilosophical relativisms that insist that truth is relative to era, culture, race, gender, ethnic origin, or even just personal “belief system.” Indeed, metaphysical antirealism is a form of relativism, but it must not be confused with any of these. It acknowledges only relativity to being human. This is why metaphysical antirealism is not a sort of subjectivism. What our cognitive faculties deliver can count as objective in the straightforward sense that, in principle, it can be and often is the same for all humans, and personal divergence from it is what we count as subjec-

 See my Skepticism about the External World (New York: Oxford University Press, 1998).

24

Chapter One: Introduction

tive. This is why metaphysical antirealism can allow for a sharp distinction between objective truth and personal opinion or fancy. We know that there are fairly advanced nonhuman cognizers, e. g., whales and chimpanzees. There may be also extraterrestrial cognizers far more advanced than humans. The world cognized by whales is relative to cetacean cognition. The world cognized by chimpanzees is relative to simian cognition. The world cognized by extraterrestrials would be relative to their cognition, The antirealist holds that the world cognized by humans is relative to human cognition. That relativity is biologically inescapable. By contrast, relativity to era, culture, race, gender, ethnic origin, or personal belief is not. We cannot literally transport ourselves to an era in the past, but we can and often do transcend the present by, for example, viewing what past architects designed and past builders built, reading what past authors wrote, and today even hearing recordings of what past singers sang. We cannot change our ethnic origin, culture, race, or gender, but we can transcend it by communicating, and often agreeing, with people of a different origin, culture, race, or gender. We can do so because, since we belong to the same species, we share roughly the same cognitive faculties. This is why relativity to era, culture, race, gender, or ethnic origin is quite unlike relativity to being human. As to relativity to personal belief system, it would differ from the triviality that one believes what one believes only if one’s beliefs really constituted a system, if one did not hold contradictory beliefs, and if there were genuine criteria for membership in the system. But don’t we sometimes believe the premises and reject the conclusion of a valid argument? And is the belief, for example, that yoghurt is better than sour cream really part of my system of beliefs? One’s beliefs are more likely to form not a system but just a “multitude”: “Do I contradict myself? Very well then I contradict myself. (I am large, I contain multitudes),” Walt Whitman wrote. I mention in this book, and sometimes discuss in detail, the views of various philosophers, from Plato, Kant, and Hegel to Moore, Wittgenstein, and Sartre. No philosophical discussion should ignore the historical context of its topic. But this is not a book in the history of philosophy. I do not engage in exegesis for its own sake. Some readers may be surprised by the considerable attention accorded to Hegel, Moore, and the early Wittgenstein (often with quotations rather than paraphrase so the readers can judge for themselves what I say about them). But Hegel’s insistence on the cognitive priority of society over the individual suggests, when generalized, the radical rethinking of the distinction between oneself and the world that is needed in an antirealist but nonanthropocentric metaphysics. Moore’s ethics remains the modern paradigm of a nonanthropocentric ethics, and his account of the nature of consciousness is invaluable for the formulation of a defensible antirealism in metaphysics. And Wittgenstein’s distinction between saying and showing provid-

4 Logical Antirealism and Semirealism

25

ed a strikingly original third alternative in the realism/antirealism debate. It has not been sufficiently appreciated, perhaps because, like Hegel’s and Moore’s views, it is found too “obscure” and “difficult.” Several chapters are rooted in previously published articles: “Metaphysical Realism and Logical Antirealism,” in Richard Gale, ed., Guide to Metaphysics (Blackwell, 2002); “Saying and Showing the Good,” in Heather Dyke, ed., Time and Ethics: Essays at the Intersection (Kluwer Academic Publishers, 2003); “Ethics Dehumanized,” in Mark Timmons and Terry Horgan, eds., Metaethics After Moore (Oxford University Press, 2006); “Bergmann and Wittgenstein on Generality,” Metaphysica, Vol. 7, No. 1, April 2006; “Epistemology Dehumanized,” in Quentin Smith, ed., Epistemology: New Essays (Oxford University Press, 2008); “Facts,” in Javier C. Arteseros, ed., Studies in the Ontology of Reinhardt Grossmann, (Frankfurt: Ontos Verlag, 2010); and “Generic Statements and Antirealism,” in Logos & Episteme, I, 1 (2010).

Part One: Epistemology and Ethics Dehumanized How can he who has magnificence of mind and is the spectator of all time and existence, think much of human life? Plato

Chapter Two: Three Varieties of Epistemology 1 Naturalistic Epistemology The subject matter of both epistemology and ethics traditionally has been considered human – the knowledge and the good, respectively, that humans, not cetaceans or angels, seek and enjoy or lack. This, I argued in the Introduction, is a mistake. Its correction calls for refocusing these branches of philosophy, their dehumanization. Such refocusing is more easily accomplished in the case of epistemology. Throughout its history, it has wrestled mainly with issues concerning the validity of certain inferences, hardly a matter to be settled by zoological considerations. In effect, epistemology has often been a sort of logic. In this chapter I shall attempt to make clear how epistemology-as-logic differs from naturalistic epistemology, which celebrates the primacy of zoological considerations, as well as from subjective, Cartesian, epistemology, which is logically incompatible with zoological considerations but thereby also lacks subject matter altogether. The case for dehumanizing epistemology is best understood in the context of the important differences among these varieties of epistemology. (Fundamental disagreements in epistemology, as elsewhere in philosophy, often arise from differences of interest, not genuine conflict.) Naturalistic epistemology is explicitly anthropocentric, humanized; the “natural” objects it considers are not cetacean or simian. To be consistent, subjective epistemology cannot be anthropocentric, though its practitioners are seldom aware that this is so. It would beg the question against the skeptic if it takes itself to be about any one or several humans. Epistemology-as-logic is as nonanthropocentric as logic; it is, of course, a human endeavor, but humans are not its subject matter. All three have been with us at least since Socrates. My chief concern in this chapter will be with epistemologyas-logic, but it is naturalistic and subjective epistemology that have represented the standard conception of epistemology. It may seem obvious that epistemology should be naturalistic. Its name is a synonym of “theory of knowledge,” the knowledge in question surely is that of humans, and humans are parts of nature, of its fauna. Epistemology naturalized is thus epistemology humanized: it is about humans. Not only does it ignore gods, angels, and extraterrestrials, it ignores also chimpanzees, whales, and bats. But thereby it also lacks the supreme generality and abstraction distinctive of philosophy. Humans already belong in the subject matter of several special sciences that seek detailed information about them, including their perception, conceptualization, and reasoning. This is why naturalistic epistemology is only programmatic. The substantive work is done by biology and psychology. As I

30

Chapter Two: Three Varieties of Epistemology

mentioned in the Introduction, Quine, who took up the case for “epistemology naturalized,” often mentioned the role in cognition of what he called “surface irritations,” but did not himself investigate these irritations. Naturalistic epistemology is focused on human matters even when straying into talk about nonhuman biological and computational states. They are of interest to it mainly for the light they might cast on the epistemic faculties and states of humans. The intrinsic interest of such nonhuman states is of course indisputable, but they even more obviously belong in the province of the empirical sciences. Naturalistic epistemology may be only programmatic, but its pedigree is impressive. Much of Aristotle’s epistemology was naturalistic. When he described the parts and functions of the soul, he was doing in principle what biologists and psychologists do today. And the rationale of the program seems impeccable. Humans, obviously, are parts of nature, they are not heavenly spirits even if they possess immaterial souls. But this is also why the proper investigation of them and of their epistemic faculties and states is empirical and belongs in the natural sciences. It would be strange to propose today investigating any part of nature nonempirically. Even if the human epistemic faculties and states were faculties and states of immortal souls, a genuine investigation of them would still be empirical. A human immortal soul is still the soul of a human being, a certain animal. Both Plato and Aquinas would have agreed. Much the same can be said about the suggestion that they are faculties and states of “persons.” Surely the persons in question are animals, not gods. If human beings, including their epistemic faculties and states, belong in the subject matter of disciplines other than philosophy, the obvious question is what room is left for naturalistic epistemology. We saw in the Introduction that concern over this question may explain the shift to the view of philosophy as just “conceptual,” not “factual” – neither about natural facts nor about nonnatural facts, but about concepts or words. Hence its preoccupation with “definitions,” “analyses,” and “elucidations” of the “workings of our language.” But if the concepts and words are in nature – presumably in human minds and languages – they, too, lie outside philosophers’ professional competence: there is psychology and neuroscience, as well as linguistics and lexicography. (If concepts are not in nature, e.g., if they are Platonic Forms, then they should be of no concern to naturalistic epistemology.) The investigation of brain-states and words calls not for “definitions” or “analyses,” to be tested by “intuitions,” but for meticulous empirical descriptions and fruitful hypotheses, tested by standard scientific methods. The very idea of aiming at definitions or analyses of brain-states is foreign to neuroscience. As to words, more than half a century ago Wittgenstein pointed out, as I noted in the Introduction, that they are not used in accordance with necessary and sufficient conditions, and thus their uses cannot be captured in definitions.

1 Naturalistic Epistemology

31

Perhaps the most familiar project in recent epistemology was the search for a definition of knowledge, which preoccupied it from the 1960s through the 1980s. It was born in the late 1950s, when A. J. Ayer’s Problem of Knowledge and R. M. Chisholm’s Perceiving appeared. But the project was out-of-date already at its birth. Thirty years earlier Wittgenstein had written: “If I was asked what knowledge is, I would list items of knowledge and add ‘and suchlike.’ There is no common element to be found in all of them, because there isn’t one.”⁴⁰ Linguists and lexicographers of course agreed. A famous paper by Edmund Gettier, a student of Wittgenstein’s disciple Norman Malcolm, argued the point in the early 60s. But few of those who wrote the thousands of pages devoted to discussion of his paper seemed aware that, whatever its author’s intentions might have been, it called not for greater diligence, sophistication, or imagination in the project of defining knowledge but for its abandonment. Epistemology is the theory of knowledge, but the word “knowledge” stands either for a disciplinary, essentially social achievement, such as grammar (Aristotle’s favorite example), astronomy, and arithmetic, or for a personal achievement. The study of knowledge as a social achievement belongs in the history of science and the sociology of knowledge. Investigation of it would be, of course, naturalistic, essentially historical and sociological. But epistemologists usually have been interested in knowledge only as a personal achievement. And this interest may assume one of two very different forms. I may ask whether, how, and what knowledge is possible for me, a certain human. If so, my epistemological endeavor would remain anthropocentric and therefore would count as naturalistic. It would be objective, though rather narrow in subject matter. But I may ask instead whether, how, and what knowledge is possible for me in abstraction from the fact that I am human and ignoring the question of whether, how, and what knowledge is possible for other humans. This would be the question that a philosophical skeptic who respects consistency would ask, especially if it concerns knowledge of the existence of an “external” world consists of rocks, trees, stars, as well as human bodies such as yours and mine. We may describe an epistemology limited to this question as subjective.

 MS 302, “Diktat für Schlick” 1931– 33. Quoted by David Stern, “Sociology of Science, Rule Following and Forms of Life,” in M. Heidelberger and F. Stadler, eds., History of Philosophy and Science, 347.

32

Chapter Two: Three Varieties of Epistemology

2 Subjective Epistemology While the subject matter of naturalistic epistemology, whether human knowledge or human concepts or language, lacks the generality expected of philosophy, subjective epistemology seems to lack generality altogether. It seems to be baldly and bleakly about only one person, myself, hardly a topic of philosophical or scientific interest, whoever and whatever I might be. But it cannot presuppose, explicitly or implicitly, my humanity, including the existence of my body, my language, and my place on earth or in history, for then it would be a naturalistic, though incompetent, epistemology limited to just one “sample” of humanity. The subjectivity of subjective epistemology does not consist in its being about a solitary human being. The term “subjective” is not a synonym of “mental” or “mentalistic.” If minds are immaterial, my study of your mind might be as objective as my study of my mind. And if I could study my brain only as my brain, the study would be subjective and form no part of biology. Quine’s rejection of Cartesian epistemology on naturalistic grounds was both too narrow and too wide. It was too narrow because what was characteristic of Descartes’ epistemology was not its subject matter, presumably what he called a “thinking thing,” but his exclusive use of first-person indexicals in its defining initial stages. In his argument “I think, therefore I am” Descartes did not argue for the existence of Descartes, a Frenchman who of course possessed a material body even if also an immaterial soul. To be consistent, Descartes could only use “I,” as of course he did. One need not be naturalistically, “scientifically,” minded in order to reject a theory dependent on the exclusive use of first-person indexicals. Quine’s rejection of Cartesian epistemology was also too wide because, as R. M. Chisholm pointed out, in its initial stages, including the proof of “I exist,” Cartesian epistemology was consistent with a naturalistic, even materialist, ontology. ⁴¹ For all Descartes knew when he was initially stating the proof, the pronoun “I” in it might have referred to a material, not a “thinking,” thing. The truth is that, pace Descartes and almost all other philosophers, his cogito had no ontological content at all. Subjective epistemology is not a capricious narrowing of the subject matter of naturalistic epistemology from all humans to just one. If it were, it would not be subjective, though also it would be of no philosophical interest. Subjective epistemology arose, most notably with Descartes, Berkeley, and Hume, as a distinct variety of epistemology in order to face the skeptical challenge. The skeptic

 Roderick M. Chisholm, A Realistic Theory of Categories (Oxford: Cambridge University Press, 1996), 99 – 105. Chisholm speculated that the mind might be “microscopic part” of the brain.

2 Subjective Epistemology

33

cannot assume that he or she is human, since being human involves having a body, which is a part of the material world the existence of which the skeptic doubts. Therefore, the subjective epistemologist also cannot make this assumption when attempting to answer the skeptic. Subjective epistemologists cannot consistently write, even though the best of them did, books titled “An Essay Concerning Human Understanding,” “A Treatise Concerning the Principles of Human Knowledge,” or “A Treatise of Human Nature.” If true to their titles, such books would beg the question against the skeptic. Subjective epistemology is essentially Cartesian, though it was anticipated by the Greek skeptics, especially Sextus Empiricus. Its raison d’être was the project of answering the skeptic. Had it succeeded, its mission would have been accomplished, and there would have been room left only for naturalistic epistemology and epistemology-as-logic. Naturalistic epistemology does not beg the question against the skeptic by taking its subject matter to be humans because it is not concerned with the skeptic’s question; Aristotle was no more concerned with skepticism than was Quine. Indeed, though a “theory of knowledge,” naturalistic epistemology need have little concern with knowledge itself. It is best understood as concerned with cognition, that is, the employment of our perceptual, conceptual, and verbal faculties, which may lead to knowledge but would be of interest even when it does not. For subjective epistemology, however, since it is mainly an attempt to answer the skeptic, only knowledge matters. Alleged cognitive states such as justified belief are at best images of knowledge. We seek them in the hope of finding something still worth having when knowledge is absent or impossible. The skeptic about the external world would not be answered by being told that one would be justified to believe that there is an external world. One reason is that the issue is too important for anything less than knowledge – it concerns the existence also of other people, including those we see daily and love. Another reason is the impropriety of such uses of “justified” and “believe.” As used in epistemology, “justified” is a technical term, of obscure meaning and uncertain reference, indeed often explicitly introduced as primitive. In everyday talk, it is a deontic term, usually a synonym of “just” or “right,” and thus “justified belief” is a solecism. For it is actions that are justified or unjustified, and beliefs are not actions. If told that the phrase stands for belief resulting from “reliable processes,” this would be a verbal stipulation, and would also involve the use of “reliable,” another technical term that has required extensive explanation. Even the word “belief,” whether or not prefaced by an adjective like “justified,” is seldom used in epistemology with clear sense and reference. In everyday talk, it may refer to a behavioral disposition, moreover a multi-track one, manifested in a great variety of events, e. g., in what one says, what one does

34

Chapter Two: Three Varieties of Epistemology

to oneself or to one’s spouse, children, colleagues, friends, and strangers, what one buys, sells, or reads, what one eats or drinks, etc. A proper translation then of “S believes that p” into logical notation would include all sorts of counterfactuals and would be too cumbersome to bother stating. It certainly would have neither the form “B(S,p)” nor the form “B(S,x,y…, R),” where S stands for the believer, p for what is believed, and x, y, and R for constituents of p. If we take belief to be a brain state, process, or condition, then the translation of “S believes that p” into logical notation, were it possible, would need to be provided by neurologists, though they would have little use for it. And if we take belief to be an irreducibly mental state or condition, plain respect for the phenomenological facts would tell us that beliefs so understood are as rare as experiences of faith, commitment, or strong conviction seem to be. The truth is that in ordinary discourse “S believes that p” (e. g., “S believes that Jones owns a Ford,” when uttered in the course of office chatter) functions roughly as a synonym of the colloquial “S thinks that p,” and is no more enlightening or needed in philosophy than is the latter. And “belief” is used more or less as a synonym of “opinion,”view,” “stand,” etc., as in “The belief that ….. originated in…..” Neither the verb nor the noun usually expresses religious faith or some other commitment or conviction that may have psychological reality and thus deserve the interest of psychology and, perhaps, philosophy. For example, it would be absurd for me to say now that I believe that I am typing, let alone that I am justified in believing that I am typing. The word “justified,” whether applied to such phantom beliefs or to the sentences, statements, assertions, judgments, hypotheses, etc., supposedly expressing them, is often prefaced by the adverb “epistemically.” But this renders its use in epistemology even less appropriate because the noun corresponding to that adverb in Greek is episteme and in English “knowledge,” exactly the word we tried to avoid. We resort to the phrase “justified belief” when aware that we have no knowledge but also hope that we are not entirely ignorant. But it is not the natural, traditional word for describing what we might have in such cases. It is the word “evidence” that comes closest to serving this function – in science, courts of law, and careful everyday discourse. The root of “evidence,” however, is the same as that of “evident.” When what is evident is a proposition, it is evident if it is or can readily be seen, whether literally or metaphorically, to be true. “Seeing is believing,” the saying goes, but what it usually means is that seeing is knowing. Hence, the traditional account of knowledge as apprehension, intuition, awareness, or acquaintance,

2 Subjective Epistemology

35

and the existence of the so-called strong sense of “know,” roughly that of certainty, as contrasted with its weak sense of some sort of true belief.⁴² A proposition is sometimes evident because it is “seen” to be true by itself, i. e., to be self-evident. More often it is evident because it is seen to follow from one or more other propositions that are evident.⁴³ We seldom say, however, that the latter are evidence for the former. We seldom call the premises of a valid deductive argument evidence for its conclusion, even if the premises are evident and the validity of the argument is itself evident, as it would be if its form were as simple as, say, that of modus ponens. Rather, we speak of evidence when what we want to know is neither self-evident nor seen to follow from anything evident, yet we think or hope that something else “supports” the proposition in some other manner. It is then that skepticism is born and so is the subjective epistemology that attempts to answer it. J. L Austin wrote, “The situation in which I would properly be said to have evidence for the statement that some animal is a pig is that, for example, in which the beast itself is not actually on view, but I can see plenty of pig-like marks on the ground outside its retreat. If I find a few buckets of pig-food, that’s more evidence, and the noises and the smell may provide better evidence still. But if the animal then emerges and stands there plainly in view, there is no longer any question of collecting evidence; its coming into view doesn’t provide me with more evidence that it’s a pig, I can now just see that it is, the question is settled.”⁴⁴ Some would say, however, that when the animal comes in view it is self-evident that the animal is a pig. Taken literally, of course, the term “self-evident” is a pleonasm, as “self-seen” and “self-visible” would be. But it does serve to mark the important difference between what is evident in virtue of being seen to be true by itself and what is evident in virtue of being seen to follow from one or more other propositions that are seen to be true. Even in a modus ponens argument with self-evident premises, the conclusion, if not self-evident itself, would not be evident unless seen to follow from the premises, i. e., unless the validity of the argument is self-evident. The conclusion would be evident only if its relation to the premises was evident.

 The distinction between a weak and a strong sense of “know” was made by Norman Malcolm in “Knowledge and Belief,” included in Knowledge and Certainty (Englewood Cliffs, NJ: Prentice-Hall, 1963), 58 – 72  Compare G. E. Moore’s explanation, to be discussed in chapter 3, of what he meant by calling the fundamental propositions of ethics self-evident, in Principia Ethica (Cambridge: Cambridge University Press, 1903), Preface.  J. L. Austin, Sense and Sensibilia (Oxford: Clarendon Press, 1962), 115. Original italics.

36

Chapter Two: Three Varieties of Epistemology

Skepticism begins by noting that usually what we want to know is neither self-evident nor made evident by anything that is self-evident. And subjective epistemology usually tries to avoid the disconcerting implications of this fact by appealing to something else that it hopes is relevant to what we want to know. We may call it “evidence,” although, even if it is itself self-evident, it does not make what we want to know evident. This is how Austin used “evidence” in his example of the pig: “a few buckets of pig-food,” “the noises and the smell.” Religion and the law are the noteworthy home of such uses of the word, which are often exquisitely circumspect. They are also common, even if less circumspect, in the lab and the street. The notion they express is a degenerate offspring of the notion of the evident. But, though degenerate, it is usually understandable and also harmless. We may need to know, not merely believe, that God exists, yet we may be aware that we do not. So we look for “evidence” of his existence. In the courtroom, a verdict of guilt or innocence may be mandatory, though neither guilt nor innocence is likely to be self-evident or made evident by anything that is self-evident. So we look for something else we hope is relevant to guilt or innocence, and call it “evidence,” whether “beyond reasonable doubt” or not, and whether just “circumstantial” or not. If we think that such degenerate evidence is “strong” enough, we may even say that we know that for which we take it to be evidence, thus making also a degenerate use of “know.” Saying this would make explicit the reason we appealed to it in the first place, i. e., our desire for truth, and it may seem to anoint the appeal. How strong the evidence must be, however, is never made clear because, given the sort of reasons that lead us to appeal to it, this cannot be made clear. As cognitive beings, we seek knowledge because it is truth we want, not such “evidence,” even if we were to dress it up with phrases like “epistemic probability” and “epistemic justification.” The idea that evidence comes in degrees and that possession of it yields an approximation to knowledge, something worth having even though knowledge is absent or impossible, may suggest to some that there can be such a thing as an approximation also of truth. But while truth may be incomplete, irrelevant, or misleading, there cannot be twothirds or 86 % truth. Nonetheless, though as cognitive beings it is truth and therefore knowledge that we seek, we are not purely cognitive beings. The degenerate uses of “evidence” and “know” in religion and the courts of law are defensible. They are often needed even in the lab and the street. To go about our business we must think of certain judgments as final, settled, even if we soon revisit them. There are practical reasons in religion, the court of law, the lab, or the street for resorting to a degenerate notion of evidence. But no such reasons exist in epistemology, which is neither a religion or a courtroom nor a lab or the street. Our con-

2 Subjective Epistemology

37

cerns in it are purely cognitive. This is why the degenerate notion of evidence is not harmless in epistemology. It gives rise to the illusion that knowledge is relatively easy to achieve, or at best that what knowledge requires is merely the limit, even if only ideal, of a range of degrees of “evidence,” of “epistemic probability,” or of “epistemic justification,” and that what falls short of that limit would nonetheless suffice. But for what it might “suffice” is unclear, since practical considerations are now irrelevant. Unsurprisingly, it has not sufficed for answering the skeptic. In everyday life and thought, the degenerate notions of evidence and evidence provide a way of clearing our epistemic conscience. They are analogous in this respect to the degenerate notion in ethics of “subjective duty,” which provides a way of clearing our moral conscience. The weak sense of “know” is analogous to the weak sense of “ought” introduced by the notion of subjective duty. If ignorant, as we often are, of our objective duty, of what we ought to do, we may settle for doing what we think, perhaps feel, we ought to do, our subjective duty. We may even insist that one always ought to do what one thinks or feels one ought to do. However, just as our concern as cognitive beings is with truth, our concern as moral beings is with doing what we really ought to do. The weak senses of “know” and “ought” are natural, in view of the scarcity of cases in which we can use “know” and “ought” in their proper, strong senses. There is no need for legislation against them. But we are deeply aware of the difference when facing matters of major importance, and then we stay faithful to the strong senses. We do not usually say that we know we will be alive tomorrow and thus that we need not pay our life insurance premium today, regardless of how healthy and safe we think or feel we are today. Serious people buy fire insurance even though they have never had a fire. And we do not say that children ought to sacrifice their lives if they think or feel they ought to do it. We can now understand why the strong sense of “know,” which requires that what we say we know is self-evident or seen to follow from what is self-evident, has been central in subjective epistemology. The attraction of religion is that it promises certainty about matters of ultimate concern, not mere probability. A religious person would not be satisfied if told that God probably exists. A missionary does not win converts by assuring them that they would be justified in believing, or that it is probable, that there is God. The attraction of subjective epistemology is that it seeks certainty about matters of ultimate concern where nothing less suffices. Its main topic was the existence of material things – of the earth and the sun, of your body and mine. Among its topics were also the existence of other minds and the validity of induction. It would be jejune in everyday life to say that we are only justified in believing, or that it is only probable, that we have bodies, that others have thoughts and feelings, or that the

38

Chapter Two: Three Varieties of Epistemology

past tells us anything about the future. Only a philosopher might be satisfied if told that other people only probably exist. And it would be outrageous in epistemology to suppose that by saying that the external world probably exists we are genuinely answering the skeptic. The focus on epistemic justification or epistemic probability, rather than knowledge, came about precisely when epistemologists concluded that we cannot have genuine knowledge of such matters, yet remained unwilling to accept skepticism. The concept of knowledge played a central role in Cartesian epistemology. But Descartes’s principal question was not “What is knowledge?” This question had been asked before and usually answered, briefly and informally but sufficiently, in the same way – knowledge is apprehension, grasping, getting hold of the truth and then steadfastly holding on to it. The Cartesian question that inaugurated subjective epistemology was rather whether we have knowledge of anything, especially of an external material world. Cartesian epistemology began by taking skepticism seriously, hoping to refute it. And for this reason it was essential that the Cartesian epistemologist ask whether I have knowledge, that is, to employ an indexical, the first-person pronoun, rather than a name like “Descartes” or a definite description like “the author of the Meditations.” For had the Cartesian epistemologist done the latter, the skeptic would have complained of question-begging. One who questions the existence of the external material world questions also the existence of human beings, including Descartes, since they, or at least their bodies, are parts of that world. In the First Meditation, Descartes could not have used “I” to refer to the Frenchman named Descartes, since that Frenchman was part of the world the existence of which Descartes was to prove later, in the Second Meditation. When he employed the argument “I think, therefore I am,” Descartes could not have been referring even to the “thinking thing” that in the Second Meditation he concluded he was. He could not have said he was referring to his thinking thing, for how would he have answered the question, which thinking thing it was? He could not have said, “my thinking thing,” for “my” is the possessive adjective corresponding to “I.” Louis XIII also was, or had, a thinking thing, but Descartes did not suppose he was proving the existence of that thinking thing and thus of Louis XIII. Nor could he infer from “I think” just “There is a thinking,” as some have suggested. Was it Descartes’s or Louis XIII’s thinking? If it was no one’s, there might be thousands of such orphaned thinkings. The existence of which one was Descartes inferring? Epistemological ventures seldom benefit from ontological adventures. To confront the skeptic without begging the question Descartes needed to begin his inquiry by renouncing claims to any subject matter. He could refer to nothing even when using “I.” Subjective epistemology must lay claim to no subject matter when attempting to refute skepticism.

2 Subjective Epistemology

39

Descartes probably did not see that he faced these difficulties because all along he thought he was “directly aware” of a thinking thing and its ideas. But even if we ignore the question about which thinking thing and ideas those were, “his” or Louis XIII’s, yet another question, also fundamental but ignored by Descartes, can be asked. If a necessary condition of awareness is that its object exists, then the skeptic would ask whether Descartes was really aware of a thinking thing and its ideas, just as the skeptic asks whether we really perceive bodies when we think we perceive them. And if the existence of its object is not a necessary condition of the awareness, then the skeptic of course would question the cogency of the inference from the occurrence of the awareness to the existence of the thinking thing and its ideas. Descartes thought he might be deceived by God or an evil demon regarding 3+2=5, but did not see that if this is so then he might be deceived also about what he thought he was aware of. This, we may note, vitiates also Descartes’ several inferences from the existence of his idea of God to God’s existence. Couldn’t God, or an evil demon, deceive him into thinking that he had that idea? Perhaps he did not really have it. The failure to see that in this way the skeptic could question any appeal to awareness vitiates also the familiar appeals in post-Cartesian subjective epistemology to “intuition,” “immediate experience,” or “direct acquaintance.” Plato pointed out the poverty of such appeals in the Theatetus, and so did Hegel,⁴⁵ but they remain common in philosophy. Thus, while naturalistic epistemology has a subject matter too limited to be philosophical, subjective epistemology appears to have no subject matter at all. Its raison d’être is to meet the challenge of skepticism. Otherwise, there would be no rationale for distinguishing it from naturalistic epistemology, albeit it would be a naturalistic epistemology concerned, inexplicably, with just one natural object, just one human being – oneself. To remain subjective, subjective epistemology must refer to the “self” only by means of indexical expressions such as “I.” To both have a subject matter and not beg the question against the skeptic, it must be satisfied with a subject matter that is an entity that can be referred to only with indexicals. But would anything be an entity if it could be referred to only with an indexical? To suppose that there could be thinkers who are only Is borders on incoherence, just as to suppose that there could be times and places that are only nows and heres borders on incoherence. Even to say this has required use of the grammatical monstrosities “Is,” “nows” and “heres.”

 See, for example, Hegel’s Logic, trans. William Wallace (Oxford: Clarendon Press, 1975), Part Five.

40

Chapter Two: Three Varieties of Epistemology

Before his optimistic inferences to the existence of God, Descartes seemed to have only himself, to be in a state of absolute solitude. But in fact he did not have even himself, whether as the writer of the Meditations or just as a certain thinking thing. He had nothing. Subjective epistemology is dependent on the use of “I” as a dangling pronoun, a pronoun without an antecedent noun. It is like a geography of here or a history of now that in principle is unable to say where is here or when is now. This is why subjective epistemology is often described as epistemology from the first-person perspective. But there are no such entities as first persons, second persons, or third persons, there are only first-person, second-person, or third-person pronouns. The vague and much abused term “perspective” can be misleading, but it is helpful because there are no obvious alternatives to the adjective “perspectival.” There is an alternative to the noun “perspective,” however, that is less affected, namely, “view,” as long as we understand this term not in the optical but its ordinary broad sense. So understood, “view” is a synonym of “cognition.” The “first person,” we may then say, is only a perspective, a view, a cognition. It is in this sense that subjective epistemology is only perspectival. There is no entity that is just the first person, and so the subject matter of subjective epistemology is not that privileged entity – the first person is not a person. It is only a view, a cognition. I shall have much more to say on this topic in Chapter Ten. Nonetheless, though subjective epistemology lacks a subject matter and is only perspectival, the rationale for it is impeccable. Lack of subject matter does not imply unimportance. The idea of a geography of here or a history of now that is unable to say where is here or when is now does seem absurd. But the idea of traveling from here and now without being able to say from where or when does not. A journey must begin somewhere and at some time, even if we cannot say where and when. Subjective epistemology may be an epistemology of pronouns without nouns, but to get nouns we must, so to speak, begin with pronouns. The first-person pronoun is indispensable, not because of what it refers to but because of the role it serves in the initiation of cognition. It is essential to all talk and thought, and thus to all inquiry. In normal contexts, to heedfully assert the sentence p one must be willing to assert “It is true that p.” But to heedfully assert “It is true that p,” one must be willing to assert “I know that p” (rather than the very different “I think that p” or even “It is probable that p”). Indeed, one must be willing to assert “I know that p” in order to heedfully assert “He (she, Jack, the expert) knows that p.” I can say that Jack knows when the train will leave even if I cannot say that I know, but I cannot say he knows that it will leave at 5 p.m. if I cannot say that I know this. In any inquiry, one must begin with the first-person view, with the use, however implicit, of “I,” even if only in judgments, implicit or not, such as “I’ll look for it in the bush,”

2 Subjective Epistemology

41

“I’ll ask Jill,” or “I’ll check the dictionary.” This is a proposition neither of physics nor of metaphysics. It’s like “Every journey must begin somewhere,” not like “Every journey must begin in Woodbury.” Yet the proposition enjoys the abstraction characteristic of philosophy and bequeaths it to subjective epistemology. “I think” must be able to accompany all of our representations, Kant held, even though, as Sartre later argued, it seldom actually does. Russell wrote, “When you are considering any sort of theory of knowledge, you are more or less tied to a certain unavoidable subjectivity, because you are not concerned simply with the question what is true of the world, but ‘What can I know of the world?’…You cannot go outside yourself and consider abstractly whether the things that appear to you to be true are true.”⁴⁶ Russell may have been wrong in thinking that there is an “inside” to be contrasted with an “outside,” but his grasp of the rationale for subjective epistemology was firm. Unless one opens one’s eyes and looks, one does not see. Unless there is a view, nothing is seen. This is why the allure of the subjective turn that Cartesian epistemology initiated is ever-present. It would be sad if subjective epistemology were all there were to epistemology, but outrageous to deny its essential insight. As a theory it is futile and usually misguided, yet it is as indispensable and unavoidable as one’s awareness that to get anywhere one must start somewhere and that to see anything one must look. The mistake is to suppose that subjective epistemology is about me, even if there is such an entity as me, whether a human being or a mere thinking thing. It is the mistake of supposing that subjective epistemology has a subject matter and thus that it is a cognitive discipline, a theory of something, presumably knowledge or cognition, when in fact it only draws attention to what is the necessary entry into any subject matter and serves as the prelude to any discipline. Subjective epistemology must use “I,” or a synonym of it, yet it can refer with it to nothing, not because there is nothing to refer to but because of the very nature of its project. That project was not a mistake. Antirealism is a metaphysical, not epistemological, theory, but it shares with subjective epistemology the peculiar feature I have just described. Its thesis is that the world, at least insofar as it is perceived, understood, or described, depends on our powers of perception, conceptualization, and speech. This is not a zoological proposition, however. It is not about humans, even though we are humans. We can now see better how it should be understood. Indeed, the proposition is not about humans, but neither is it about nonhumans. It is not about entities at all. Rather, it is about the necessary conditions of all thought and talk about entities. As such, it is intimately related to subjective epistemology, as intimately as Kant’s phi-

 Bertrand Russell, The Philosophy of Logical Atomism (Chicago & La Salle: Open Court, 1985), 37.

42

Chapter Two: Three Varieties of Epistemology

losophy was related to Hume’s, and can cast further light on it. That Hume’s skepticism led to Kant’s transcendental idealism was not just an event in Kant’s personal life. It manifested a turning-point in the history of philosophy, just as more than a century earlier Descartes’ method of doubt had manifested its turn to epistemology. Hegel’s absolute idealism manifested a third turning-point. The subjective epistemologist makes essential use of the first-person singular pronoun “I,” but Hegel saw that it is the first-person plural pronoun “we” that is essential to full-fledged cognition. The self-centered focus on the conditions that my cognition, my heedful thought and talk, must satisfy was broadened as well as deepened by Hegel’s focus on the conditions that our cognition, our heedful thought and talk must satisfy. To heedfully assert p, one must indeed be willing to assert “I know that p.” But this is only the necessary prelude to full-fledged cognition, which would be expressed by “We know that p.”

3 Epistemology-as-Logic Subjective epistemology does not have a subject matter. Naturalistic epistemology does, but its subject matter is human and thus lacks the supreme abstraction and generality characteristic of philosophy. The third variety of epistemology, epistemology-as-logic, has a subject matter that exceeds the bounds of the special sciences. Like formal logic, it is unambiguously “dehumanized” and belongs in philosophy. All three varieties of epistemology, however, are defensible within the limits of their very different yet not incompatible concerns. Their differences call for mindful distinctions, not mindless quarrel. In the Introduction I noted that in ethics some have eschewed the need for empirical investigation by taking “the deontic way,” focusing on what we ought to or at least are morally permitted to do. In the present chapter we saw that some have taken the deontic way in epistemology by focusing on what we ought to or at least would be “justified” to believe. They have thus avoided competing with the empirical sciences. Whatever its merits in in these cases, the deontic way is natural in epistemology insofar as its concerns have to do with the validity of certain inferences and thus are essentially logical. Logic, the “art of reasoning,” is commonly said to tell us what may and what may not be inferred from a given proposition or set of propositions, clearly a deontic matter. The inferences of interest in epistemology are seldom deductive, and for this reason the need for grounding judgments about their validity is especially evident. Much of traditional epistemology has consisted in the search for such grounding. Contrary to what textbooks sometimes say, formal logic is concerned not with inferences as actions, presumably human and thus properly studied by psy-

3 Epistemology-as-Logic

43

chology, but with their formal validity, the relation of the truth-value of the premises to the truth-value of the conclusion, in particular, the formal consistency of the conjunction of the premises and the negation of the conclusion. Its general subject matter thus consists of alethic relations, in the broad and etymologically proper sense of relations between propositions with respect to their truth-value. If some propositions, or at least sentences, are neither true nor false, as a consequence of the truth-value of other propositions or sentences, this fact too would belong in its subject matter. Formal logic thus exemplifies the generality and abstraction definitory of philosophy to the highest and purest degree. This is why Aristotle assigned “the principles of the syllogism,” especially that of noncontradiction, to “the science of being qua being.” This is why Frege wrote, “Just as ‘beautiful’ points the way for aesthetics and ‘good’ for ethics, so do words like ‘true’ for logic…[I]t falls to logic to discern the laws of truth…The Bedeutung [reference, meaning] of the word ‘true’ is spelled out in the laws of truth.”⁴⁷ Elsewhere, Frege explained: “What is distinctive about my conception of logic is that I begin by giving pride of place to the content of the word ‘true’ …”⁴⁸ If metaphysics is the science of being qua being, logic may be said to be the science of being qua truth, ethics of being qua goodness, and aesthetics of being qua beauty. Indeed, all four – Being, Truth, Goodness, and Beauty – belonged among what the medievals called transcendentalia. Epistemology-as-logic differs from formal logic by focusing on the validity – legitimacy, cogency, worth – of certain nonformal inferences, but its subject matter, like that of formal logic, consists of alethic relations, in particular, the relation of the truth-value of the premises of the nonformal inference to the truthvalue of its conclusion. It too enjoys the level of generality and abstraction characteristic of philosophy. Like formal logic, it is concerned not with inferences as human actions but with the alethic relations they exemplify. Unlike subjective epistemology, it does not lack subject matter, it is not just perspectival. And unlike naturalistic epistemology, which does have a subject matter, it is not just programmatic. Of course, epistemology-as-logic does apply to human matters, just as formal logic does. But it is not about them. There is nothing puzzling about this. Arithmetic also applies to humans, as well as to bats and stars, but it is neither about humans nor about bats or stars. It is about numbers. In attempting to answer the skeptic, subjective epistemology hoped to find cogent inferences, formal or nonformal, from premises it deemed known to be

 Gottlob Frege, “Thought,” in M. Beaney, The Frege Reader (Oxford: Blackwell), 325 – 326.  Beaney, 36.

44

Chapter Two: Three Varieties of Epistemology

true, even if as minimal as “I think.” Epistemology-as-logic, however, does not ask whether the premises of such an inference are true, nor does it agonize if the inference is not deductive. It is free from obsession with skepticism, just as naturalistic epistemology enjoys such freedom. In this it follows the lead of its older and more mature sibling, formal logic. In evaluating a deductive argument, formal logic is not concerned with the truth of its premises. And, we may note in passing, it also does not fret that, even if they are true, the formal validity of the argument might not suffice for the truth of its conclusion because God might deceive us about logic just as he might deceive us about arithmetic, the latter a possibility Descartes did worry about. Epistemology-as-logic may aim at providing eventually a general theory of the alethic relations exemplified in nonformal inferences. But it began by examining particular kinds of such inference, just as formal logic began with the examination of Aristotelian syllogisms and only later, mainly through Frege’s work, offered a general theory of the alethic relations they exemplify. Inferences involving probability, induction, and abduction are standard topics in subjective epistemology. They would also be topics in epistemology-as-logic, though in abstraction from possible use against skepticism, which was the reason for subjective epistemologists’ interest in them. Appeals to probability, of course, have failed to answer the skeptic, who would either deny that mere probability suffices or question the truth of one or more of the premises. Nonetheless, the calculus of probability remains an established discipline of some distinction. Appeals to induction and abduction, notoriously, also have failed to answer the skeptic. But they remain standard topics in the philosophy of science, which seldom strays into Cartesian doubts. But of special interest to epistemology-as-logic would be relations of nonformal entailment. That there are such relations is usually acknowledged independently of epistemological concerns. A standard example is the entailment of being colored by being red, and anyone who, like Kant, regards mathematical truths as necessary but “synthetic,” i. e., not reducible to logical truths, allows also for nonformal entailments in mathematics. Neither mathematicians nor philosophers of mathematics worry that God might be deceiving us about 7 + 5 = 12. An inventory, much less detailed discussion, of all nonformal alethic relations is neither possible nor needed here. I shall limit myself to an especially important one: the relation of presupposition. Bringing to light and focusing on that relation was the turning-point in the development of epistemology beyond its subjective stage. If we call subjective epistemology Cartesian, then epistemology-as-logic, insofar it focuses on the relation of presupposition, may be called Kantian. Hence the application of Kant’s term “transcendental” to recent arguments from presupposition. Kant’s project in the Critique of Pure Reason can

3 Epistemology-as-Logic

45

be described as discovering the presuppositions of the several levels of cognition, though instead of “presupposition” Kant used “condition” and would have described the project as discovering the conditions that the several levels of cognition must satisfy. But epistemology-as-logic need not adopt Kant’s essentially mentalistic approach to epistemology or any of his specific doctrines. The relation of presupposition became a major, though also controversial, topic in 20th century philosophy because of Strawson’s criticism of Russell’s theory of descriptions.⁴⁹ Strawson gave as examples the presupposition of “There is a king of France” by “The king of France is wise,” and the presupposition of “He is not dead” by “He cares about it.” (“Does he care about it? He neither cares nor doesn’t care; he is dead.”)⁵⁰ If France does not have a king, neither “The king of France is wise” nor “The king of France is not wise” is true. A proposition p presupposes a proposition q, according to Strawson, when both “If p then q” and “If not-p then q” are true, in other words, when q is a necessary condition of both the truth and the falsity of p, thus of p’s being a coherent proposition at all, even though p does not formally entail q and not-p does not formally entail q.⁵¹ In the example of the king of France, the presupposition manifests itself in our dismissing as confused, not inconsistent, both anyone who said that the king of France is wise but denied that France has a king and anyone who said that the king of France is not wise but denied that France has a king. Our judgment of anyone who said, “He cares about it but he is dead,” would be similar. Presupposition is neither formal entailment nor a probabilistic, inductive, or abductive relation. This is why appeals to it have seemed to provide answers to skepticism entirely different from the usual answers. The latter are almost certainly either formally invalid or contain premises the skeptic finds as questionable as the conclusion. The anti-skeptic’s predicament has been that to answer the skeptic one must assume more than the skeptic would allow, but if one assumes less then the answer does not follow from the assumptions. In appeals to presupposition, however, the consequence of denying the presupposed proposition is not the falsity but the incoherence of the proposition that presupposes it, whether a trifling incoherence as in the example of the king of France, or a deep one as in the examples I shall sketch shortly.

 P. F. Strawson, “On Referring” (Mind, 1950) and Introduction to Logical Theory (London: Methuen, 1952), 175 – 179. In the former, more influential work, Strawson did not use the term “presupposition,” and wrote instead of “some sense of ‘imply’” that is “not equivalent to ‘entails’ or ‘logically implies.’”  Introduction to Logical Theory, 18.  Introduction to Logical Theory, 175.

46

Chapter Two: Three Varieties of Epistemology

Some have said that presupposition is merely a feature of language, just “internal” or “pragmatic,” not “logical” or “semantic,” as if pervasive features of language are ever merely features of language. Aristotle defended the principle of noncontradiction not by trying to infer it from “more certain” propositions, but by showing that it is presupposed even by any reasoning intended to cast doubt on that principle. Russell repeatedly pointed out that the proposition “what follows from a true proposition is true” is primitive and presupposed by all deductive reasoning.⁵² That the natural sciences are rife with presuppositions has always been evident. Physics presupposes, it does not discover, the existence of a spatiotemporal world. Psychiatrists presuppose, they do not discover, that it is not evil spirits that cause mental illness. The examples from Strawson that I have mentioned may be of little intrinsic interest but this cannot be said of those in his major metaphysical work, Individuals, or his book on Kant, The Bounds of Sense. And certainly it cannot be said of the examples in Kant’s own works. Kant defended important but controversial philosophical propositions on the ground that they are presupposed by other propositions that are not controversial. A simple example is the presupposition that the objects of sense perception (“outer sense”) are in space. “This page is white” would not be true if this page were not in space, but neither would “This page is not white” be true if this page were not in space. Kant’s argument that morality presupposes freedom is another and famous example. Freedom seems to be presupposed by all genuine actions, moral, immoral, or nonmoral. It is what seems to distinguish actions from mere movements. A no less famous but more difficult example is Kant’s argument that objective order in time presupposes causal necessity. It is complex and not to be dealt with lightly, whether in agreement or disagreement, but we need not go into its details to get a glimpse of it. If we ask whether Jack met Jill before or after she moved to town, the answer would depend in part on reasoning about when and where he could have met her. When such questions and their answers really matter (as they often do in courts of law), it would be foolhardy to rely on memory impressions, what Kant called the subjective play of fancy. As these examples show, what is presupposed need not be a single proposition, just as a deductive proof ordinarily does not rest on a single premise. It might even be a system of propositions, and what presupposes it might itself be such a system. This is why the philosophically interesting examples of presuppo-

 Bertrand Russell, The Problems of Philosophy (Oxford: Oxford University Press, 1953), 71– 77. See Alfred North Whitehead and Bertrand Russell, Principia Mathematica to *56 (Cambridge: University Press, 1962), 1.1.

3 Epistemology-as-Logic

47

sition seldom have the simple structure of the examples about the present king of France and the man who is dead. Here are three other examples. The first is a presupposition especially relevant to traditional epistemology: the existence of a material world. It assumes many forms, and is neither simple nor obvious. G. E. Moore noted that in doubting the existence of the material world Descartes would have had to doubt the existence of other philosophers, past and present, including those he had read, heard, argued with, and whose works and views were the context of his doubt, whether through agreement or disagreement.⁵³ Philosophers are human beings, therefore inhabitants, parts, of the material world. The history of philosophy is not a history of angels. Descartes could not have taken his doubt seriously as a philosophical doubt if he had considered what would be the case with respect to his own doubt if there were not a material world. If genuinely philosophical, skepticism about the material world questions its own existence. The historical context of philosophical thinking, such as Descartes’ doubt, is essential to it, however original the thinking may be. It is even more obviously essential to it than, as contemporary essentialists have argued, the biological origin of an organism is essential, “metaphysically necessary,” to that organism. The “historicity” of a philosophical view is no more a contingent fact than the historicity of a political event. Both bear necessary relations to their past. Neither Cartesian epistemology nor Democratic or Republican politics in the 21st century would be comprehensible if stripped of such relations. The skeptic questions what makes it possible for skepticism to be the philosophical view it is: its roots in what some other philosophers have held. It would not exist if those philosophers had not existed, and thus if the material world did not exist. Descartes’ methodological doubt would not have occurred if the proposition he doubted had not been true. Indeed, the very language Descartes used to develop and explain his doubt would not have existed. And employment of language is essential to philosophical thought, even if rudimentary thoughts are possible without language. Philosophical thought, whether superior or mediocre, involves argumentation, good or bad, which has a fairly complex structure, distinct premises and conclusions, each with its own structure, and logical connections rooted directly or indirectly in that structure. The terms employed in the argumentation are chosen usually with deliberation and discretion from a fairly extensive and often highly technical lexicon. Any language employed in philosophy, say, Descartes’ French or

 G. E. Moore, “A Defense of Common Sense,” in Philosophical Papers (London: Allem & Unwin, 1959), 32– 60. I discuss the argument in Skepticism about the External World.

48

Chapter Two: Three Varieties of Epistemology

Latin, also involves phonemes and inscriptions, and has been shaped by a human community. All three – phonemes, inscriptions, and the human community – are parts of the material world. Could the argumentation be stated and explained in a private language, in the minimal, not necessarily Wittgenstein’s, sense of a language created by the philosopher alone without reliance on a public language, as in devising a secret code? Surely such a private language, even if possible, would be too primitive. The reader is invited to attempt constructing a fragment of one and then translating a paragraph from Descartes’ Meditations into it. Writing philosophy is not like recording one’s aches and pains. A private language for the latter, perhaps invented by a hypochondriac, might be possible, but to say that the sophisticated language employed by the philosophical skeptic could be such as language would be mere posturing. Might the argument take place just in the skeptic’s thought, without use of language, public or private? Even if some thought without language, e. g., recalling a sensation, were possible, to suppose that philosophical thought might be such thought would be like supposing that we can understand differential equations without using symbols. The skepticism considered in Descartes’ First Meditation was not a tipsy sailor’s declaration, “Maybe I know nothing.” It was a professional, serious and informed, philosophical view. This is why we still take it seriously. Of course, that philosophical skepticism about the material world questions its own existence does not entail that it is false. It does not make it self-contradictory. But it does make it deeply incoherent. If the material world did not exist, then it itself would not exist. The second example of philosophically interesting presupposition can be found in Sartre’s strikingly original defense of the existence of other minds. One’s acceptance of the “Other” is not discursive, he pointed out. It is presupposed by many of one’s own psychological states, it is essential to them. Sartre dwelt at length on the experience of shame when looking through a keyhole but seeming to hear footsteps. It is “an immediate shudder that runs through me from head to foot without any discursive preparation.” It is “shame of oneself before the Other,” even if I know that no one is actually looking at me.⁵⁴ To say that what is presupposed here is only the possibility of being the object of another’s look would be to deny that the experience is genuine shame. There is almost always such possibility. At any rate, the interesting sort of skepticism about the existence of other minds, one that is not just a trivial consequence of skepticism about the existence of other human bodies, questions even that possibility. For it questions the very intelligibility of there being anyone “other than myself.”

 Being and Nothingness, 22. Italics in original.

3 Epistemology-as-Logic

49

It denies that we can even “conceive” of another mind, somewhat as Berkeley denied that we can conceive of unperceived bodies. Such a skeptic owes us an account of that “immediate shudder that runs through me from head to foot.” My third example of presupposition belongs to an even deeper level. We may call it conceptual presupposition. What is presupposed is not a proposition but a particular understanding of a certain concept. For example, all discussions, skeptical and anti-skeptical, of the existence of bodies presuppose a particular understanding of the concept of existence. Therefore, to ask whether bodies exist presupposes some answer to the question of what it is for a body to exist, and ultimately of what it is for something to exist. Is existence a property? If it is not, then what are skepticism and anti-skepticism about? Would it be intelligible to speak of the truth or falsity of any sentences asserting or denying existence, and thus of knowledge or ignorance of their truth? Would such sentences express genuine propositions? Standard epistemology, whether naturalistic or subjective, provides little guidance on these questions. It usually takes for granted Kant’s view that existence is not a real predicate, meaning that it is not a property, a “determination,” of a thing, res, but fails to consider the epistemological implications of the view.⁵⁵ Kant himself wrote that the existence of a thing has to do “only with the question whether [the] thing is given to us in such a way that the perception of it could in any case precede the concept.”⁵⁶ This was not Berkeley’s “to be is to be perceived.” It was closer to Mill’s “matter is a permanent possibility of sensations.” Kant held that to be actual (wirklich) a thing must “stand…in accordance with the laws of empirical progression.”⁵⁷ We cannot go here into the details of these laws, but like any other laws they would involve, according to Kant, applications of the pure concepts of the understanding. If so, the actuality or existence of a thing, like its causality, was for him not so much a matter of discovery of empirical fact as the imposition of a concept. “Exists” expresses not a property we find in a thing, but rather a conception of the thing that we contribute.⁵⁸ If, following etymology, we use “object” for anything that is or can be “placed before or presented to the eyes or other senses” (Oxford English Dictionary) or to consciousness generally, the “mind,” we can use “entity,” again following etymology, for any object to which the concept of existence applies. Obviously, there are objects, “things,” that do not exist, e. g., the Easter Bunny or a child’s favorite but imaginary friend Jack. Objects and entities may not differ with re   

For a succinct statement of Kant’s view, see Critique of Pure Reason A 598/B 626. Critique of Pure Reason A 225/B 272. Critique of Pure Reason A 493/B 521. I develop this view in detail in Skepticism about the External World.

50

Chapter Two: Three Varieties of Epistemology

spect to aesthetic qualities, but it is entities that are of interest in cognition, as children soon learn. This does not mean that how we apply the concept of existence is mere caprice, any more than that how we apply the concept of causality is one. What is meant is closer to what Nelson Goodman meant by the “entrenchment” of the predicate “green,” which the predicate “grue” lacks, though for Kant the application of the concept of existence was grounded in the activity of our faculty of the understanding, not in our linguistic practices. Clearly, Kant’s view of existence required that skepticism and subjective epistemology, insofar as they concern the existence of bodies, be drastically rethought – or just altogether bypassed by him, as in fact they were. Is existence just what the existential quantifier expresses, the “satisfaction” of a propositional function, as Russell argued and most contemporary epistemologists take for granted?⁵⁹ To suggest that it is would be a nonstarter. Whether the propositional function “x is a horse” is satisfied depends on what we allow as values of the variable x. Is “x is a horse” satisfied by both Secretariat and Pegasus, or only by Secretariat? If we say the latter, our reason would be that Pegasus does not exist, but now in a sense of “exist” obviously other than yet presupposed by Russell’s. It is its ordinary sense, which we employ in saying that the Easter Bunny and the child’s friend Jack do not exist. In any genuine case of inference involving presupposition, there is a natural desire to think of it as a case of formal entailment, since this is the sort of entailment we find most familiar and understand best. And when we see that the conditional corresponding to the inference is not a tautology we are tempted to declare the inference invalid. Or, if we still find it compelling, we are tempted to change the conditional so that it becomes a tautology, and bless the change by calling it an “analysis” or “translation.” This is what Russell did in his theory of definite descriptions. He saw that if the present king of France is wise then, of course, France has a king. He also saw, however, that this is not a tautology. So, he proceeded to “translate” it into one. In the philosophically substantive cases, however, like those Kant and Sartre made familiar, no such analyses or translations are plausible. The proposition “If there is objective order in time, then there is causal necessity,” for example, obviously is not a tautology, and to try to change it into one by “analysis” would hardly be a task worth attempting. So, the remaining option is to just deny the proposition. When the presupposition is conceptual, however, there is no such third option. For example, the skeptic can deny, however implausibly, that skepticism,

 Bertrand Russell, Introduction to Mathematical Philosophy (New York: Simon and Schuster, 1971), 164.

3 Epistemology-as-Logic

51

like all philosophical thought, presupposes the existence of the material world, that objective order in time presupposes causal necessity, or that the phenomenon of shame presupposes the Other’s look. But the skeptic is not likely to question the presupposition of a particular understanding of the concept of existence when doubting or affirming the existence of an external world. An answer, whether affirmative or negative, to a question presupposes understanding the question. The skeptic cannot deny the central place of the concept of existence in any discussion of what does or does not exist, or of what we can or cannot know to exist. Berkeley’s “to be is to be perceived” was not a casual remark made when defending his immaterialism.

Chapter Three: The Property Good 1 Anthropocentrism and Conceptual Analysis in Ethics Although ethics usually has concerned itself explicitly with human matters, there have been noteworthy exceptions. Plato’s theory of the Form of the Good in The Republic is a famous example. In the 18th century, as we saw in the Introduction, Kant distinguished sharply between the metaphysics of morals and the empirical discipline of “practical anthropology.” In the 20th century, G.E. Moore offered in Principia Ethica a detailed exposition and defense of an ethics concerned with what he called the nonnatural property good. It is well-known but usually vigorously rejected. And fifteen years later Wittgenstein outlined in Tractatus Logico-Philosophicus a subtle conception of ethics as concerned with the world, not oneself or other humans. It has received little attention, unlike the rest of the book. Plato, Kant, Moore, and Wittgenstein provided both the rationale and the outline of a dehumanized ethics, though they differed much in other respects. In this chapter I shall consider Moore’s ethics, and Wittgenstein’s in Chapter Four. My aim will not be to offer exegesis but to provide detailed examples of dehumanized ethics at its best and explain why and how they avoided anthropocentrism. In Principia Ethica Moore proposed that good (I shall follow his use of the adjective instead of the noun “goodness”) is a simple, indefinable, nonnatural property, and that this property is the proper subject matter of ethics. The book was published in 1903, and became a signpost in the philosophy of the following 100 years. It may still be too early to judge how 20th century philosophy ended, but its beginning was remarkable. Russell’s Principles of Mathematics also appeared in 1903, the first volume of Husserl’s Logical Investigations in 1900 – 01, and four of William James’s major philosophical books in 1902– 09. Wittgenstein’s Tractatus Logico-Philosophicus was written between 1914 and 1918. There was not a significant difference, except perhaps in style and temperament, between Anglo-American and European philosophers. The analytic/continental schism came much later. In the Preface of Principia Moore wrote that his ethics was closest to Franz Brentano’s. Both Russell and Husserl began as mathematicians. Frege was a German philosopher-mathematician to whom, by their own admission, Russell and Wittgenstein were heavily indebted. Russell studied and discussed Meinong in detail. James was admired in both Britain and Europe, influenced Husserl and Wittgenstein, and was the subject of articles by Moore and Russell. In Principia Ethica Moore dehumanized ethics even more clearly than Plato had done. But the book also inaugurated what has come to be known as analytic ethics, the sort of ethics characteristic of 20th century analytic philosophy. Analytic ethics

1 Anthropocentrism and Conceptual Analysis in Ethics

53

began largely in opposition to Moore’s thesis that the property good is indefinable. But Moore had made clear that he had no interest in what he called verbal, and the tradition calls nominal, definitions. They are the business of lexicography, he wrote. Yet it was just such definitions that analytic philosophers sought, sometimes calling them “analyses.” The most familiar example, noted in chapter 2, comes from analytic epistemology, not ethics: the definitions of “S knows that p” that preoccupied epistemologists from the late 1960s to the early 1980s. The definitions sought and offered by analytic philosophers were not even lexicographical, definitions that record lexical facts and are tested by empirical investigation of speech and writing. Rather, they recorded their authors’ impressions of lexical facts, and were tested by the authors’ “intuitions” about what we would or would not say, usually in some hypothetical situation. The question “How do I know what we would or would not say in that situation?” was usually ignored. For it could be answered properly only by appealing to what others have said in similar situations, and thus making an appeal, however amateurish, to lexical fact. Even the Oxford English Dictionary is valuable mainly for the examples of usage it lists, not the definitions it tries to distill from them. Searches for such definitions became alien to contemporary philosophy of language because of the three trailblazing developments in it more than half a century ago that were mentioned in the Introduction. The first was Quine’s article “Two Dogmas of Empiricism,” published in 1951, which attacked appeals in philosophy to meanings. It is widely accepted today, though often only pro forma. Phrases such as “conceptual question,” “conceptual content,” and “conceptual connection” still abound in the literature, with the noun “concept” explicitly used for the meaning or use of a word. The second development, also often accepted but just pro forma, was Wittgenstein’s relentless argument in Philosophical Investigations, posthumously published in 1953, that words are not used in accordance with necessary and sufficient conditions. He gave “game” as an example, but the argument applies also to “good,” “right,” “reason,” “know,” “exist,” and other denizens of the philosopher’s lexicon that, like “game,” are everyday words, not technical terms introduced as abbreviations of multi-clause descriptions. The third development was Chomsky’s linguistics, first proposed in that same decade. It marked a striking advance by insisting on the biological basis of linguistic competence and the use in the study of language of the standard methods of scientific research. The kind of definition Moore did allow in philosophy was an account of the constitution of what is defined, a listing of its parts. It was closer to what the philosophical tradition calls real definition, though it mentioned not genus and differentia but parts. Such a definition can be called an analysis, in a sense reasonably similar to that employed in chemistry. In later years analyses were offered mainly of

54

Chapter Three: The Property Good

facts and propositions, which were taken to be nonlinguistic entities categorially different from those chemistry analyzes. Their analysis was intended to reveal logical form, and for this reason was called logical analysis. It was in such analyses that analytic philosophy took root, beginning in 1905 with Russell’s theory of definite descriptions and perhaps culminating in Moore’s claim two decades later, in “A Defense of Common Sense,” that he knew the proposition “This is a hand” to be true but did not know how to analyze it. In Principia Ethica, however, Moore’s example was the definition of a horse and consisted of an anatomical inventory. Moore in effect agreed with Kant that “in matters of morality it is always real definitions that must be sought.”⁶⁰ Despite its inattention to what he meant by “definition,” analytic ethics did begin and develop in relation to Moore’s ethics. But discussions of Principia seldom ventured beyond Chapter I, which alone is included in most anthologies. Usually ignored was the crucial Preface, where Moore explained what he meant by “intuition” and “self-evidence,” and thus what anyone calling him an intuitionist and foundationalist ought to mean. Also usually ignored were the beginning of Chapter 2, where he explained what he meant by “natural” and “nonnatural,” thus what anyone calling his ethics nonnaturalistic ought to mean, and Chapter V, where he explained his theory of right on the basis of the theory offered in Chapter I. By “intuitions,” Moore wrote, he meant self-evident propositions, and “nothing whatever as to the manner or origin of our cognition of them.”⁶¹ And a self-evident proposition, he explained, is one that is evident but not by virtue of inference from other propositions.⁶² Moore did not say what he meant by “evident,” perhaps thinking it unnecessary. I suggested in the previous chapter that a proposition is evident if it is or can readily be seen to be true, either literally or metaphorically. Therefore, it may also be said to be known, in the serious and traditional sense of “know,” which sharply distinguished between knowledge and belief or opinion. I also noted that the noun “evidence,” as used in a court of law or in the lab, has a wider meaning, but the same root. Moore described as “self-evident,” if evident at all, those propositions in ethics that state “what kind of things ought to exist for their own sakes,” i.e., are intrinsically good. As to the meaning he attached to calling something “natural,” Moore wrote he meant that the thing is in time. Therefore, what is nonnatural is what is not in time. This is not an idiosyncratic sense of “natural.” To be sure, we usually think  Immanuel Kant, Logic, trans. Robert S. Hartman and Wolfgang Schwartz (New York: Dover, 1974), 144.  G. E. Moore, Principia Ethica, Revised Edition (Cambridge: University Press, 1993), 36.  Principia Ethica, 193.

1 Anthropocentrism and Conceptual Analysis in Ethics

55

of the things in nature as being in both time and space. But there might be things that unquestionably are in nature but not in space because they are not physical. The obvious example would be irreducibly mental states, the existence of which most psychologists at the time accepted. They would not be in space but they would be in time. (This is why in later philosophy the paradigm of nonnatural entities were abstract entities such as numbers.) The mere fact that a thing falls outside the subject matter of physics does not make it nonnatural. If we define a natural thing as one belonging in the province of the natural sciences, we would need a noncircular account of what is meant by “natural sciences,” as Moore was doubtless aware and for this reason did not offer such a definition. But the fact is that the distinction between the natural and the nonnatural did not play a central role in his book, though the phrase “naturalistic fallacy” did. As Moore made clear in the also usually ignored Chapter IV, which was devoted to what he called metaphysical ethics, even ethical theories concerned with the “supersensible” committed the fallacy. The fallacy was merely a case of confusing two things: the property good and some other property. In Chapter I of Principia Ethica Moore argued that the property good is nonnatural and simple, therefore indefinable if defining something is listing its parts, that almost all earlier ethical theories had committed the naturalistic fallacy of confusing it with some other property, and that they could be refuted with what later was called the open-question argument, which in effect encouraged the reader to pay close attention to the property the theory confuses with the property good in order to see that they are two properties, not one. It may be worth mentioning that Moore’s contemporaries in the Society of Apostles and the Bloomsbury Circle, who included Russell, Keynes, and Virginia Woolf, found more important, not these metaphilosophical generalities, but the substantive view, defended in Chapter VI, that personal affection (love, friendship) and aesthetic appreciation (contemplation of beauty, in art and in natural objects, human and nonhuman) are the greatest goods. In contrast with Kant’s view that a good will is the only thing that is unconditionally good and Mill’s that pleasure alone is good, Moore held that love and the contemplation of beauty are the Ideal. But he rejected Sidgwick’s contention that nothing “appears to possess this quality of goodness out of relation to human existence, or at least to some consciousness or feeling.”⁶³ This may be what prompted Keynes to rate Moore higher than Plato.

 Principia Ethica, 133.

56

Chapter Three: The Property Good

Chapter VI has also been ignored by analytic ethics, which has focused instead on the preliminary discussions in Chapter I, especially the “objectivity of value” it took Moore to be defending there. But, as Brian Hutchinson points out in a recent book, “Moore never even entertained doubts about the objectivity of value.” Hutchinson acknowledges that for us this may be “a mystery difficult to fathom,” but suggests that the mystery “is to be savored rather than solved.”⁶⁴ While the central tenet of Moore’s theory of good was that good is a simple, indefinable, and nonnatural property, the central tenet of his theory of right was that the right action, i. e., duty, in a particular situation, is the action that “will cause more good to exist in the Universe than any possible alternative,”⁶⁵ that “is the best thing to do,” what “together with its consequences presents a greater sum of intrinsic value than any possible alternatives,” either because it “itself has greater intrinsic value than any alternative” or because “the balance of intrinsic value” of its consequences does, so that “more good or less evil will exist in the world” if it is adopted.⁶⁶ “Cause” and “produce” are used in the broad sense of “contribute,” since the action might be the best thing to do because of its own goodness or because of its “organic,” not causal, relationships. Moore has been called a utilitarian, but unlike Bentham’s, Mill’s, or Sidgwick’s utilitarianism, Moore’s presupposed a theory of good that places no limits on what items might be intrinsic goods, thus allowing that some may be actions. Moore’s theory of right may be called cosmological. It tells us that we ought to do what would be best, all things in the Universe considered. It is not idiosyncratic. It accords, for example, with Aquinas’s first principle of natural law: “Good is to be done and promoted, and evil is to be avoided.”⁶⁷ It does imply, as Moore noted, that justice is not to be done if the heavens should fall – unless, he wryly added, “by the doing of justice the Universe gains more than it loses by the falling of the heavens.”⁶⁸ As we shall see, the ethical views of Russell and Wittgenstein, the other two founders of analytic philosophy, were also nonnaturalistic and cosmological, perhaps because of Moore’s influence. But, with the exception of H. A. Prichard, a philosopher of unsurpassed acuity, and W. D. Ross, whose terminology and distinctions are still found indispensable, later AngloAmerican ethics diverged in both respects. They are related. If ethics is natural-

 Brian Hutchinson, G.E. Moore’s Ethical Theory: Resistance and Reconciliation (New York: Cambridge University Press, 2001).  Principia Ethica, 198.  Principia Ethica, 76 – 77.  Thomas Aquinas, Summa Theologica, Part Two, Question 94, Article  Principia Ethica, 197.

1 Anthropocentrism and Conceptual Analysis in Ethics

57

istic, it is not likely to be cosmological. And if it is cosmological, it is not likely to be naturalistic. Naturalistic ethics is almost certainly ethics humanized: it is about humans, not cats or bats. So it is not cosmological. Not only does it ignore the good of the universe, it ignores that of extraterrestrials, angels, and gods, if there are any, and usually also that of nonhuman animals, plants, and rivers. Thus it lacks the supreme generality and abstraction distinctive of philosophy and alone justifying its existence alongside the other cognitive disciplines. A cosmological ethics can be expected, of course, to have application to humans, just as chemistry and mathematics can. But this makes none of them about humans. Of course, we are human, thus interested in ourselves and other humans. But this is not a reason for making ethics be about us, any more than it is a reason for making astronomy to be about us. Indeed, as I noted in the Introduction, there is a special, deep, and often misunderstood sense in which humans may be cosmically central – namely, that leading to Kant’s transcendental idealism and its recent versions, e. g., in Nelson Goodman’s and Hilary Putnam’s works. In that special sense, Kant, Putnam, and Goodman may be said to have humanized even astronomy. But they did not hold that astronomy is about humans. If humans are natural objects, a species of animal, we can hardly expect to have special philosophical knowledge of them, just as we can hardly expect to have special philosophical knowledge of stars or bats. Accounts of human well-being do not belong in philosophy, just as accounts of human anatomy and human evolution do not. They belong, I have argued, in the empirical sciences. The argument does not depend on a narrow use of the words “natural” and “science.” If mental states are not reducible to physical states, there could still be a natural science of them, in Moore’s sense of “natural” and the traditional sense of “science,” in which history and political geography are social sciences. In fact there was such a science in Moore’s time – namely, introspective psychology. The argument did depend, however, on the fact that qualification for research in empirical matters requires special training, for example, in chemical analysis or the use of telescopes. Nothing analogous with respect to humans occurs in philosophy seminar rooms. How to achieve happiness, in the ordinary sense, recognized by both Kant and Mill, of enjoyment or satisfaction of our needs and desires, has been a stock question in ethics, with both Epicurus and Plato offering advice, but arguably the invention of aspirin and contraceptives, tractors and pesticides, air conditioning and spreadsheets, answered it more effectively. In Buddhist ethics, sadly but realistically, suffering seems the primary concern, not pleasure, as in Western ethics. But Indian utilitarians who hope to learn from Americans how to reduce suffering presumably go to American colleges of agriculture

58

Chapter Three: The Property Good

and public health, not American philosophy departments. One may ask, indignantly, what about loftier goods, not Bentham’s but certainly Plato’s and Kant’s, such as justice, authenticity, salvation? Especially in India, a deeply religious country, they are often thought far more important. But these loftier goods call for nonzoological considerations. Neither psychology nor lexicography would illuminate them. Philosophers who avow allegiance to naturalistic ethics do write about some loftier goods, at least about justice. But, as we saw in the Introduction, they are more likely to adopt a conception of ethics, far removed from both naturalism and nonnaturalism, as a “conceptual,” not “factual,” discipline. This allows them to avoid both commitment to nonnatural facts and responsibility for competence regarding natural facts. But if concepts are in nature – presumably in human minds, brains, or languages – they too lie outside philosophers’ competence: there is psychology (introspective or not), neuroscience, as well as linguistics and scholarly lexicography. If they are not in nature, Moore’s venture into the nonnatural was at least straightforward. It was also not dated, though its critics often call it “obsolete.” Like the 17th century way of ideas, conceptual analysis perhaps went out of date when Kant pointed out in 1787 that our business is not merely to analyze concepts but to extend our knowledge.⁶⁹ Like post-Gettier analytic epistemology, post-Moorean analytic ethics was unfazed by misgivings such as Quine’s about meanings. It clung to conceptual, in fact linguistic, analysis. It went through several stages. The first began in Vienna, soon after the publication of Wittgenstein’s Tractatus and with some personal involvement by him. Ethical statements were rejected as nonsense, or at least as lacking cognitive sense. The subtlety of Wittgenstein’s verbally similar position, however, was missed altogether. The second was to offer a positive characterization: they are expressions of emotion. But the rich literature already in existence on the emotions in psychology (from James to Arnold) and phenomenology (from Meinong to Sartre) was ignored, even though it seems to show that the emotions are not, as the emotivists thought, self-contained subjective episodes, Humean “impressions of reflexion,” but rather intentional states, directed upon objects, with character dependent on that of their objects, and thus in principle cognitive. The third stage, perhaps motivated by the experience of World War II, which made both the outright rejection and the emotivist interpretation of ethical statements appear jejune, was to claim that they express a special “moral point of view,” something psychologically no less genuine than the emotions but supposedly less subjective, and that their function is to “guide,

 Critique of Pure Reason, B 18.

2 The Good and the World

59

not goad.” In effect, this third stage consisted in denying ethical statements a full-fledged, unqualifiedly cognitive status, yet conceding that their function is not merely that of imperatives or exclamations. In all three stages it was taken for granted that the job of ethics is to describe the content of ethical concepts, or the meanings or uses of ethical words, the “workings” of ethical discourse. We cannot give people what really interests them, namely, an ethics that says what they should do, Moore’s heirs held, but we can give them an ethics that says what they mean, a “metaethics.” This was the message even of the more recent fallback positions of “projectionist” antirealism and “supervenience” realism, where the focus remained metaethical, not substantive. Few worried that the very idea of telling people what they mean seems paradoxical.

2 The Good and the World By taking the conceptual turn analytic ethics did not provide a genuine alternative to “ethics humanized.” The alternative provided by Moore remained. Let us consider some of its details. I suggested earlier that the role of the indefinability and nonnaturalness of the property good was relatively minor, given Moore’s explanation of what he meant by them. Less familiar is that Moore proposed a criterion, a test, for determining whether a thing exemplifies that property, whether it is intrinsically good. The criterion was “the method of isolation.” It consisted in asking whether a world – a whole world – that contains the thing but otherwise was just like a world that lacks it would be better.⁷⁰ The two worlds might be wholly inanimate, and even considered “apart from any possible contemplation…by human beings.” The focal good is that of the world, not that of the human or sentient parts of it. It is independent even of possible human consciousness. An essential part of Moore’s ethics was the principle of organic wholes: “the value of a whole may be different from the sum of the values of its parts.” The method of isolation suggests that the world itself is an organic whole. In addition to (1) being good intrinsically, independently of anything else, by exemplifying the property good, a thing (object, action, state, property) may be said to be good because (2) it has a totality of consequences that exemplify that property, or because (3) it noncausally enhances the goodness of the organic unities or wholes of which it is an element. Its overall goodness thus depends on the actual or possible goodness of the world, the “Universe.”

 Principia Ethica, 135– 136, 143 – 147, 236 – 238, 245 – 247.

60

Chapter Three: The Property Good

Pleasure is a familiar example of something that is good in sense (1), surgery of something good in sense (2), and the presence of attractive surroundings might be an example of something that is good in sense (3). A thing could be said to be good overall only if considered in all three respects. In the case of actions, consequentialism ignores sense (1), deontological ethics ignores sense (2), and sense (3) is ignored by both. Acknowledging all three, as Moore did, gives both consequentialism and deontological ethics their due. In particular, consequentialism is given its due because the goodness or badness of an action overall (in W. D. Ross’s terminology, if not meaning, its being actually, not just prima facie, right) would depend, in part, on the goodness or badness of its consequences – all of them, for anything less would be morally unacceptable. All the consequences of an action and all the organic wholes to which it belongs are relevant to its being good or bad overall. The action will have consequences even after the human species has become extinct, a fact that thoughtful environmentalists would not find irrelevant. An explicit provision of Moore’s method of isolation was that the two hypothetical worlds might be supposed to be unpopulated, even considered “apart from any possible contemplation…by human beings,” as Moore did when applying the method to beauty.⁷¹ The action belongs to organic wholes possessing parts, both near and remote, past and future, with an indeterminate range, and thus requiring reference to all space and time. Moore defined duty in Principia Ethica as the action that will cause more good to exist in the Universe than any possible alternative, but he also described what “is good in itself or has intrinsic value” as what “ought to exist for its own sake.”⁷² In his later work Ethics he repeated that “it is always our duty to do what will have the best possible consequences,” but denied that this is “a mere tautology.”⁷³ In the preface to the second edition of Principia Moore wrote that he had used “good” in a sense that bears an “extremely important relation to the conceptions of ‘right’ and ‘wrong.’”⁷⁴ Clearly, what he offered was not a stipulative definition, but neither was it a lexicographical report. This is what we should expect. The connection between good and right cannot be trivially definitional, but it also cannot be just happenstantial. And even if good were a natural property, right would remain nonnatural if understood in terms of good in the manner Moore proposed. For it would still involve reference to all the consequences of an action and all the organic wholes of which it is a part – to all space and time,

 Principia Ethica, 133. The propriety of Moore’s use of beauty as an example here is astutely discussed by Robert Audi in “Intrinsic Value and Reasons for Action,” in Metaethics after Moore.  Principia Ethica, 34.  G. E. Moore, Ethics (Oxford: Oxford University Press, 1912) 73.  Principia Ethica, 4.

2 The Good and the World

61

the whole world. But these totalities of consequences and organic wholes, indeed the world itself, would not be natural objects in Moore’s sense. They would not be natural because they would not be in time, even if they consisted only of things that are in time. They may be not natural also in a larger sense. As we shall see in the next chapter, Wittgenstein held that, although sentences about such totalities show something, indeed the “higher” (das Höhere), they say nothing. The idea that the rightness of an action involves reference to the whole world is not purely philosophical. It is supported by scrupulous moral thought, which sets no time or place beyond which it cares not what happens. Some Americans do care about the floods in Bangladesh, and many people, wherever they may be, care about the climate on earth a century from now. Authentic environmentalists do not say that when humans become extinct, whales and prairie grass might as well. Many believe honesty would be owed to, and expected of, even extraterrestrials, angels, and gods, should they exist. I mentioned earlier that Russell’s and Wittgenstein’s views were also nonnaturalistic and cosmological, perhaps partly because of Moore’s influence. I am interested here not in their historical connections, but in the light they may cast on each other. In 1903, when Principia was published, Bertrand Russell wrote, “Man’s true freedom … [lies] in the determination to worship only the God created by our own love of the good.”⁷⁵ In 1914, after two years of intense discussions with Wittgenstein, he attributed to “the ethical work of Spinoza… the very highest significance,” as “an indication of some new way of feeling towards life and the world.”⁷⁶ This new way of feeling, Russell added, lay outside the scope of “the scientific method.” Hegel wrote that “to be a follower of Spinoza is the essential commencement of all philosophy.”⁷⁷ Spinoza defined God as “substance consisting of infinite attributes, each one of which expresses eternal and infinite essence,” and declared that “love towards God is the highest good which we can seek.” The method of isolation implies important and illuminating, but seldom noticed, similarities of Moore’s views also to Kant’s and Wittgenstein’s. For Kant, “the ultimate end of the pure use of our reason” was ethical, but he resolved to “[keep] as close as possible to the transcendental and [to set] aside entirely what might…be psychological, i. e., empirical,”⁷⁸ since “the metaphysic of ethics

 Bertrand Russell, “A Free Man’s Worship,” in Mysticism and Logic (London: Allen & Unwin, 1917), 50.  Bertrand Russell, “Scientific Method in Philosophy,” in Mysticism and Logic, 109.  Hegel’s Lectures on the History of Philosophy, (London & New York: Routledge and Kegan Paul ; Humanities Press, 1974), 257.  Critique of Pure Reason, A 797/B 825-A 801/B 829.

62

Chapter Three: The Property Good

is really the pure morality, which is not grounded on any anthropology.”⁷⁹ Wittgenstein’s position will be discussed in chapter 4. Suffice it here to point out that, since Moore’s theory of right action involves reference to the world as a whole (duty is “that action, which will cause more good to exist in the Universe than any possible alternative”), he might have agreed with Wittgenstein that in some sense ethics concerns the “limits of the world,” not its contents, that it is “transcendental.”⁸⁰ It may be worth noting that as a student Moore studied Kant assiduously, even attending a course on Kant in Germany, and that he and Wittgenstein enjoyed a close relationship. Moore and Russell were not Hegelians, but they grew in a philosophical culture dominated by Hegel’s philosophy. Wittgenstein had close ties to both. Spinoza’s Substance, Hegel’s Absolute, and Moore’s, Russell’s, and Wittgenstein’s worlds were very different, but all five were taken to be reality as a whole. That Russell and Wittgenstein may have shared a nonanthropocentric view of ethics such as Moore’s should not be surprising. The view was not entirely novel. Plato held that the philosophic life culminated in a glimpse of the Form of the Good, which he held to be indescribable. Aquinas placed Good in the company of Being, One, Truth, and Beauty, the so-called transcendentalia, which were said to range across the categories, i. e., the highest genera, thus to lack even the status of categories of things in the world, much less the status of things, and therefore to be indefinable per genus et differentiam. In philosophical theology God had been described as a being of infinite goodness that is the source and measure of all other goodness, earthly and unearthly. And Kant, as if using words from Wittgenstein’s Tractatus, wrote that, unlike what he called practical anthropology, moral thought is concerned, not with what happens, but with what ought to happen, even if it never happens.⁸¹

3 The Relevance of the Property Good It is often asked whether a dehumanized ethics such as Moore’s could be “relevant” to action. The question is as ancient as Aristotle’s complaint about the “inefficacy” of Plato’s Form of the Good. But it is ambiguous. It may be asking (1) whether Moore’s property good could motivate actions by itself, (2) whether one could be motivated by cognitive states of which it is an object, by itself or

 Critique of Pure Reason, A 841– 842/B 869 – 70.  Tractatus, 6.13, 6.421.  Critique of Pure Reason, A 802/B 830.

3 The Relevance of the Property Good

63

as a constituent of a state of affairs, or (3) whether there could be such cognitive states in the first place. The idea of being motivated to action by Moore’s property good indeed has doubtful coherence, but only because the idea of being motivated by any property seems incoherent, for purely logical, not ethical or psychological, reasons. A sentence of the form “FRx” would be ill-formed. But it is not incoherent to ask whether one can be motivated by cognitive states of which the property good, or a state of affairs that includes it as constituent, is the object. Could such cognitive states be “reason-giving”? Could they be “action-guiding”? The question is difficult to answer because those who specialize in such matters, namely, psychologists working on motivation, still know too little, perhaps because of the moral and social constraints on serious experimentation with human subjects. Hume did announce that reason is, and ought to be, the slave of the passions, but we may wonder how he knew this. Perhaps Hume could be excused for thinking that “passions” are discoverable through introspection. But his own views on causation should have kept him from thinking that their being motives, i. e., their motivating, could be discovered in such manner. Some have denied that there can be cognitive states of which the property good, or a state of affairs that includes it as constituent, is the object on the grounds that the property good could have no causal efficacy. They usually rely on causal or quasi-causal theories of knowledge, which are accepted either because they seem “scientific” or because of thought-experiments about what we would or would not say in imaginary situations. Some even deny the cognitive status of arithmetic if it is taken to be about abstract, nonspatiotemporal entities, since they assume that causation is a spatiotemporal relation. But this is to assume that the nature of causality is clearer than the nature of knowledge, or at least that we understand the former better than the latter. Many, of course, just deny that there is such a property as Moore’s property good. Elsewhere I have suggested that it should be considered a generic property, though one on a very high level of generality.⁸² It is the genus of which Moore’s personal affection and aesthetic appreciation, Aristotle’s eudaimonia, Mill’s pleasure, and other goods reasonably proposed by reasonable people would be species. This is why we cannot “see” it in the way we can see a shade of yellow. Neither can we see color, the genus of which a shade of yellow is a species, in that way, though color is a generic property on a much lower level of generality than goodness. Perhaps generic properties have no causal powers by themselves but their species do. A tire’s having shape helps no car roll but its

 Skepticism in Ethics (Bloomington & London: Indiana University Press, 1989).

64

Chapter Three: The Property Good

being round does. A traffic light’s being colored stops no driver but its being red sometimes does. If Moore’s ideal goods – personal affection and aesthetic appreciation – are species of goodness, perhaps they have causal powers. If they do, as seems obviously to be the case, but their genus does not, the reason may be that the latter is a generic property, like shape or color. Whether or not this suggestion is right, those who deny that we can have cognitive access to Moore’s property good on the grounds that it would have no causal efficacy need to pay more attention to such metaphysical details. They also owe us answers to questions like those explored by David Armstrong⁸³ and Evan Fales.⁸⁴ The first is what is causation? Is it, as Armstrong and Fales argue, a relation between universals, properties? If not, is it nonetheless grounded in such a relation? And are properties universals in the first place, or are they rather particulars, “tropes,” perhaps both as Moore in fact held?⁸⁵ How would the arguments against the cognitive accessibility of Moore’s property good read in the case of each possibility? According to Fales, there must be properties we can identify independently of their causal powers if a vicious infinite regress is to be avoidable. His examples are the properties characterizing the content of sense perception, though his ultimate concern is with properties in physics.⁸⁶ But if some properties can be identified without reference to their causal powers, so might Moore’s property good, whether or not it has causal powers. Needless to say, I shall not attempt to answer these questions here, but answers, detailed and carefully worked out, are needed. Otherwise it is difficult to take seriously the complaint that Moore’s property good can have no causal efficacy, that we could have no cognitive access to it, that therefore that it would be irrelevant to action. Appealing to “naturalism” or “the scientific point of view,” much less to “our intuitions,” is not enough. Without such answers the complaint might be like the 17th century natural philosophers’ complaint that Newton appealed to occult and immaterial gravitational forces, rather than to intelligible and robust bumping, or like H. A. Prichard’s complaint that Einstein’s theory of relativity was unintelligible because it relied on a non-Euclidean geometry. How a body could “motivate” another body at a distance was incomprehensible to those 17th century natural philosophers, and Prichard could form no mental image of a non-Euclidean space.

 David M. Armstrong, Universals and Scientific Realism (Cambridge: Cambridge University Press, 1978).  Evan Fales, Causality and Universals (London: Routledge, 1990).  G. E. Moore, “Identity,” Proceedings of the Aristotelian Society, n.s., v. I, 1900 – 1901.  Causality and Universals, 221– 224.

3 The Relevance of the Property Good

65

Therefore, I shall ignore the complaints about the relevance of Moore’s property good to action. But another question does arise. It is both legitimate and deep. By requiring reference to the whole world in judgments of duty, Moore could tell us nothing specific about how we ought to act in a particular situation. This is why he virtually admitted that, unlike his theory of good, his theory of right was profoundly skeptical. There might be an action we ought to do, but as a matter of empirical fact, not philosophical theory, we could not know what it is. In view of the mind-boggling range of its consequences and organic relationships, even probability statements about it could not be seriously made. Radical moral skepticism thus seems inevitable. But Moore’s moral skepticism does not lead to amoralism. His position is often described as “ideal utilitarianism” because it was not mere consequentialism. An action may be intrinsically good even if it ought not to be done, even if it did not make the world better. In Ross’s terminology, if not meaning, it may be a prima facie duty even if not an actual duty. This is why respect for the good may continue to inform one’s actions. Such respect would be akin to love, whether practical or pathological, not to calculation. It would have as its object the intrinsic goodness of the action, its being a “prima facie duty,” even if not an actual duty. Only a part of the world, not the whole world, can be loved. I am not suggesting an inference, which surely would be specious, from the intrinsic goodness of an action to its rightness. No claim is being made that the former makes the latter “probable,” or that it “justifies” or is a “reason” for the action. In Hutcheson’s useful terminology, if not meaning, it is at most an “exciting reason,” not a “justifying reason.” If in acting one is motivated and guided only by respect for the good, yet only the intrinsic goodness of an action is intellectually visible, then one is motivated and guided only by respect for the intrinsic goodness of the action. One has no knowledge of the totality of its consequences and organic relationships, indeed one has even no genuine conception of that totality. Thus, qua agent, the ideal utilitarian can only be a deontologist, not a consequentialist. We have to settle for right-minding, even if it does not coincide with right-acting. This is why Moore’s ideal utilitarianism was not inimical to moral common sense, which views with distaste the spirit of calculation, of cost-benefit analysis, that ordinary consequentialism cultivates. The ideal utilitarian has no “justifying reasons” but plenty of “exciting reasons” for doing good particular actions, namely, their goodness. Thus Moore’s dehumanized ethics may be seen as the marriage – of love, not convenience – of the two great ways of moral thinking: the utilitarian and the deontological.

Chapter Four: Saying and Showing The Good 1 The distinction explained In Tractatus Logico-Philosophicus Wittgenstein made the case for a dehumanized ethics more briefly but also more powerfully than Moore had done fifteen years earlier. He knew Moore well. And Principia Ethica had been published nine years before he met him. But he did not owe his views about ethics to Moore, even if they influenced him in some respects. They involved tying the notions of ethics to the notion of the world as a whole. As we saw in the previous chapter, this had been done also by G. E. Moore, key features of whose ethics were the notions of worlds “considered in isolation” and of the totality of the consequences of an action. But Wittgenstein’s ethics was largely based on his views in logic, in particular, that the world is the totality of facts (Tractatus 1.1), that “fact” is a formal concept (4.1272), that “world” therefore is also a formal concept, that the metaphysical subject or philosophical self does not exist (5.633), that therefore there is no philosophical distinction to be made between me or my life and the world and ethical statements can only be about the world, and that like all sentences employing a formal concept they say nothing even if they show much. Wittgenstein’s first stay at Cambridge was in 1912– 14. Moore’s ethics was a central topic of discussion in Britain at the time, not only among philosophers but intellectuals generally, especially in the Bloomsbury Circle. Both Moore and Wittgenstein were members of the Society of Apostles, together with Russell, Whitehead, and Keynes. Wittgenstein had read Principia Ethica. In a 1912 letter to Russell, he wrote that he hadn’t liked book, because Moore repeated “himself dozens of times.”⁸⁷ Wittgenstein also attended Moore’s lectures. They camped together in Norway, Wittgenstein dictating philosophical notes to Moore. Moore might well have agreed with Wittgenstein that by requiring reference to the world as a whole ethics is about “the limits of the world,” not its human inhabitants. It would be surprising if they did not notice the similarity of their views on this deepest level, despite their many differences, philosophical and personal. At a meeting of the Apostles in 1912 Wittgenstein heard Moore’s paper “Is Conversion Possible?” which Moore had first read to the Apostles in 1900 while working on Principia. That Moore read the paper again suggests he had not abandoned its ideas. Moral conversion, he said in it, “is not unlike religious conversion,” even though it “is not necessarily connected with any religious ideas.” It is “both a great good in itself and it secures all other goods which de Letters to Russell, Keynes and Moore, ed. G.H. von Wright (Oxford: Blackwell, 1974), 9.

1 The distinction explained

67

pend on one’s own mind alone….You see ‘life steadily and whole’ and can feel neither desire nor fear of what you see to be bad in it.”⁸⁸ At about the same time, Wittgenstein wrote in his Notebooks: “To believe in a God means to understand the question about the meaning of life…to see that the facts of the world are not the end of the matter…to see that life has a meaning,”⁸⁹ and later, in the Tractatus: “The sense [Sinn] of the world must lie outside the world” (6.41). The sense (or meaning) of the world is not something in the world because it is the sense of the whole world. He went on to remark that “It is not how things are in the world that is mystical, but that it exists” (6.44). Indeed, even if we refrain from calling the world mystical, we should acknowledge that it is mysterious. (It was one of the three spurious objects of knowledge Kant thought required treatment in the transcendental dialectic, the other two being the Self and God.) The reason the world is mysterious is logical, not mawkish or cabalistic. It is not that the world is too big or too unlike what we take it to be. Not its size or content, but its logical/ontological category, or rather its failing to fall in any category, is what makes it mysterious. We may say that the world is everything, but this would only acknowledge its peculiarity. For to speak of everything is to employ the predicate “is a thing” or “is a fact,” depending on whether we think the world is the totality of things or of facts. Both predicates, Wittgenstein held, express only formal concepts, corresponding to formal or internal properties, and thus the sentences in which they occur say nothing, though they show much. This is why Wittgenstein held that genuine propositions about it, and thus ethical propositions as he understood them, are impossible. Contrary to received opinion, neither this distinction between saying and showing nor the picture theory of meaning and thought on which it is based is idiosyncratic or obscure. The distinction has a straightforward, noncontroversial application even to ordinary pictures, say, paintings and photographs, indeed to representations generally. And the picture theory is merely a subtler version of the traditional theory of meaning and thought, which was unabashedly representational, “pictorial”: thought involves “ideas,” often explicitly understood as mental images or pictures, and the meaning of an expression is what it stands for. The distinction between what can be said and what can only be shown provided a welcome alternative to the stark choice between realism and antirealism, in logic, metaphysics, ethics, even the philosophy of religion. Wittgenstein

 Tom Reagan, Bloomsbury’s Prophet (Philadelphia: Temple University Press, 1986), 144.  Ludwig Wittgenstein, Notebooks, 1914 – 1916, trans. G. E. M. Anscombe (New York: Harper & Row, 1961), 74.

68

Chapter Four: Saying and Showing The Good

thought it concerned “the cardinal problem of philosophy.” Many of his readers find it obscurantist. Tough-minded philosophers ask, “What are those things that can only be shown? “How can there be such things?” But these questions miss the point of the distinction, namely, to provide an alternative to both realism and antirealism. What only shows itself is not a part of reality, like Secretariat or oxygen. But neither is it unreal, like Pegasus or phlogiston. This is why in the Introduction I called Wittgenstein’s position semirealism, distinguishing it from both standard realism and standard antirealism. The distinction becomes less puzzling when stated as the distinction, familiar to kindergarteners as well as film producers, between telling and showing – “show and tell.” Kindergarteners and film producers might or might not agree that some things can only be shown, but they would agree unanimously that showing is often much more effective than telling and sometimes alone possible. Of course, they would not have Wittgenstein in mind. But Wittgenstein’s picture theory of meaning does suggest a reasonably clear account of what it is to be something that can be told or said, namely, to be capable of being pictured, represented, whether in an ordinary picture such as a painting or photograph, or in a “logical picture” such as a sentence, and of what it is to be something that cannot be told or said but could be shown, namely, to be capable of being shown in a picture but incapable of being pictured. That there are “things” of the latter sort should be noncontroversial. In a painting, much is shown that is not and cannot be pictured by the painting or any part of it. For example, the painting may represent a tree next to a barn, each represented by a part of the painting, and the spatial relation between the parts of the painting that represent the tree and the barn would represent their relation of being next to each other. But nothing in the painting represents that relation’s being a relation, nothing “says” that their being next to each other is a relation (rather than, say, a shape or color). Yet the painting shows this, indeed must show it in order to represent what it does represent. What it shows cannot be denied as one might deny, for example, that the painting is a portrait of Churchill. The absence from the painting of what it only shows would not be like Churchill’s absence. Of course, paintings do not consist of words, and sentences are only “logical” pictures. But like all pictures, physical or mental, paintings are logical pictures, though not all logical pictures are paintings. The picture theory of meaning was offered in the Tractatus first as an account of thought in general: “A logical picture of facts is a thought [Gedanke]” (3). Only then did Wittgenstein also offer it as an account of sentential meaning: “In a proposition [Satz] a thought finds an expression that can be perceived by the senses” (3.1). “I call the sign with which we express a thought a propositional sign” (3.12).

1 The distinction explained

69

The theory was in accord with the received, traditional doctrine that thinking consists in operating with “ideas,” “mental representations,” perhaps even “mental images.” It implied, as that doctrine also did, the no less traditional doctrine that the truth of a judgment consists in correspondence to reality, to be established, if possible, by comparison. But Wittgenstein applied these traditional doctrines to language. He did so by proposing an unusual but not implausible conception of declarative sentences as logical pictures. He thus made the distinction between thinking and speaking seem insignificant. It was a “linguistic turn.” To think of something, traditional philosophy of mind held, is to represent it in the mind. Notoriously, this was most plausible and least unclear in the case of the representations called mental images. But to speak of something, Wittgenstein held, is also to represent it in a picture, though the picture would not be a painting or a photograph, or even a mental image. It would be a logical picture – a “propositional sign,” a sentence. Wittgenstein in effect strikingly broadened the traditional conception of a picture. A sentence (“propositional sign”) is a logical picture because it “depicts” what it says by sharing with it logical form, rather than, say, shape or color, as paintings, photographs, and perhaps mental images do. But Wittgenstein’s view of a sentence was also strikingly broad, it was not the grammarian’s: “The essence of a propositional sign is very clearly seen if we imagine one composed of spatial objects (such as tables, chairs, and books) instead of written signs” (3.1431). It would be unwieldy but possible to use tables, chairs, and books instead of words to refer to objects, and to use configurations of tables, chairs, and books instead of sentences to make statements. A sentence could also be composed of mental objects. In a letter to Russell, Wittgenstein wrote that a thought consists of “psychic constituents that have the same sort of relation to reality as words,” though he added: “What those constituents are I don’t know”⁹⁰ This is why Wittgenstein’s often used “thought” [Gedanke] and “proposition” [Satz] interchangeably. The linguistic turn that was implied by his assertion “The limits of my language mean the limits of my world,” therefore, must not be confused with the linguistic turn that became fashionable later, even though it was initiated by the Tractatus. If our cognitive access to reality consists in “representing” it, but the representations need not be more than logical, then whether the access is psychological or linguistic becomes irrelevant. This was the linguistic turn that Wittgenstein took. Descartes, Locke, and Kant accepted the first part of the antecedent of this conditional, but the second part did not even occur

 Letters to Russell, Keynes and Moore, 7.

70

Chapter Four: Saying and Showing The Good

to them. Had it done so, they might have accepted both, and then the history of modern philosophy would have been dramatically different. Wittgenstein’s distinction between “saying” and “showing” may be initially mystifying, but one of its applications is familiar even to beginning students of logic. That a statement is logically true, a tautology, i. e., that its negation is a contradiction, often “shows” itself and is immediately “seen” in its logical form, without reference to what the statement is about. But that a statement is true in some other way, presumably empirically, does not show itself, its truth cannot be just seen in its form – we must also know what it is about, what it “says,” by attending to what the descriptive expressions in it stand for. Although in accord with the traditional views of thought as representation and truth as correspondence, Wittgenstein’s picture theory of meaning and his distinction between saying and showing were a major, far-reaching revision of those views. He accepted them only on the level of atomic sentences, where indeed alone they are plausible. But this is an extraordinarily primitive level. Sentences on the highest levels, like those of logic and ethics, allow only showing, not saying, he held. Even ordinary molecular and general sentences fail to be pictures in Wittgenstein’s or indeed any view. A sentence of the form “If p then q” is not itself a picture, even if p and q are. Not surprisingly, in his later work Wittgenstein abandoned the picture theory of meaning and the correspondence theory of truth. They are not defective, but their applicability is limited. This is true, however, also of the traditional theory of thought and meaning: no physical or mental representation can be made of what a sentence of the form “If p then q” says, let alone the sentences of logic and ethics We can now understand the Vienna positivists’ passionate opposition to what can only be shown, especially the ethical. No picture of any kind – physical, mental, or merely logical – can be made of it. No painting can literally depict the goodness of a person or the rightness of an action. It follows that what can only be shown is not observable, since presumably anything observable can, at least in principle, be pictured, physically or mentally, and therefore also logically. The traditional empiricists denied that there are unobservable entities – we cannot have “ideas” of them, since ideas are copies of sensory impressions. The 20th century logical empiricists denied that there are things that only show themselves, and in particular that there are “ethical objects.” Their most familiar claims about ethics did appear to coincide with Wittgenstein’s. If propositions are pictures, then there can be no propositions of ethics. The ethical cannot be said. But Wittgenstein held that it can be shown. He avoided unbridled realism in ethics sufficiently to inspire logical positivism, yet he also avoided unbridled antirealism sufficiently to protect the ethical.

2 Logic and the World

71

Wittgenstein used “shown” (zeigt), of course, as a metaphor. He resorted occasionally to other terms, as in Tractatus 4.121: “Propositions cannot represent [kann nicht darstellen] logical form: it is mirrored [spiegelt sich] in them….They display it [Er weist sie auf].” And “say” (sagen), even if not used metaphorically, is sufficiently vague in everyday language, German as well as English, to allow for a variety of uses, as does also the closely related in meaning “tell.” A sentence says or tells us something, but so do a list of names, a thermometer, and a human face. Rudolf Carnap chided Wittgenstein for writing a whole book and then concluding that what he had said in it could not be said. But a simple way of avoiding the appearance of paradox would be to introduce, in place of “say,” the two verbs “say1” and “say2,” explaining that what is said1 is what according to Wittgenstein is said, and what is said2 is what according to him is only shown. For example, the sentence “This page is white” says1 that this page is white but says2 that the page is an individual object. Then Wittgenstein’s otherwise puzzling conclusion in the Tractatus would be that what he had said2 in the book could not have been said1. The difference would be explained sufficiently as the difference between what can be pictured and what cannot. There can be a picture of what the sentence “This page is white” says1, perhaps a photograph of it, but there can be no picture of what “This page is an object” says2.

2 Logic and the World In the Tractatus, as the title makes explicit, Wittgenstein was mainly concerned with logic. No explanation is needed, therefore, of the brevity of his remarks about ethics. Even in metaphysics, his focus was on its most abstract level, that of logical form. Not only would nothing be perceived or thought in a world without logical form, nothing would be a world without it. For nothing that violated or could not be captured by logic could be a world. Even a world consisting solely of immaterial or nonspatiotemporal objects must conform to logic. Wittgenstein’s logical and metaphysical views would have remained unchanged if he had supposed that his simple objects were angels. Unlike the logical antirealist, he did not deny that there is logical form in the world. Rather, he drew attention to the radical difference between logical form and the things that exhibit it by pointing out that, unlike them, it cannot be pictured, though it can be shown: “There are, indeed, things that cannot be put into words. They make themselves manifest. They are what is mystical” (6.522). If we take for granted that truth is correspondence to fact, as Wittgenstein did, then we must take for granted also that there is something in the world

72

Chapter Four: Saying and Showing The Good

to which the logical form of a true sentence corresponds, viz. the logical form of the fact it asserts. Otherwise, there would not be a fit sufficiently specific and definite for truth. Yet the logical form of the fact is quite unlike the things the sentence is about, viz. the constituents of the fact. It is invisible, as is the logical form of the sentence, even if those things as well the sentence and its grammatical form are visible. But it is also quite unlike the usual examples of what is invisible. The subject-predicate form would be present also in the putative facts asserted by the theological sentence “God is wise” and the mathematical sentence “3 is an even number,” and while God and numbers are invisible there is an obvious difference between the reason they are invisible and the reason logical form is invisible. The page and color that the sentence “This page is white” is about are visible, and so is its surface grammar, but its logical form is not. In the case of “3 is an even number,” both the logical form of the sentence and what the sentence is about are invisible. In the Introduction I called Wittgenstein’s position logical semirealism in order to distinguish it from both logical realism, which cheerfully allows for such statements as “This page is an individual object,” and logical antirealism, which no less cheerfully dismisses them. It is a sophisticated view. It should not be confused with mere denials of reality. Its aim is not to fight superstitions or fairy tales. Denying that some things can be said is not like denying that there is such a horse as Pegasus or such a substance as phlogiston. What is denied is that they are like Secretariat or oxygen. This is not to assert that they are like Pegasus or phlogiston. It is to assert that they are different from all four. Wittgenstein applied his semirealism chiefly to what he called “logical objects.” But it also had important implications for ethics – for what might be called “ethical objects.” Indeed, shortly after completing the Tractatus, he wrote that the point, the meaning, of the book was “an ethical one.”⁹¹ In this chapter I am mainly concerned with this ethical “point.” But to understand it, we must pay some attention to the logical/ontological views on which it rests, though much more will be said about them in chapters 6 and 7. The point of the Tractatus may have been ethical, but Wittgenstein declared that its “fundamental idea” was that the “logical constants” (the sentential connectives like “not,” “and,” “or,” “if…then”) and the quantifiers (“all,” “some”) stand for nothing in the world.⁹² More obviously relevant to ethics was that both “object” (“thing”) and “fact” were included in Wittgenstein’s list of formal

 Paul Engelmann, Letters from Ludwig Wittgenstein, 143 – 144. Cf. David G. Stern, Wittgenstein on Mind and Language (New York: Oxford University Press, 1995), 8, 70 – 74.  Tractatus, 4.0312.

2 Logic and the World

73

concepts (4.1272). It follows that the distinction between saying and showing applies also to the concept “world,” since the world is presumably either the totality of things or the totality of facts. It is “nonsensical” to speak of all facts or of all objects. We cannot speak of “totalities” determined by properties that can only be shown, i. e., formal properties. Therefore, we also cannot speak of the world. The totality of facts is the totality determined by the one-place predicate “is a fact,” and the totality of objects is the totality determined by the one-place predicate “is an object.” But being a fact and being an object are formal properties, which can only be shown. “Object,” “complex,” “fact,” “function,” “number” signify formal concepts, represented in logical notation by variables, for example, the pseudo-concept object by the variable “x” (4.1272). The properties they appear to stand for are formal, internal, such that it is unthinkable that what they are attributed to should not possess them (4.123). For this reason it would be just as nonsensical to assert that something has a formal property as to deny it (4.124). The statement “This page is white,” for example, does say something. What it says can be pictured literally, in a painting or a photograph. But the putative statement “This page is an (individual) object” does not, for it presupposes what it purports to say, its having sense, in particular the use of “this page” as its subject, depends on its being true.⁹³ Yet it is not gibberish. The motivation behind the two most common reactions to Wittgenstein’s distinction between saying and showing was plain, though often tacit, empiricism. The first holds that what only shows itself is not anything at all, since neither logical nor ethical objects can be observed. Prima facie, there is little to be said in favor of this view. Surely, what only shows itself, in logic or in ethics, is not like Pegasus or phlogiston. Its claims to a status in reality would remain even if they could not be fully met. The second, less straightforward, reaction is to hold that what only shows itself cannot be said for reasons of surface grammar. It resembles the common interpretation of Frege’s claim that the concept horse is not a concept: “is a horse” is a grammatical predicate, not subject. For example, Warren Goldfarb writes, “All we are doing [in speaking of logical form] is noting that names have to be put together in one way or another in order to make sentences.”⁹⁴ Of course, Wittgenstein was “noting” this, but it was not all that he was doing. He was also trying to explain why we put names together, why we need sentences rather than just names.

 In Principia Mathematica an individual is defined as anything that is neither a proposition nor a function (9.13).  Warren Goldfarb, “Metaphysics and Nonsense,” Journal of Philosophical Research XXII (1997), 66.

74

Chapter Four: Saying and Showing The Good

Even when something can be said with a sentence, why could it not be said with a list of names? This would be an awkward way of saying something, but why is it not a genuine way? Lists of names sometimes do seem to say, “tell,” something, e. g., in inventories. Wittgenstein’s answer was that we need sentences, not just names, because the world is the totality of facts, not of things. It was a metaphysical, not grammatical, answer. A world in which Jack is the father of Jim and a world in which Jim is the father of Jack may contain the same things and thus the same inventory, but they are different worlds. And why is it that only some ways of putting words together say anything, i. e., count as wellformed sentences? Wittgenstein’s answer was that the formal properties of what names stand for, i. e, objects, allow only some configurations of objects into states of affairs and thus only some logical pictures. Perhaps there are better answers to these questions, but answers are needed and few other than Wittgenstein’s have been given. The etymology of the word “reality” deserves attention by both realists and antirealists. The root of “reality” is res, the Latin for “thing,” and no argument is needed that “logical things” such as negation and “ethical things” such as goodness are not things like Secretariat or oxygen. To this extent logical and moral realism are unacceptable. But also no argument should be needed that they are not things like Pegasus or phlogiston. To this extent logical and moral antirealism are unacceptable.

3 The World and the Good Duty, Moore held, is the action that “will cause more good to exist in the Universe than any possible alternative.”⁹⁵ However, as we noted in the previous chapter, a sophisticated but sensible moral realism like Moore’s would allows that the action itself might be intrinsically good (prima facie right, in W. D. Ross’s terminology). This would be a sufficient reason for doing it even if we do not know what its consequences would be. Doing justice would be bad (“wrong”) if the heavens should fall, though presumably it would remain intrinsically good. To say that justice is to be done even if the heavens should fall would be moral posturing, not moral thinking. As Kant noted, Frederick the Great’s committing suicide, if captured by the enemy, in order to protect his country from extortion, might be good (“right”), even if suicide is bad (“wrong”) in itself. Nevertheless, such moral realism does remain beholden to the future. For it still enjoins us to take into account all the consequences of an action, since gen-

 Principia Ethica, 198.

3 The World and the Good

75

uinely moral thought sets no date and no place beyond which what happens would not “matter.” Today’s small children would be retiring more than halfa-century from now, but if we can we ought to help now to make their retirement possible. The disastrous floods in Bangladesh occur thousands of miles from Missouri, but Missourians who can ought to provide aid. The moral realism just described, however, inherits the major defects of ordinary consequentialism. The first is epistemological: we cannot know what we ought to do because we cannot know all the consequences of our actions; we cannot even make serious probability judgments about such an indefinite, possibly infinite, totality. The problem is familiar, it was discussed at length by Sidgwick. It arises because of commonsense considerations, not philosophical theories. It renders literal cognitivism questionable. I have discussed it in detail elsewhere.⁹⁶ The second major defect of ordinary consequentialism is metaphysical and not familiar: with respect to such a totality, realism, not just literal cognitivism, is questionable. Wittgenstein was the first to see this. Wittgenstein described the concerns of ethics and religion as “the higher” (das Höhere). But his reason was logical, nor ethical or religious: they are about the limits of the world and thus exceed the limits of what is sayable. A logical category, such as object or fact, might not seem to be something “higher” in the way the concerns of ethics and religion do, but it does enjoy highest generality. Whether the world is the totality of facts or of objects, “world” is a formal concept. But that something falls under a formal concept cannot be said. According to Wittgenstein, ethical statements involve putative reference to the world. Therefore, they attempt to say what cannot be said. They say nothing, even if they show much. The controversy between moral realism and moral antirealism thus becomes a special case of the controversy between metaphysical realism and metaphysical antirealism. If ethics involves putative statements employing formal concepts about the totality of objects or of facts, that is, about the world, as both Moore and Wittgenstein believed, then according to Wittgenstein, though not Moore, both moral realism and moral antirealism must be rejected. Contemporary antirealism in ethics usually begins by rejecting Moore’s view that goodness is a nonnatural property. But, as I pointed out in the previous chapter, even if goodness were a natural property, rightness might still be nonnatural. For, as understood by Moore it involves reference to all the consequences of an action and all the organic wholes to which it belongs. If Wittgenstein is right, such reference would be impossible. There could be no genuine statements about such a totality, just as there could be no genuine statements be about the

 Cf. Skepticism in Ethics.

76

Chapter Four: Saying and Showing The Good

totalities of objects and of facts. Thus, in a different though no less important sense of the word, those totalities could be said to be nonnatural. Yet rightness, natural or nonnatural, is the raison d’être of ethics as a discipline insofar as it is about moral appraisal of action. In the Tractatus Wittgenstein wrote, “[I]t is impossible for there to be propositions of ethics” (6.42). But what he meant was not at all what his positivist successors thought – indeed, it was just the opposite. He went on to explain: “Propositions can express nothing that is higher….Ethics cannot be put into words. Ethics is transcendental” (6.42– 6.421). So was also logic: “Logic is not a body of doctrine, but a mirror-image of the world. Logic is transcendental” (6.13). Wittgenstein took this to follow from the very nature of logic: “The propositions of logic describe the scaffolding of the world, or rather they represent it. They have no ‘subject-matter’” (6.124). Propositions cannot represent logical form. But they show it. Wittgenstein was not a realist, whether in logic or in ethics, but neither was he an antirealist. He was a semirealist. The difference between goodness or rightness and oxygen or yellow color that moral antirealism insists on, even if grossly misunderstands, has been amply demonstrated in its long history. It is also felt by anyone who asks the familiar and seductive but logically puzzling question, “Why should I do what I ought to do?” Few agree with moral realism that goodness or rightness is at all like oxygen or yellow. Yet, that the goodness and rightness are like Pegasus and phlogiston is also seldom taken seriously, except by enfants terrible and philosophes subtile. Few are unable to understand the difference between what ought to be the case and what is the case. Few think of it as resembling the difference between phlogiston and oxygen. Moral antirealists are seldom sensitive to the distinctions that led Wittgenstein to logical and then to moral semirealism. In this respect they differ radically from the philosophical roots they share with him in Kant’s transcendental idealism. Kant proposed a unified account of the “empirically real” and the “transcendentally ideal,” and he vigorously defended the need for acknowledging both. Nothing comparable in scope or depth can be found in standard moral antirealism. Perhaps the reason is the tacit prejudice that the distinction between reality and nonreality does not admit of refinement. Meinong called it prejudice in favor of the actual. By “the actual” he meant, as Moore did, what is in time. Early in the twentieth century, philosophers on both sides of the English channel, including Moore, Russell, Meinong and Husserl, routinely distinguished between being and existence, meaning by the latter what Meinong meant by “actuality,” and unhesitatingly asserting, for example, that relations and numbers have being, but denying that they exist, that they are in time. What exists, they held, is in time. What has being but is not in time does not exist, though

3 The World and the Good

77

some said that it “subsists.” Indeed, usually we speak of the existence of spatiotemporal items, not of items such as relations or numbers. But, unless we are philosophers, we do not say that relations or numbers do not exist, e. g., that there is no such relation as fatherhood or such a number as 5. We usually just ignore the issue and say nothing. Wittgenstein wrote the Tractatus when the distinction between being and existence was familiar, and undoubtedly he was sensitive to the reasons for making it. It is not the same as his distinction between what can be said and what can only be shown, but resembles it in motivation. Contemporary antirealists ignore both distinctions, and assume that being, existence, reality, and actuality are the same, all expressed by the particular (“existential”) quantifier. Wittgenstein in effect showed that this assumption is not so much wrong as primitive. He was particularly sensitive to its failure to fit the special status of “the logical” and “the ethical.” As Kant pointed out, ethics is concerned not with what does happen, but with what ought to happen, even if the latter has never happened and will never happen. Wittgenstein made essentially the same point by saying that the ethical is not in the world. It certainly is not the sort of thing that can be observed and pictured in ordinary physical or mental pictures. But, going beyond anything that Kant or anyone else had held, Wittgenstein concluded that it cannot be pictured even in logical pictures, that it cannot be said: “Ethics cannot be put into words.” If this conclusion was mystical, its mysticism was grounded in logic. In Tractatus 5.62 Wittgenstein wrote, “what the solipsist means is quite correct; only it cannot be said, but makes itself manifest.” The sentence was preceded by “The world and life are one” (5.621) and that “I am my world” (5.63), and was followed by “There is no such thing as the subject that thinks or entertains Ideas.” (5.631). In Part Three I shall have much to say about these seemingly alarming statements. Suffice it here to observe that Hume and Sartre also denied that there is what Wittgenstein called “the subject that thinks or entertains Ideas.” If they are right, then the world and life may indeed be said to be one. And if by “value” is meant the sense or meaning of life, value would not be in the world (6.41), because the sense of the world would not be the sense of any “subject’s” life. Of course, Wittgenstein did not deny that there were human beings, whose lives may be said to have a sense. The Tractatus was not about them. It was about the world and its logic. Clearly, few if any of the usual specific or informative statements can be made about ethical issues if the world is so understood, nor can useful examples can be given. No one should be surprised by the obscurity of Wittgenstein’s ethics. It could not have been clear. Wittgenstein reached the strikingly Spinozistic conclusion that to ask about value and thus the sense of the world requires “view[ing] the world sub specie aeterni,” as well as “feeling” it as a limited whole,” which, he added, is some-

78

Chapter Four: Saying and Showing The Good

thing “mystical” (6.45). Ethics does ask what makes life good, worth living, but “the good life is the world seen sub specie aeternitatis”⁹⁷ This is why Wittgenstein also wrote: “If the good or bad exercise of the will does alter the world, it can alter only the limits of the world, not the facts – not what can be expressed by means of language. In short the effect must be that it becomes an altogether different world. It must, so to speak, wax and wane as a whole…” (6.43). Wittgenstein’s point was not that an action, an “exercise of the will,” does not alter the world, which of course it does. Nor was it that actions are ethically irrelevant. His point was that the value, the goodness or rightness, of an action does not consist in its producing some particular event or events in the world, as standard consequentialism holds; rather, it consists in the world itself becoming different – as a whole, at its limits. It is in this sense that “How things are in the world is a matter of complete indifference for what is higher…” (6.432). Realization of value, whether goodness or rightness, consists, not in the occurrence in the world of some particular event, but in the world itself becoming different, at its limits, in its waxing and waning as a whole (6.43). However, all this can only be shown. There cannot be ethical propositions. The reason is not that, as Wittgenstein’s early followers thought, there is nothing for such propositions to be about. The reason is that what they purport to say cannot be said, though it can be shown (6.42). “The sense [Sinn] of the world must lie outside the world… For all that happens and is the case is accidental” (6.41).” The sense of the world is not something in the world, because it is the sense of the whole world. This is why it may be said to “lie outside the world,” at its “limits.” Wittgenstein’s assertion that value is not anything that happens because it is not accidental need not mean more than what in the Foundations of the Metaphysics of Morals Kant called the third fundamental proposition of morality: duty is the necessity of acting from respect for the law. Duty is what ought to happen, “must” happen. What happens is already in the world, what ought to happen is not yet, or perhaps ever. In the Notebooks, Wittgenstein had written: “To believe in a God means to understand the question about the meaning of life… to see that life has a meaning.”⁹⁸ The sense or meaning of life is the ultimate value. It constitutes “[t]he solution of the riddle of life,” but that solution “lies outside space and time” (6.4312). It is not an item in the world because it is the sense of the whole world. What Wittgenstein called the riddle of life presumably concerns the sense or meaning of life. This, indeed, was a central topic in traditional ethics. To ask about ultimate value is to ask

 Notebooks, 83e. Wittgenstein adds: “the work of art is the object seen sub specie aeternitatis.”  Notebooks, 74.

3 The World and the Good

79

about the meaning of life, what makes life worth living. (Plato’s Republic began with Cephalus’s claiming that what makes life worth living is not pleasure, as commonly thought, but justice.) One who asks about the meaning of one’s life sometimes phrases the question as asking about “the sense of it all.” Life can hardly be fully meaningful in a meaningless world. Indeed, that the world exists at all, that there is something rather than nothing, has been for some the ultimate object of joy or sorrow, and certainly of wonder. Much later, in 1929, Wittgenstein tried to explain: “What is good is also divine. Queer as it sounds, that sums up my ethics. Only something supernatural can express the Supernatural.”⁹⁹ This “supernatural” was as central to Wittgenstein’s ethics as the “supersensible” was to Kant’s. Elsewhere, also in 1929, he wrote, “[Attributions of] absolute value are nonsensical but their nonsensicality [is] their very essence… [A]ll I wanted to do with [those attributions] was to go beyond the world and that is to say beyond significant language.”¹⁰⁰ Wittgenstein meant, however, not that such attributions would be gibberish, but that they would not be logical pictures, in the sense required by his theory of meaning. Wittgenstein’s remarks about ethics left most of his readers bewildered, and his claim that there is something “mystical,” which cannot be expressed, was unpalatable to his heirs in analytic philosophy. But that there is what he called “the mystical” is also a consequence of his fundamental views about logic, and these cannot be easily rejected. The existence of the (“actual”) world, indeed of something, may not be mystical, but it is a basic presupposition of logic. And how to express existence (with a predicate or just the existential quantifier?) has remained deeply controversial since Anselm proposed his ontological argument for the existence of God. Noteworthy signposts in the ongoing discussion have been Kant’s denial that existence is a “real predicate,” Meinong’s theory of objects, Russell’s vehement attack on it, and present-day quandaries about what it is for one world to be “actual” and the rest just “possible.” Children find the difference between what exists and what does not, say, between their real and imaginary playmates, simple and easy to comprehend, but philosophers do not.

 Ludwig Wittgenstein, Culture and Value, edited by G. H. von Wright and translated by Peter Winch (Chicago: University of Chicago Press, 1980), 3e.  Ludwig Wittgenstein, Philosophical Occasions 1912 – 1951 (Indianapolis, IN: Hackett, 1993), 40, 44.

Part Two: Metaphysics Humanized It is solely from the human standpoint that we can speak of space, of extended things Kant

I have argued that insofar as they are acceptable, epistemology is about certain nonformal inferences and ethics is about the world. The case of metaphysics, however, is different and much more complex. In the Introduction I suggested that while anthropocentrism in epistemology and ethics is natural and understandable but indefensible, in metaphysics it is unnatural and almost incomprehensible but at least as defensible as Kant’s transcendental idealism and contemporary antirealism. In metaphysics it is due not to humans’ natural interest in themselves, as it is in epistemology and ethics, but to purely philosophical considerations. The present Part is devoted to making these considerations clear. In effect, it explains and defends antirealism, the humanization of metaphysics. For there can be no justification for just rejecting it and returning to pre-Kantian metaphysics. Yet, antirealism seems to lead, however tacitly, to the absurdity of human creationism. The explanation of how antirealism may avoid anthropocentrism, and thus how antirealist metaphysics may be dehumanized, will be our task in Part Three. The subject matter of metaphysics is traditionally described as the world or reality. According to the Oxford English Dictionary, “world” usually signifies “The material universe; the cosmos,” or just a planet, perhaps Earth, but it applies also to “Any state or realm of existence.” In the latter sense it is a synonym of “reality,” in the OED sense of “the aggregate of real things or existences.” Reality may include, in addition to the spatiotemporal world, abstract, nonspatiotemporal entities such as Platonic universals and numbers, as well as God, who by being its creator is not in the spatiotemporal world. In discussions of our topic, however, the terms “world” and “reality” are usually interchangeable. The term “realism” applies to the metaphysical view that the world is independent of our faculties of cognition, in the contemporary sense in which cognition includes perceiving and understanding (conceptualizing) as well as describing (by writing or speaking). I have used the term “antirealism” for the opposite view. Its versions are as diverse as Berkeley’s “immaterialism,” Kant’s “transcendental idealism,” Hegel’s “absolute idealism,” and recent positions such as Hilary Putnam’s “internal realism,” Michael Dummett’s “antirealism,” and “Nelson Goodman’s “irrealism.” Metaphysical realism and antirealism are usually understood as concerned with all reality, and therefore should be distinguished from realism and antirealism regarding just an alleged part of it, e. g., moral values (moral realism/antirealism), abstract entities (Platonic realism/antirealism), the theoretical entities of physics (scientific realism/antirealism), and so on. My concern in this and the next Part is with metaphysical realism and antirealism. But, as we saw in the Introduction, even within them, important distinctions are needed. One can be a realist regarding “things” but an antirealist regarding “facts” and thus regarding the “world,” if the world is the totality of

84

Part Two: Metaphysics Humanized

facts, not of things. One can be a realist regarding the spatiotemporal/physical and causal structure of the world but antirealist regarding its logical structure: One may believe that there are physical objects but deny that there are logical objects. Moreover, asserting and denying the reality of an item are not the only options. Bertrand Russell held that some things do not exist but subsist, and Wittgenstein, as we saw in the previous chapter, that some things cannot be said but show themselves.

Chapter Five: The Role of Language in Cognition 1 The empirical and the a priori question The most readily understood and least controversial version of metaphysical antirealism today is linguistic antirealism. It has dominated contemporary philosophy. Its immediate appeal rests on the plausible belief that cognition, or at least thought, is impossible without language. The present chapter will be devoted to linguistic antirealism insofar as it rests on this belief. The next two chapters will consider the more technical considerations that may lead to it. Unlike Kant’s transcendental idealism, which focused on the role of our faculties of sensibility and understanding in shaping cognition and thus the world as cognized, linguistic antirealism focuses on the role of our language faculty, or “organ” as Steven Pinker has called it. Contemporary antirealists thus avoid Kant’s questionable appeal to inhabitants of consciousness such as concepts, let alone sensations and ideas. For, as David Armstrong has remarked, “Concepts [understood as mental items] are a more mysterious sort of entity than linguistic expressions.”¹⁰¹ The language faculty may in some sense be “mental,” but its role in cognition can be understood and defended, as indeed it was by Wittgenstein, Ryle, Austin, and Strawson, without appeal to any manifestations of its work in consciousness. Linguistic antirealism thus seems to avoid the absurd implication that the world is an inhabitant of one or several humans’ consciousness. Moreover, even if the human mental faculties are in some sense properties of the human brain, the role of the language faculty in cognition can be understood and defended without appealing to neuroscience. Linguistic antirealism thus seems to avoid also the absurd implication that the world is in one or several humans’ brains. Of course, it would be no less absurd to hold that the world is in language, that it is linguistic. But in Part Three we shall find that this seeming implication of linguistic antirealism can be avoided. Wittgenstein did write: “The limits of my world are the limits of my language,” but he also wrote, “I am my world,” which makes no mention of language. And Heidegger insisted that it is language, not we, that “speaks.” These are obscure pronouncements, to which I shall return. They suggest that linguistic antirealism can be understood in such a way that it involves no reference to language as a human, zoological, phenomenon. But there would have been little motivation for accepting it if human cognition were not believed to be essentially or at least importantly linguistic, even though  David Armstrong, Universals and Scientific Realism, 25.

86

Chapter Five: The Role of Language in Cognition

this is obviously an empirical topic and neither Wittgenstein nor Heidegger thought of himself as engaged in empirical research. I shall therefore begin the present Part by considering the role of language in cognition. The linguistic turn was the central event in both Anglo-American and continental 20th century philosophy. According to Michael Dummett, after Frege philosophy of language replaced epistemology as the central branch of philosophy, just as epistemology had replaced metaphysics after Descartes. The phrase “linguistic turn” was coined by Gustav Bergmann and popularized by Richard Rorty in order to characterize recent Anglo-American philosophy, but as we saw it applies also to much of contemporary continental philosophy. The turn began with the publication of Wittgenstein’s Tractatus Logico-Philosophicus after World War I and the birth shortly thereafter of logical positivism in Austria, but it hastened with the migration in the following two decades of Wittgenstein and most of the positivists to Great Britain and especially the United States. In addition to claiming, as the 19th century positivists had done, that all knowledge is empirical and in principle scientific, the logical positivists held that philosophy, which obviously is not an empirical science, nonetheless does have a subject matter: language. To distinguish philosophy from the empirical disciplines of linguistics and lexicography, however, they identified that subject matter more specifically as what they called the logic of language. For logic, of course, has always been a branch of philosophy, and a paradigm of an a priori discipline. But the linguistic turn probably would not have been taken were it not for the assumption, often tacit, that cognition, or at least thought, without language is impossible. The philosophers who took the turn considered this assumption true as a matter of empirical fact, as indeed it seems to be, except, as we shall find in section 4 of this chapter, in the crucial case of what I have called logical cognition. Yet few attended to the relevant empirical facts. For example, a highly relevant fact, to which I shall return, is the apparent occurrence of cognition of numbers, or at least of “numerosity,” in prelinguistic children and even some nonhuman animals. No armchair philosophy can discuss it responsibly, yet obviously it is an empirical datum crucial for the general question whether cognition without language is possible. And the ultimate value of the linguistic turn, indeed much of the subsequent course of Anglo-American philosophy, did seem to depend on how that question should be answered. I have said repeatedly that philosophy cannot competently investigate empirical facts, and this applies also to the facts about the dependence of cognition on language. But it does not mean that philosophy should be unaware of them. After all, they were the larger context of the linguistic turn and thus of the varieties of antirealism most common today. Some of these empirical facts are so obvious as to require no special competence or investigation, e. g., that mathemat-

1 The empirical and the a priori question

87

ics requires symbols. But other relevant facts are not obvious, including the alleged fact that speakers of different human languages perceive and think of the world differently. In this chapter we shall look at what has been said about these issues, first by philosophers and then by linguists and psychologists. I shall not attempt to provide a comprehensive account. For that a separate book, indeed several books, would be needed. And we must keep in mind throughout that the linguistic turn and linguistic antirealism are logically independent of each other, even though the former encouraged the adoption of the latter. Linguistic antirealism is a metaphysical position. The linguistic turn was essentially the adoption of a philosophical method. Unfortunately, the discussions by philosophers of the question whether cognition without language is possible are seldom detailed, and the discussions of it by scientists are seldom useful. Philosophers have relied usually on speculations or, at best, exercises in introspection. Scientists have been hampered by the virtual impossibility, moral and social, of serious experimentation with children before and after they acquire language. That the most advanced, the “hard,” sciences – physics, chemistry, biology – are experimental, not just empirical, while the “soft” sciences – economics, sociology, political science, and much of psychology – are not, is not an accident. Experiments vastly augment not only the number but, more importantly, the variety of relevant empirical data and thus make detection of significant, rather than merely coincidental, correlations much easier. That there is a connection between cognition and language was hinted by the historically influential use of logos in Greek for both reason and word. The definition that man is a rational animal could be understood also as saying that man is a speaking animal. This is why the question whether cognition without language is possible has often been taken to ask whether thought without language is possible. The first definition of “thought” in the Oxford English Dictionary reads as follows: “The action or process of thinking; mental action or activity in general, esp. that of the intellect; exercise of the mental faculty; formation and arrangement of ideas in the mind.” We are also told that “think” is “The most general verb to express internal mental activity, excluding mere perception of external things or passive reception of ideas.” As the OED implies, “thought” and “think” have both a wide sense, that of mental action or activity in general, and a narrow sense, that of mental action or activity of the intellect. They have these two senses also in philosophy, though philosophers, especially Brentano and Husserl, usually add that the action or activity is “intentional,” directed upon an object, real or not, and describe being conscious of something as an “act.” Descartes used the words in their wide sense. After he offered the argument “I think [cogito, pense], therefore I am,” he explained, “What is a thinking thing [res cogitans, une chose qui pense]? It

88

Chapter Five: The Role of Language in Cognition

is a thing that doubts, understands, conceives, affirms, denies, wills, refuses; that imagines also, and perceives.” Clearly, by “thought” Descartes meant what later philosophers, in particular, Brentano, Husserl, and Sartre, meant by “consciousness.” Philosophical theories of thought in this sense are essentially philosophies of mind. For example, soon after Descartes offered his argument, the British empiricists proposed theories of mental activity as a “succession” or “association” of discrete “ideas,” whether abstract ideas (roughly the “concepts” of later philosophy), ideas of “sensation” (later called sense-data), or ideas of the imagination (later called mental images). Descartes drew a sharp distinction between a “thinking thing” and the human body. A thinking thing is “simple,” it has no parts. The body, in contrast, is a paradigm of an entity that has parts, “organs” – an “anatomy,” in the modern literal as well as the etymologically pregnant sense. Indeed, one of the standard arguments for the immortality of the soul was that it is simple and therefore incapable of “corruption,” “disintegration,” “falling apart,” which was the common conception of how a thing ceases to exist. Later philosophers more commonly used “thought” in its narrow sense, that of mental action or activity of the intellect. In that sense, thought is the level of cognition expressed in Greek by noēsis and in English by “understanding,” “conceptualization,” or “judgment.” Unlike a neonate’s sense perception, which seems to be much like that of nonhuman animals, thought understood as activity of the intellect involves the application of concepts and seems distinctively human. Indeed, in the 17th century John Locke declared that “brutes think not,” on the grounds that they are incapable of entertaining “abstract ideas” and using language. In the 18th century, Kant rejected the atomism of the empiricists’ view of mental activity as a succession of ideas by pointing out the synthesizing function of the “understanding” (Verstand). In the 19th century, Franz Brentano distinguished among presentation, judgment, and phenomena of interest. Judgment necessarily involves presentation, because one must be aware of what the judgment is about (whether by sensing it or through a concept). And judgment makes possible the characteristically mental acts of acceptance or rejection, belief or disbelief.¹⁰² In the 20th century, H. H. Price followed Kant by regarding thinking as the employment of concepts, even in recognition, and argued that a concept is essentially a principle of classification.

 Franz Brentano, Psychology from an Empirical Standpoint (London: Routledge, 1973), trans. by A.C. Rancurello, D.B. Terrell, and L. McAlister. 2nd ed., intr. by Peter Simons, 1995.

1 The empirical and the a priori question

89

Unlike primitive sense perception and imagination, a judgment has as its object a proposition, what is “judged” to be true or false. And propositions usually are understood in philosophy as the primary vehicles of truth-value. Judgments, sentences, beliefs, opinions, etc., are said to be true or false in virtue of the propositions they “express.” It may be said, however, that while a declarative sentence expresses a proposition, the assertion of that sentence expresses a judgment. Frege used “thought” (Gedanke) in a sense close to that of “proposition.” Thoughts, he held, belong in a “third realm,” neither mental nor physical, and thus they are not judgments, which presumably are mental.¹⁰³ If a judgment is expressed by the assertion of a declarative sentence, it can be expected to have parts and a structure analogous to those of a sentence. Indeed, Jerry Fodor has argued that there is a “language of thought,” thus attributing to thought the sort of structure characteristic of language, though he regards thought to be a state or activity of the brain.¹⁰⁴ And Gustav Bergmann analyzed all mental acts as “complexes,” facts or states of affairs, each consisting of a momentary particular and two attributes that the particular exemplifies: a “species” that determines the kind of mental act it is (e. g., perceiving, remembering, imagining) and a “proposition” (sometimes called “text” by Bergmann) that determines which state of affairs is its object (“intention”). He held that even the objects of ordinary sense perception and imagination are states of affairs, and thus require sentences for their expression in the “ideal language.”¹⁰⁵ In this book the question whether there is thought without language would have a clearer and more promising focus if we ask, instead, whether there is cognition without language. For it is the term “cognition,” not “mind” or even “judgment,” that captures literally what is relevant to the epistemological and metaphysical issues that give rise to antirealism. Yet, even then, the question would remain too broad, partly because the affirmative answer seems obvious. Neonates do enjoy sense perception, surely a level of cognition, however rudimentary, months or years before they learn to speak; they also communicate, though not by speaking but often by crying. Later, though still before being able to speak, they enjoy recognition, e. g., of Mother, a primitive level of conceptual cognition, and perhaps they even engage in rudimentary reasoning, e. g., that crying leads to being fed. We shall find in section 4 that only logical cognition can be considered clearly impossible without language, because only there the impossibility is logical.  “Thought,” in Beaney, The Frege Reader.  Jerry Fodor, The Language of Thought (Cambridge: Harvard, 1975).  Gustav Bergmann, Meaning and Existence (Madison, University of Wisconsin Press, 1959). See also New Foundations of Ontology.

90

Chapter Five: The Role of Language in Cognition

Since language is an empirical subject matter, the study of its role in cognition should also be empirical. In keeping with the stance explained in the Introduction regarding philosophical inquiries into empirical matters, the present chapter will offer no theories about language. It will be limited to a brief review of what others have said. Careless observation and incautious generalization from one’s own experience can be grossly misleading. A useful discussion, scientific or philosophical, of the question whether cognition without language is possible must be specific regarding the respects in which it might or might not be possible. It must also be specific regarding the kinds of cognition, language, and possibility that are at issue. Elementary and obviously relevant distinctions must be made. Many of the most familiar views tend to be intolerably general and vague. Surely, there are fundamental differences between sense perception and calculation, between the language of babes and the language of bards, and between causal impossibility and logical impossibility. Judgments about the causal possibility or impossibility of cognition without language are especially hazardous. One may be unable to balance one’s checkbook without talking to oneself, but other people can do it. When driving in a strange city, one may be compelled to issue to oneself instructions like “Turn left on 3rd street, then right at second stop-light,” but other people might not. The professional activities of some people, especially philosophers, consist almost entirely of talking, reading, and writing. (Sartre titled his autobiography Words.) But the activities of farming, fishing, and acrobatics do not. Language might not be causally necessary even for some relatively advanced levels of cognition, such as designing a house and planning a trip. But surely it is necessary for doing physics or mathematics. A fly’s perceptual cognition does not require language, but any conceptual cognition that whales perhaps engage in might require it. In the case of extraterrestrial life forms, angels, or God, language might not be necessary for any cognition. To be sure, a human being who lacks a language cannot do physics or mathematics, but surely God can, and for all we know so can intelligent beings outside our solar system. We may not understand what such cognition might be like, but neither can we form auditory images of the high-frequency sounds that dogs but not humans can hear. Humans certainly need language, symbols, for cognition in physics and mathematics, but this may be due to their limited cognitive powers. A god might cognize directly, for example, the ultimate constitution of matter, without relying on language or any symbols, including those of mathematics. The truth is that, so far, there is neither sufficient knowledge nor conceptual maturity in psychology or biology to judge responsibly whether thought or cognition is causally possible without language. But judgments about whether some

2 Philosophical opinions

91

forms of it are logically possible can be made in philosophy. This question is not empirical, it is properly philosophical. We will come to it in section 4.

2 Philosophical opinions Let us briefly review what philosophers have said about the role of language in cognition. In the Cratylus Plato sharply distinguished between names and the things they name, considered whether names are arbitrary and conventional, and argued that the study of names is inferior to the study of the things named. But, at the beginning of modern philosophy, Bacon wrote, “Men believe that their reason governs words; but it is also true that words react on the understanding; and this it is that has rendered philosophy and the sciences sophistical and inactive.”¹⁰⁶ Hobbes dwelt on language in detail, arguing that “‘true’ and ‘false’ are attributes of speech, not of things.”¹⁰⁷ Locke held that brutes abstract not, because they talk not: “the power of abstracting… [is] an excellency which the faculties of brutes do by no means attain to.”¹⁰⁸ Leibniz wrote, “Nothing exists in the intellect that was not before in the tongue – except intelligence itself.” Hegel remarked that “[t]he forms of thought are, in the first instance, displayed and stored in human language.”¹⁰⁹ This last opinion was accepted by most 20th century philosophers. As we have seen, Wittgenstein declared that “the limits of the language … mean the limits of my world,” and Heidegger announced that “Language is the house of being.” But in Thinking and Experience, published in 1953, H. H. Price wrote, “it is sometimes supposed that no intellectual activity of any kind can occur without the use of words. This is not true of recognition… Recognition is a prelinguistic process in the sense that it is not dependent on the use of words… [W]ords themselves have to be recognized.”¹¹⁰ Price’s view was plain common sense. Recognition is a level, indeed a fundamental level, of all cognition, primitive or advanced. It is present in children long before they acquire language. In adults it often occurs when no name or even description of what is recognized is availa-

 Works, Novum Organum, Aphorisms Concerning the Interpretation of Nature and the Kingdom of Man, LIX.  Leviathan, Part I, Chapter IV.  John Locke, An Essay Concerning Human Understanding Vol. 1 (London: Dent, 1960), 126. But in his Treatise of Human Nature, Hume disagreed: “no truth appears to me more evident, than that beasts are endowd with thought and reason as well as men.”  Science of Logic, trans. A. V. Miller (London: George Allen, 1969), 31.  Thinking and Experience, 36 – 38.

92

Chapter Five: The Role of Language in Cognition

ble, for example when recognizing, “reading,” facial expressions.¹¹¹ And creative work in music or painting, as well as in mathematics and advanced theoretical physics, only minimally involves the use of a natural language, though in mathematics and physics it does require mathematical symbols. Price felt obliged to say what I quoted because of the dominance at the time of the philosophy of ordinary language, according to extreme versions of which, he wrote, “an intelligent being…must always be talking to himself or to others.” Indeed, in the Philosophical Investigations, which by coincidence was published (posthumously) also in 1953, Wittgenstein wrote, “When I think in language, there aren’t ‘meanings’ going through my mind in addition to the verbal expressions: the language is itself the vehicle of thought…..”¹¹² Expounding what he took to be Wittgenstein’s view, Renford Bambrough argued in 1960 that objects called by the same name are distinguishable from other objects just by the fact that they are uniquely called by that name: they neither have a common property nor bear to each other resemblances except for what Wittgenstein had called family resemblances.¹¹³ In 1973 Michael Dummett wrote that “to possess a concept is to be a master of a certain fragment of language,”¹¹⁴ and in 1975 that “Only with Frege was the proper object of philosophy finally established: namely, first, that the goal of philosophy is the analysis of the structure of thought; secondly, that the study of thought is to be sharply distinguished from the psychological process of thinking; and, finally, that the only proper method for analyzing thought consists in the analysis of language.”¹¹⁵ In 1982 we find Donald Davidson declaring that “a creature cannot have a thought unless it has language.”¹¹⁶ In 1995 Quine wrote that “thought, as John B. Watson claimed, is primarily incipient speech,” though Quine allowed that in the case of artists, acrobats, and engineers, sometimes they are “thinking with nonverbal muscles.”¹¹⁷ However, after thus paying homage to muscular thought, Quine declared that be-

 Noam Chomsky, Rules and Representations, 248. Chomsky appeals to innate structures and speaks of a possible “universal grammar of faces.”  Ludwig Wittgenstein, Philosophical Investigations, trans. G.E.M. Anscombe, (Oxford: Blackwell, 1991), 329 – 330.  J. R. Bambrough, “Universals and Family Resemblances,” Proceedings of the Aristotelian Society, LXI (1960 – 61): 207– 223.  Michael Dummett, Truth and Other Enigmas (Cambridge: Harvard University Press, 1978), 438.  Truth and Other Enigmas 458.  Donald Davidson, “Rational Animals,” in Subjective, Intersubjective, Objective (Oxford: Oxford University Press, 2001), 100.  W.V. Quine, From Stimulus to Science (Cambridge, Mass.: Harvard University Press, 1995), 88 – 89.

2 Philosophical opinions

93

liefs, meanings, ideas, properties, and propositions are all entia non grata, adding that “believing, doubting, hoping, expecting, regretting, all continue alive and well, and their objects, by my lights, are sentences.”¹¹⁸ Even Nelson Goodman, who for years had held that what he called “symbol systems” need not be linguistic, slipped in 1988 and denied that “there is a readymade world beyond discourse.”¹¹⁹ Nevertheless, analytic (though perhaps not continental) philosophy soon went beyond these extremist views. In 1992 we find Dummett himself asserting that “linguistic practice is no more sacrosanct, no more certain to achieve the ends at which it is aimed, no more immune to criticism or proposals for revision, than our social, economic or political practice.”¹²⁰ Though intended as an attack on the philosophy of ordinary language (especially J. L. Austin’s), Dummett’s assertion had independent philosophical significance by tacitly implying that criticism of language presupposes cognition that is not dependent on language. It is noteworthy that, except perhaps for Quine’s casual reference to Watson, these opinions about language were offered without mention of any empirical data or of the relevant empirical sciences. Some have come to believe that, though all cognition, from simple perception to mathematical reasoning, is necessarily representational, symbolic, the representations need not belong to a natural language – they can belong to a “language of thought,” in Jerry Fodor’s phrase.¹²¹ For Fodor this was just a philosophical hypothesis, though he thought it has the support of neuroscience. Tim van Gelder has explained: “Contemporary orthodoxy maintains that it [cognition] is computation: the mind is a special kind of computer, and cognitive processes are the rule-governed manipulation of internal symbolic representations.”¹²² But then he drew the natural conclusion “that because the cognitive system traffics only in symbolic representations, the human body and the physical environment can be dropped out of consideration; it is possible to study the cognitive system as an autonomous, bodiless, and wordless system whose function is to transform input representations into output representations.”¹²³ If cog From Stimulus to Science, 93.  Nelson Goodman and Catherine Z. Elgin, Reconceptions in Philosophy (Indianapolis & London: Hackett, 1988), 154. But elsewhere, including this same work (see vii, 9, 11, 155, and section VIII, 3), Goodman points out that what he calls “symbol systems” need not be linguistic, that they could be, e. g., notational or pictorial.  Michael Dummett, The Logical Basis of Metaphysics (London: Duckworth, 1991), 214– 215.  The Language of Thought (Thomas Y. Crowell Co., 1975).  Tim van Gelder, “What Might Cognition Be, If Not Computation?” The Journal of Philosophy, Volume XII, number 7 (July 1995), 345.  van Gelder, 373.

94

Chapter Five: The Role of Language in Cognition

nition is what such a cognitive system does, cognition does not involve language, even if some neural states are described as “symbolic,” though neither is it also directed upon objects and thus it is not cognition of reality – or of anything. At most these neural states would be like the characters employed by a computer, but as John Searle had shown in 1980, the idea that they mean anything other than what the programmer means by them is grossly mistaken.

3 Scientific opinions Let us now briefly review what linguists and psychologists have said about the role of language in cognition. More than a century after Locke announced that brutes abstract not, the German linguist Wilhelm von Humboldt wrote, “Language is the formative organ of thought…. Thought and language are therefore one and inseparable from each other.”¹²⁴ Humboldt offered no evidence in support of these opinions. He did speculate, however, that existing languages differ from each other markedly in regard to “perfection,” which he attributed to national, ethnic, and racial differences among their speakers. Early in the 20th century, Ferdinand de Saussure wrote (or was reputed to have said – the text was reconstructed from lecture notes): “Psychologically, setting aside its expression in words, our thought is simply a vague, shapeless mass. Philosophers and linguists have always agreed that were it not for signs, we should be incapable of differentiating any two ideas in a clear and constant way. In itself, thought is like a swirling cloud, where no shape is intrinsically determinate. No ideas are established in advance, and nothing is distinct, before the introduction of linguistic structure.”¹²⁵ Saussure did not say how he (or the philosophers and linguists to whose authority he appealed) learned all this. For example, did he study human infants before they learned to speak? William James famously announced that the world perceived by an infant is a “blooming buzzing confusion,” but he also failed to say how he knew this.¹²⁶ James did think, however, that thought without language is “perfectly possible.” In support, he quoted at great length a deaf-mute instructor, “Mr. Ballard,” who was reported to have written: “It was during those delightful rides, two or three years before my initiation into the rudiments of written language, that I began to ask myself

 Wilhelm von Humboldt, On Language, (Cambridge: Cambridge University Press, 1988), 54– 55.  Ferdinand de Saussure, Course in General Linguistics (Chicago and LaSalle, Ill.: Open Court, 1986), 111 [155].  William James, The Principles of Psychology (New York: Henry Holt, 1890), 46.

3 Scientific opinions

95

the question: How came the world into being? When this question occurred to my mind, I set myself to thinking it over a long time.” Concerning this passage, Wittgenstein later wrote, “Are you sure – one would like to ask – that this is the correct translation of your wordless thought into words? Do I want to say that the writer’s memory deceives him? I don’t even know if I should say that. These recollections are a queer memory phenomenon, – and I do not know what conclusions one can draw from them about the past of the man who recounts them.”¹²⁷ Despite Wittgenstein’s misgivings, such reports of personal experience, if truthful, are useful. They must be read with caution, but the evidence they provide is unavailable elsewhere. For example, if Helen Keller’s acquisition of language was indeed as she and her teacher Anne Mansfield Sullivan claimed, then it was an extraordinary example of how, despite the absence of the usual sort of exposure to language, one can acquire a rich, indeed stylistically first-rate, linguistic proficiency. Yet Keller’s editor did write: “Philosophers have tried to find out what was her conception of abstract ideas before she learned a language. If she had any conception, there is no way of discovering it now; for she cannot remember, and obviously there was no record at the time. She had no conception of God before she had the word ‘God.’”¹²⁸ But another report of personal experience, by the mathematician Kalvis M. Jansons, is instructive and may serve as an example of what is needed: “From an early age I found that many things were easier to think about without language. This usually, but not always, meant thinking in terms of pictures and was particularly true when trying to make or understand intricate mechanisms. … To me, abstract pictures and diagrams feel more important than words.”¹²⁹ Not all reports of personal experience on our topic are as credible as Jansons’. This is especially true of some familiar claims by leading psychologists, which must be understood as reports of personal experience since no other evidence is cited. For example, J. B. Watson wrote that “thinking is merely talking, but talking with concealed musculature.”¹³⁰ Leonard Bloomfeld held that thinking is talking to ourselves, “suppressing the sound-producing movements and replacing them by very slight inaudible ones.”¹³¹ (I should mention that today

 Philosophical Investigations, 34.  Helen Keller, My Story, ed. John Albert Macy, 234. Mark Twain is reported to have said that the two most interesting characters of the nineteenth century were Napoleon and Helen Keller (The Principles of Psychology, 225).  Kalvis M. Jansons, “A personal view of dyslexis and of thought without language,” in L. in Thought without Language, ed. L. Weiskrantz (Oxford: Clarendon Press, 1988), 502– 503.  J. B. Watson and W. McDougal, The Battle of Behavior (New York: Norton, 1929), 33.  Leonard Bloomfield, Language (Chicago: University of Chicago Press, 1933), 28.

96

Chapter Five: The Role of Language in Cognition

there is evidence that thinking is not prevented by drugs that suppress muscular activity.) B. F. Skinner wrote, “thought is simply behavior – verbal or nonverbal, covert or overt.”¹³² Edward Sapir appealed to “the frequent experience of fatigue in the speech organs” after “intensive thinking.”¹³³ This, presumably, was a report of personal experience. But elsewhere Sapir went much farther: “Human beings do not live in the objective world alone…..but are very much at the mercy of the particular language which has become the medium of expression for their society…..We see and hear and otherwise experience very largely as we do because the language habits of our community predispose certain choices of interpretation. “¹³⁴ This was a crude version of linguistic antirealism, but offered as a scientific hypothesis and then left unsupported by empirical evidence. In 2008, field linguists discovered a “new” language, Koro, spoken in some villages in northeastern India.¹³⁵ One of the linguists was reported as saying that languages like Koro “construe reality in very different ways…They uniquely code knowledge of the natural world in ways that cannot be translated into a major language.” But we are not told how the linguist knew this. Sapir’s student, Benjamin Lee Whorf, did engage in extensive empirical work on Indian languages. But he cited no particular evidence when he wrote, “Thinking…follows a network of trails laid down in the given language…The individual is utterly unaware of this organization and is constrained completely within its unbreakable bonds.”¹³⁶ Whorf thought that he and a Hopi could not discuss the “same” world. According to him, “[S]egmentation of nature is an aspect of grammar…We cut up and organize the spread and flow of events as we do, largely because, through our mother tongue, we are parties to an agreement to do so, not because nature itself is segmented in exactly that way for all to see.”¹³⁷ But, again, no evidence is provided.¹³⁸ Perhaps it was natural that

 B.F. Skinner, Verbal behavior (New York: Appleton-Century-Crofts, 1957), 449.  Edward Sapir, Language (New York: Harcourt, Brace & World, 1921), 19.  Edward Sapir, Language, Culture and Personality (ed. David G. Mandelbaum, Berkeley & Los Angeles: University of California Press, 1963), 16. (Reprinted from “The Status of Linguistics as a Science,” Language 5, 209.) See also A. H. Bloom, The Linguistic Shaping of Thought: A Study in the Impact of language on Thinking in China and the West; Laurence Erbaum, 1981; Richard B. Brandt,  New York Times, October 11, 2010.  Benjamin Lee Whorf, Language, Thought and Reality: Selected writings of Benjamin Lee Whorf, ed. John B. Carroll, (New York: John Wiley and Sons, 1956), 226. Zellig Harris described Whorf’s view as “occupational imperialism for linguistics,” in ‘Distributed Structure’, in The Structure of Language: Readings in the Philosophy of Language, eds., Jerry A. Fodor and Jerrold J. Katz (1965), 38, n. 1.  Language, Thought and Reality, 240, reprinted from Language 5 (1929): 207– 214).

3 Scientific opinions

97

what came to be known as the Sapir-Whorf hypothesis was offered in American linguistics. It was in America that non-Indo-European languages were plentiful and readily available for study: there were more than 100 American Indian languages. In this respect American linguistics had an important advantage over European linguistics. Watson, Bloomfield, Sapir, Whorf, and Skinner were the most influential scientists on our topic in the first half of the 20th century. This may explain in part the unbridled linguisticism characteristic of American philosophy during that period. Respect for science was characteristic of the logical positivism dominant then. The supposedly scientific view offered by the psychologists seemed identical with the philosophical view that Hilary Putnam described as holding that “objects arise out of discourse, rather than being prior to discourse.”¹³⁹ But were Watson, Bloomfield, Sapir, Whorf, and Skinner right? It is generally agreed that to answer this question on empirical grounds and with confidence we would need to study the prelinguistic child’s cognition. And such studies have been notoriously unsuccessful, probably because serious experimentation with children, e. g., delaying a child’s learning a language, would be grossly unethical. Less alarming experiments tend to be inconclusive, perhaps because small children, not just infants, seldom cooperate with or even take seriously the investigators, who often happen to be their parents. This seems true, for example, in the case of the theorizing about early cognition and language acquisition by Bärbel Inhelder and Jean Piaget, which tends to be little more than speculation.¹⁴⁰ The most important development in linguistics, of course, took place in the second half of the 20th century. It was due to Noam Chomsky. Chomsky is not only a major linguist, he is also a philosopher, whose views (especially in his later writings) on scientific methodology and the mind-body problem are sophisticated and subtle. But it is his work in linguistics that is of special relevance to the topic of this chapter. Its central thesis was anticipated by Humboldt: “Since the natural disposition to language is universal in man, and everyone must possess the key to the understanding of all languages, it follows automatically that

 But Whorf did add: “Our Indian languages show that with a suitable grammar we may have intelligent sentences that cannot be broken into subjects and predicates…When we come to Nootka, the sentence without subject or predicate is the only type…Nootka has no parts of speech; the simplest utterance is a sentence, treating of some event or event-complex” (Language, Thought and Reality, 242).  Hilary Putnam, Realism and Reason. Philosophical Papers. Volume 3 (Cambridge: Cambridge University Press, 1983), xvi.  For example, Bärbel Inhelder and Jean Piaget, The Growth of Logical Thinking (New York: Basic Books, 1958), trans. Anne Parsons and Stanley Milgram.

98

Chapter Five: The Role of Language in Cognition

the form of all languages must be essentially the same, and always achieve the universal purpose.”¹⁴¹ This thesis is essentially Kantian in spirit, and it has been defended throughout Chomsky’s works. Discussions of Chomsky’s linguistics often bog down in what he rightly regards as misguided puzzling over whether what he calls universal grammar has “psychological reality,” rather than just utility as a theoretical hypothesis. Surely he is right that there is no principled difference between the propriety of hypotheses in linguistics and the propriety of hypotheses in physics.¹⁴² Chomsky has always held that the status of innate linguistic competence is ultimately biological.¹⁴³ The “innate structures” to which he appeals are not accessible to consciousness, presumably because they are in the brain.¹⁴⁴ Chomsky’s nativism is not what is argued in this book, if for no other reason than that, unlike what a philosophical book can properly attempt, it is essentially scientific and subject to empirical confirmation. But it does imply that what I have called the logical structure of language is not learned from experience. Chomsky has claimed, for example, that “the familiar [quantifier-variable] notation is ‘read off of’ the logical form that is the mental representation for natural language.”¹⁴⁵ But this is offered as a substantive scientific hypothesis. And the mental representation in question is ultimately identified with a state or feature of the brain. The hypothesis is hardly philosophical, even though, as Chomsky makes clear, it is proposed on a very high level of abstraction. Chomsky’s nativism is not as novel as many take it to be. It would not seem innovative to anyone familiar with Aristotle’s distinction between first actuality and second actuality; one of Aristotle’s applications of it was to knowledge of grammar. Nelson Goodman wrote about Chomsky’s view: “until the term ‘innate

 Wilhelm von Humboldt, On Language, 215.  See, for example, his Rules and Representations, chapters 3 and 5, and “Language and Nature.”  In Rules and Representations, Chomsky wrote: “What many linguists call ‘universal grammar’ may be regarded as a theory of innate mechanisms, an underlying biological matrix that provides a framework within which the growth of language proceeds.” (187). For more recent examples, see his “Language and Nature,” Mind, 104, 1– 60; Powers and Prospects (London: Pluto, 1996); “Comments: Galen Strawson, Mental Reality”, Philosophy and Phenomenological Research, vol. LVIII, no. 2 (June 1998). Chomsky often uses the phrase “mind/brain,” but when questioned about what the slash in it means, he admits that this is a problem and disavows any inclination to accept reductionism or eliminative materialism. See Noam Chomsky, Language and Thought (Wakefield, Rhode Island & London: Moyer Bell, 1993), 79 – 87; the question was asked by James H. Schwartz, Language and Thought, 71– 77.  Rules and Representations, 244.  Rules and Representations, 165.

3 Scientific opinions

99

idea’ is applied, what is advocated is the rather trivial truth that the mind has certain faculties, tendencies, limitations.”¹⁴⁶ But even if this truth is trivial, its importance was neglected in philosophy and psychology until Kant drew attention to it. Indeed, Chomsky’s appeal to innate psychological structures is a sort of biological “Kantianism,” a biological transcendental idealism.¹⁴⁷ His application of it is not restricted to universal grammar. He applied it to perception: “we know that the visual system of a mammal will interpret visual stimulations in terms of straight lines, angles, motions, and three-dimensional objects” ¹⁴⁸ He applies it also to conceptualization, in ways Kant would have found surprising: “[H]uman nature gives us the concept “climb” for free. That is, the concept “climb” is just part of the way in which we are able to interpret experience available to us before we even have the experience. That is probably true for most concepts that have words for them in language. This is the way we learn language. We simply learn the label that goes with the preexisting concept. So in other words, it is as if the child, prior to any experience, has a long list of concepts like “climb,” and then the child is looking at the world to figure out which sound goes with the concept. We know that the child figures it out with only a very small number of presentations of the sound.”¹⁴⁹ The role of language in cognition was discussed by Chomsky’s student Steven Pinker in his widely read book The Language Instinct. He wrote, “Sometimes it is not easy to find any words that properly convey a thought. When we hear or read, we usually remember the gist, not the exact words, so there has to be such a thing as a gist that is not the same as a bunch of words. And if thoughts depended on words, how could a new word ever be coined? How could a child learn a word to begin with? How could translation from one language to another be possible?¹⁵⁰ This is plain good sense. We may wince, however, when Pinker goes on to claim that psychologists have shown that babies can do arithmetic: “The developmental psychologist Karen Wynn has recently shown that five-month-old babies can do a simple form of mental arithmetic. She used a technique common in infant perception research. Show a baby a bunch of objects long enough, and the baby gets bored and looks away; change the scene, and if the baby notices the difference, he or she will regain interest. The methodology has shown that babies as young as five days

    

Nelson Goodman, Problems and Projects (New York: Bobbs-Merrill: 1972), 74. Cf. Rules and Representations, 251. Language and Problems of Knowledge, 171. Language and Problems of Knowledge, 191. Steven Pinker, The Language Instinct (Harvard University Press, 2007), 57– 58.

100

Chapter Five: The Role of Language in Cognition

[sic] old are sensitive to number. In one experiment, an experimenter bores a baby with an object, then occludes the object with an opaque screen. When the screen is removed, if the same object is present, the babies look for a little while, then get bored again. But if, through invisible subterfuge, two or three objects have ended up there, the surprised babies stare longer.”¹⁵¹ Wynn’s experiments had obvious merits, but Pinker’s interpretation of their results is naïve. The naiveté is not philosophical but scientific. It fits Mark Twain’s description of science as an endeavor in which “one gets such wholesome returns of conjecture out of such trifling investment of fact.” It is an exaggerated abductive inference from meager and ambiguous data, not much better than Benjamin Whorf’s inference from what he knew about Indian languages that thought depends on language, an inference Pinker justly criticizes. To claim, on the basis of experimental results like Wynn’s, that babies have knowledge of arithmetic is like claiming that since birds know how to fly they have knowledge of aerodynamics. Indeed, another scientist, Alan M. Leslie, has written of “[Baillargéon’s] important discovery about the young infant’s understanding of mechanics,” on the basis of an experiment similar to Wynn’s.¹⁵² In the same collection of papers we also find Gabriel Horn arguing that thought does not require language on the grounds that mice “distinguish self from other objects” and thus are aware of their bodies because they (generally) don’t get stuck in holes too small for them to get through.¹⁵³ Such opinions in psychology bring to mind recent reports of discoveries in genetics of a “gambling-gene” and an “alcoholism-gene.” But we should not complain inordinately about them. Despite the exaggerated conclusions drawn, the underlying facts are important in their own right. While crediting babies with cognition of arithmetic at best ignores the nature of arithmetic, there is no doubt that any prelinguistic cognition of what in the scientific literature is called “numerosity” would be relevant to the question about the dependence of cognition on language. More recent experiments, similar to Wynn’s but involving monkeys, suggest that monkeys also are capable of cognition of numerosities. Pinker writes: “Grammar offers a clear refutation of the empiricist doctrine that there is nothing in the mind that was not first in the senses.”¹⁵⁴ This implies that syntactical, and therefore presumably also logical, form corresponds to nothing perceived or otherwise experienced in the world. But Pinker’s claim

 The Language Instinct, 124. Karen Wynn’s piece appeared in Nature 358 (1992), 749 – 750.  Thought Without Language, 194. Baillargéon’s works are cited at the end of Leslie’s.  Gabriel Horn, “Thought without Language in Birds,” in L. Weiskrantz, Thought without Language (Oxford: Clarendon Press, 1988), 299.  The Language Instinct, 77.

4 Language in logical cognition

101

that therefore the “seat” of the language instinct is in the brain casts little light on the status or nature of cognition. To be told that logical cognition is grounded, not in perceived objects in the world, but in a structure of the brain is no more illuminating than to be told that it is grounded in evolution.

4 Language in logical cognition Our brief survey of what philosophers and scientists have said about the general question of the role of language in cognition suggests that it has received at best inconclusive treatment. But there is a much more specific question that can be answered with reasonable confidence. It concerns the role of language in logical cognition. To answer that question requires no empirical research and remains properly a task of philosophy, the home of logic. Wittgenstein held that the logical constants do not “represent,” that there are no “logical objects.” In the Introduction, I suggested that this is true of all logical expressions – variables, quantifiers, sentential operators, the verb “to be” in its senses of identity, existence, and predication, and the symbols or syntactic structures expressing these senses, as well as of declarative sentences themselves – and that we count as logical objects what all logical expressions so understood might be taken to represent. I called the kind of cognition that requires logical expressions for its expression “logical cognition.” It includes but must not be confused with the sort of cognition pursued by logic. The scope of the latter is much narrower. Statements expressing logical cognition usually contain also nonlogical expressions, while those in logic contain only logical expressions. I suggested that we call the former logical statements and the latter statements of logic. For example, “All men are mortal” is a logical statement because it includes the logical expression “all,” but it is not a statement of logic. Statements of logic are purely formal and usually employ only technical symbols. An example would be (($x) Φx ≡ ~ ("x) ~ Φx)), i. e., “Something is Φ if and only if it is not the case that nothing is Φ.” I shall discuss logical expressions, logical statements, and logical cognition in detail later. What is relevant here is that if logical expressions do not “represent,” i. e., if they stand for nothing, then logical cognition would indeed seem impossible without language, language would seem to be “all there is to it” since there would be nothing else that might be pertinent. We would have no “access” to what logical cognition is about except through language because it is not about anything. In this respect logical cognition is dramatically different from, say, perception, where in addition to talking about its objects we can also perceive them. Not even God could know that all men are mortal by “perceiving”

102

Chapter Five: The Role of Language in Cognition

all men. God might perceive all individual things, perhaps an infinity of them, but it is not by perception that he could know that they are all the individual things that there are: this is not something that can be perceived because there is nothing, perceivable or unperceivable, that the word “all” stands for. (I shall leave it to theology to tell us whether or how then God would know that they are all the individual things that there are.) If the logical expressions do not “represent,” then a limited but far reaching linguistic species of conceptual antirealism, the species I called logical antirealism, ought to be accepted. Any advanced cognition requires statements that make essential use of logical expressions. If logical expressions stand for nothing, then those statements have no extralinguistic significance even if they contain nonlogical components that do. Many expressions, e. g., in fiction, also stand for nothing, but this is why they are not taken to serve a cognitive function. The function of logical expressions, however, is unquestionably cognitive. The structure of a sentence depends, at least in part, on the logical expressions the sentence contains. And it is relevant to cognition because it is relevant to the truth-value of the sentence. We may call the role of logical expressions in cognition “transcendental,” in a sense related to Kant’s yet different in that it applies to language, not mental items or faculties. Indeed, the linguistic turn has often been described as a transcendental turn. If limited to logical cognition, it is more measured, discerning, cautious, and therefore more plausible than the linguistic turn exemplified in the statements quoted earlier from Wittgenstein and Heidegger. The limits of my language may mean the limits of my world, but surely there is more to the world than language. Language may be the house of being, but surely it is not also the furniture in that house. Logical antirealism rests on philosophical, a priori, not empirical considerations. That logic is an a priori discipline has seldom been questioned. To be sure, in recent years doubts have been expressed, most notably by Albert Casullo, that any knowledge is a priori.¹⁵⁵ But the fact remains that questions in logic are usually discussed and settled without making empirical appeals and certainly without engaging in empirical research. Despite the extensive debates over the general issue of realism/antirealism, little has been said about the specific issue of logical realism/antirealism, even though all advanced cognition involves cognition of logical structure. Perhaps this neglect has been due to the comforting thought that logic, which is a branch

 Albert Casullo, A Priori Justification (New York: Oxford University Press, 2004). But see also Laurence BonJour, In Defense of Pure Reason. (London: Cambridge University Press, 1998).

4 Language in logical cognition

103

of philosophy so advanced that it is often considered a separate discipline, has already dealt with and settled the issue. This, of course, is an illusion. Indeed, in the Tractatus, Wittgenstein held that the logical constants do not represent, and offered his truth tables to explain them. But we shall see in the next chapter that Russell felt compelled to allow for the existence of irreducibly negative and general facts, which surely would count as logical objects. He showed that the attempts to reduce negative facts to “incompatibility” and general facts to conjunctions or disjunctions of singular facts are unsuccessful.

Chapter Six: Metaphysical Realism and Logical Antirealism 1 The Logic of Realism In Chapter Five I considered the belief that cognition, or at least “thought,” involves language. This belief encouraged and remains characteristic of the version of metaphysical antirealism that came to dominate 20th century philosophy when it took the linguistic turn. Linguistic antirealism is the most plausible version of metaphysical antirealism. Language is indispensable at least in the case of logical cognition, the sort of cognition that involves notions that are characteristically logical, such as negation and generality. Antirealism with respect to logical cognition, which I called logical antirealism, is the most plausible version of linguistic antirealism. Metaphysical antirealism is best understood in the context of the development of its opposite, metaphysical realism. The latter’s guiding principle may be put as follows: acknowledge as being there what must be there if cognition and truth as correspondence are to be possible. This principle governed the construction of the ontological inventories, categorial schemes, of Aristotle, Frege, Russell, and Bergmann. I select these philosophers because they exemplify especially clearly the role of the principle in the development of realist ontology. In the Categories Aristotle used the notions of said of and present in as primitive, and with unsurpassed elegance proposed the following ontological inventory: (1) items said of but not present in something else (what he called “secondary substances,” i. e., substance universals, such as cat), (2) items present in but not said of something else (particular “accidents,” in any of the nine categories of accident, e. g., this cat’s whiteness, its white color), (3) items both present in and said of something else (universal accidents, e. g., the color white), and (4) items neither said of nor present in something else (what he called primary substances, i. e., particular substances, such as this cat). Frege’s inventory included (1) objects (roughly, what we call particulars or individual things), (2) first-level functions (he called them “concepts” but in a nonpsychological sense closer to that of “property” in current philosophy), which take objects as arguments and yield truth or falsity as values, (3) second-level functions, expressed by what we call quantifiers, which also yield truth or falsity as values but take first-level functions, not objects, as arguments,

1 The Logic of Realism

105

(4) thoughts, roughly what we call “propositions,” the vehicles of truth-value, and (5) functions that take thoughts as arguments and yield compound thoughts, what the propositional connectives express.¹⁵⁶ Russell’s inventory, at least at the time of The Philosophy of Logical Atomism, included (1) particulars, (2) universals (properties), (3) atomic facts, (4) negative facts, thus at least one kind of molecular facts, and (5) general facts. His notion of fact was briefly explained in the Introduction. Bergmann proposed a much richer inventory, which is why he is especially relevant here. It included (1) particulars, (2) universals, (3) the “ultimate sorts” of particularity and universality, in virtue of which an item is a particular or a universal, (4) facts (also called complexes), which might be only “potential,” possible but not actual, (5) what the propositional connectives express, (6) what the quantifiers express, (7) the “modes” of actuality and potentiality, which “pervade” facts and render them actual or potential, and (8) the three nexus of exemplification, set-membership, and meaning.¹⁵⁷ The logic of the development of realism that these inventories exemplify now emerges. All four include the category of particulars. Even sense data, with which, a la Berkeley, Russell toyed, are particulars. The privileged status of the category is due presumably to the fact that particulars are the standard examples of what is perceived or imagined. This may be why some opponents of antirealism (perhaps motivated by the title of Nelson Goodman’s Ways of Worldmaking) object to it by saying, “Surely there were a sun and an earth before there were human beings and thus, for all we know, before there were minds, or language and concepts.” But the antirealist claims, not that the sun and the earth were created by our thought or language (Goodman was not a sort of creationist), but that they cannot be perceived, conceived, and described independently of our perception, thought, and language – which is hardly more than, in Hilary Putnam’s phrase, a virtual tautology.¹⁵⁸ Such opponents of antirealism presumably would be willing to say (1) that the sun and the earth are particulars, (2) that they have properties, (3) that the sun exemplifies the property of being a star and the earth the property of being a planet, (4) that it is a fact that there was a sun and an earth before there were human beings, (5) that the sun is not a planet and the earth not a

 See “Concept and Object,” “Function and Object,” “Thought,” and “Negative Thoughts,” in Beaney, The Frege Reader.  See New Foundations of Ontology. Also “An Ontological Inventory,” Journal of the British Society of Phenomenology and Existentialism, 1974, and Logic and Reality (Madison: University of Wisconsin Press, 1960).  Hilary Putnam, “The Dewey Lectures,” Journal of Philosophy, 1994, 513.

106

Chapter Six: Metaphysical Realism and Logical Antirealism

star, (6) that they and other stars and planets were all that there was before there were human beings, (7) that they actually, not just possibly, existed then, (8) that the sun and the earth constitute a pair of which they are the members, and (9) that some things, mental or verbal, mean, refer to, the sun and the earth. If the opponents of antirealism refuse to make such additional assertions, and especially to acknowledge items that might correspond to the italicized words in them, what would be their justification? Might they be merely displaying, however unwittingly, the continuing hold of 17th and 18th century empiricism, in particular its confusion of knowing with perceiving and of thinking with imagining? We are comfortable with particulars because they come to mind first when asked what we perceive or can imagine. But they are hardly all that we can know or think. The opponents of antirealism who confidently assert the existence of the sun and the earth ignore the question whether the sun and the earth would have existed if there were not also items corresponding to the italicized words. At least they owe us a serious discussion of how sparse their ontology could be. That it cannot limit itself to such items as the sun and the earth – i. e., to particulars – became evident in the logical development of realism when it left the terra firma of particulars. The first move beyond particulars was to introduce the category of properties, a move with which Plato dazzled philosophers. It was almost inevitable because we think and speak not only of particulars but also of what they are. The move to Fregean thoughts and Russellian or Bergmannian facts came more than two millennia later but seemed also inevitable. We speak in sentences, not lists of names. If the description of the world requires sentences, what items in the world require this and might be said to correspond to sentences? As we saw in the Introduction with the example of Jack’s admiring Jill but Jill’s not admiring Jack, surely they are not the items that correspond to names and predicates. At least atomic facts must be allowed. But a further move to molecular, even if only negative, and then to general facts also seemed required. Atomic sentences are woefully inadequate for any cognition that is at all advanced. Any language beyond that of babes requires molecular, at least negative, as well as general sentences, whether universal or particular. And so a move to negation and generality themselves, i. e., to what makes negative facts negative and general facts general, seemed needed. This move was made explicitly by Frege and Bergmann, as well as by Russell in Theory of Knowledge ¹⁵⁹ though not in The Philosophy of Logical Atomism. Bergmann even included in his inventory particularity, univer-

 Bertrand Russell, Theory of Knowledge: The 1913 Manuscript (London: Allen & Unwin, 1984), 99.

1 The Logic of Realism

107

sality, actuality, and potentiality, as what allows us to distinguish respectively between particulars and universals and between actual and merely possible facts, as well as exemplification, set-membership, and meaning, which respectively “tie” particulars to their properties, members of sets to their sets, and thoughts to what they are thoughts of. Like Meinong, Bergmann has often been accused by other realists of “going too far,” but those realists do not explain where and how the line should be drawn. The usual half-hearted attempts to rescue realism from “excesses” by adopting some sort of reductionism, for example, of negative facts to incompatibility and of general facts to conjunctions or disjunctions of atomic facts, were shown by Russell in The Philosophy of Logical Atomism to fail because they presuppose what they claim to “reduce.” Aristotle’s, Frege’s, Russell’s, and Bergmann’s ontological inventories are in many ways similar, despite their major differences. All four include particulars and properties, though Frege called particulars “objects” and properties “concepts.” Frege’s, Russell’s, and Bergmann’s include a category of items that correspond to sentences, though Aristotle’s does not. Frege called these items thoughts, Russell and Bergmann called them facts. Fregean thoughts were said by Frege to belong in a “third realm,” which is neither mental nor physical, but this would be true also of Russellian and Bergmannian facts. A fact may include constituents that are mental or physical, but in no clear sense can itself be called mental (admiration may be something mental, but is the fact that Jack admires Jill also mental, in whose mind would it be?) or physical (what are the weight, size, and location of the fact that Jack admires Jill?) This is why facts are sometimes said to be only in logical space, just as Fregean thoughts were said to be in a third realm. Frege’s, Russell’s, and Bergmann’s inventories included negative and general facts, but only Frege’s and Bergmann’s included negation and generality themselves. Except in the abandoned Theory of Knowledge, Russell committed himself only to negative and general facts, not to what makes them negative or general. And only Bergmann followed the logic of realism further by including particularity, universality, exemplification, actuality, potentiality, set-membership, and meaning. Realists were required by the realist principle to acknowledge such additional “items,” though only Bergmann acknowledged all of them. The reason for the requirement was that we can hardly suppose that our thought or language corresponds to the actual, objective world of particulars if we do not suppose that most, if not all, of those additional items are also in it. There would not be a sufficiently specific, definite fit between thought or language and the world without them. For example, the sentence “Jack admires Jill” could not, if true, correspond just to Jack and Jill, or even to Jack, Jim, and the relation of admiring, since the sentence “Jill admires Jack” also would correspond to those things.

108

Chapter Six: Metaphysical Realism and Logical Antirealism

The motivation behind antirealism is now easy to discern. As Russell said about disjunctive facts, it is not plausible that somewhere “in the actual objective world” there are all those things “going about” that the realist principle seems to require.¹⁶⁰ It is implausible because, except for particulars and perhaps properties, the “things” in question are all logical, in the broad sense of “logical” explained in the Introduction. It becomes tempting, therefore, to conclude that in the case of such things there are only the words or syntactic structures supposed to correspond to them. But Wittgenstein made evident in the Tractatus that this conclusion is at least hasty. By distinguishing between saying and showing, he proposed a third alternative, what I called semirealism. I argued in chapter 4 that the distinction is neither idiosyncratic nor obscure. It has an obvious application even to ordinary pictures. I also argued that the related picture theory of meaning and thought conforms to the traditional view that thought involves representations, “ideas,” perhaps even mental images or “pictures.” Wittgenstein’s rejection of realism in logic was unequivocal and explicit: “My fundamental idea is that the ‘logical constants’ are not representatives [nicht vertreten]; that there can be no representatives of the logic of facts” (4.0312). But his rejection of antirealism in logic was also unequivocal and explicit: “…is it really possible that in logic I should have to deal with forms that I can invent? What I have to deal with must be that which makes it possible for me to invent them” (5.555).¹⁶¹ Even though the logical constants represent nothing, there must be something that makes our use of them possible and thus is presupposed by it. Wittgenstein went to some lengths telling us what that “something” is. It included the existence of objects, which are (1) the meaning (reference, Bedeutung) of names, (2) the values of the individual variables in general propositions, including “the laws of logic,” (3) and the constituents of atomic facts (Sachverhalte, “states of affairs”), and thus (4) the “substance of the world.”¹⁶² It includ-

 The Philosophy of Logical Atomism, 71.  “In logic nothing is accidental: if a thing can occur in a state of affairs, the possibility of the state of affairs must be written into the thing itself…” (2.012). “…We have said that some things are arbitrary in the symbols that we use and that some things are not. In logic it is only the latter that express: but that means that logic is not a field in which we express what we wish with the help of signs, but rather one in which the nature of the absolutely necessary signs speaks for itself [zeichnen sich aus]…” (6.124).  “The propositions of logic…presuppose that names have meaning and elementary propositions sense; and that is their connexion with the world….” (6.124). “A name means an object. The object is its meaning….” (3.203).

1 The Logic of Realism

109

ed the senses of atomic propositions,¹⁶³ the existence or nonexistence of the atomic facts such propositions assert,¹⁶⁴ and thus their truth or falsity. ¹⁶⁵ It included the truth-possibilities of atomic propositions, the combinations of which are made explicit in the truth tables for the molecular propositions composed of elementary propositions. It included the totality of the singular propositions that general propositions show but do not assert.¹⁶⁶ It included formal concepts like those of being an object and being a fact. It included therefore also the totality of objects and the totality of facts, i. e., the world itself. All these can be “shown,” but not “said.” The central doctrine in the Tractatus was that what logic is about is inexpressible. This motivates the antirealist reading of the book, for example, Cora Diamond’s¹⁶⁷ and Warren Goldfarb’s,¹⁶⁸since one might reasonably conclude that what cannot be said lacks reality. But the Tractatus also tells us that what cannot be said nonetheless can be shown or, as Wittgenstein sometimes writes, e. g., in 5.5561 and 6.522, that it can show itself (zeigt sich). One might then no less reasonably conclude that what cannot be said but can be shown is part of reality. After all, it is not like Pegasus or phlogiston. It does show itself. This motivates the realist reading, for example David Pears’¹⁶⁹ and P. M. S. Hacker’s.¹⁷⁰ Wittgenstein, however, avoided both antirealism and realism. He did write “There are, indeed, things that cannot be put into words. They make themselves manifest [Es gibt allerdings Unaussprechliches],” but his next sentence was “They are what is mystical [Dies ‘zeigt’ sich, es ist das Mystische]” (6.522). These “things” are not of concern just to “soft” disciplines like ethics and religion. They include logical form and formal properties, the subject matter of logic. We cannot say of an item that it is an object, a complex, fact, function, number, or the totality of objects or of facts. Therefore, we cannot even say of the world that it is a world. But what we cannot say shows itself in what we can say. It is not nothing.

 “A proposition shows [zeigt sich] its sense. A proposition shows [zeigt] how things stand if it is true. And it says that they do so stand…” (4.022).  “The existence and non-existence of states of affairs is reality” (2.06). “The simplest kind of proposition, an elementary proposition, asserts the existence of a state of affairs” (4.21).  “4.25 If an elementary proposition is true, the state of affairs exists: if an elementary proposition is false, the state of affairs does not exist.”  Letters to Russell, Keynes and Moore, 71.  Cora Diamond, The Realistic Spirit: Wittgenstein, Philosophy and the Mind (Cambridge, MA: MIT Press, 1991).  Warren Goldfarb, “Metaphysics and Nonsense,” Journal of Philosophical Research XXII (1997).  David Pears, The False Prison (Oxford: Oxford University Press, 1987).  M.S. Hacker, Insight and Illusion (Oxford: Clarendon, 1972), 20 – 24.

110

Chapter Six: Metaphysical Realism and Logical Antirealism

Surely, the world is “something.” Surely, so is what Wittgenstein called “the higher,” in ethics or in logic. Yet, neither is it straightforwardly a part of reality. If it were, it would not be inexpressible. Cora Diamond calls such a reading of the Tractatus a “chickening-out,”¹⁷¹ and Warren Goldfarb calls it “irresolute.”¹⁷² But I doubt they would say that logical form is like Pegasus or phlogiston. Wittgenstein is explicit that in some sense it is in “reality” [Wirklichkeit]: “Propositions can represent the whole of reality, but they cannot represent what they must have in common with reality in order to be able to represent it–logical form” (4.12), “Propositions show the logical form of reality. They display it” (4.121). Logical form can only be shown, not because it is nothing but because its presence in both language and reality is what makes any saying possible. Saying presupposes showing. Something is said only if something else is shown. But while logical form is not like Pegasus or phlogiston, neither is it like Secretariat or oxygen. Although Wittgenstein was not an antirealist, he also was not a realist. This should not surprise us. In both his early and his later works he avoided simplistic distinctions: “And yet you again and again reach the conclusion that the sensation itself is nothing. – Not at all. It is not a something, but not a nothing either!”¹⁷³ Wittgenstein’s distinction between saying and showing amounted to a rejection of standard realism, but it did not amount to acceptance of standard antirealism. In this respect, he was not unlike his distant precursor Kant, who described his position as transcendental idealism but also empirical realism.¹⁷⁴ Had Wittgenstein provided an explicit inventory of what according to him is in the world, which he did not because the inventory would have employed formal concepts, it would have been sparse. At most, it seems, it would have contained two categories: objects and facts. Moreover, it is not clear whether the category of objects included both particulars and properties. Indeed, it is not clear that Wittgenstein allowed this distinction. Nor is it clear that he allowed for molecular and general facts, rather than just for atomic facts, though he did seem to allow for negative facts: “We call the existence of states of affairs a positive fact, and their non-existence a negative fact” (2.06).

 The Realistic Spirit, e. g., 181, 194 ff..  Warren Goldfarb, “Metaphysics and Nonsense,” Journal of Philosophical Research XXII (1997), 66. See also, in the same issue, Cora Diamond, “Realism and Resolution: Reply to Warren Goldfarb and Sabina Lovibond,” 75 – 86.  Philosophical Investigations, 304.  Critique of Pure Reason, A 369 – 370.

2 Antirealism: ontological, cosmological, and logical

111

Frege also allowed for what cannot be said, though not quite in Wittgenstein’s sense.¹⁷⁵ One of his ontological categories, as have seen, was that of concepts. But we cannot speak about concepts: “The concept horse is not a concept,” Frege wrote. What expresses a Fregean concept is a grammatical predicate, and “the concept horse” is not a predicate. The relevant predicate is “is a horse,” but it cannot serve as the grammatical subject of a well-formed sentence, including a sentence of the form “x is a concept.” Yet, of course, there is such a concept. Some objects are horses. Frege did not deny that there are concepts. He was an unmitigated realist. But neither did he explain their peculiar status. This is why his readers have found his assertion that the concept horse is not a concept confusing.

2 Antirealism: ontological, cosmological, and logical According to metaphysical realism, the existence and nature of things (res), reality, the world, is independent of our cognition of them, in the broad contemporary sense of “cognition” that includes sense perception, introspection, intellectual intuition, imagination, memory, recognition, conceptualization, inductive and deductive reasoning, the use of language and other symbolism. Metaphysical antirealism denies this. I call both “metaphysical” in order to distinguish them from realism and antirealism in ethics, aesthetics, the philosophy of science, the philosophy of mathematics, and other branches of philosophy, where realism and antirealism are restricted to a specific subject matter. The thesis of metaphysical antirealism is that the world insofar as it is knowable by us depends on our capacities and ways of knowing, our cognitive faculties. The inclusion of “insofar as it is knowable” does not produce the sort of equivocation present, for example, in the sentence “As king of England, John was subject to no one,” which is true but compatible with “As duke of Aquitaine, John was subject to the king of France.” We know both that John was king of England and that he was duke of Aquitaine. But we know the world only insofar as it is knowable. And this is a tautology, unlike “As king of England, John was subject to no one.” The standard argument for the thesis of metaphysical antirealism, in both the Kantian and its more recent versions, was sketched in the Introduction as fol-

 Cf. P.T. Geach, “Saying and Showing in Frege and Wittgenstein,” in Essays on Wittgenstein in Honor of G.H. von Wright, ed. Jakko Hintikka, Acta Philosophica Fennica 28 (Amsterdam: North Holland, 1976).

112

Chapter Six: Metaphysical Realism and Logical Antirealism

lows: (1) We cognize only what we have the capacity to cognize. This is a tautology. Therefore, (2) there is no reality, no world, that is independent of our cognitive capacities. But (2) does not follow from (1). What follows is another tautology, (3) that we cannot cognize reality independently of our cognitive capacities. Contemporary antirealists argue on the basis of (1) for (2), not for (3), probably because the negation of (2), Kant’s view (4) that there is a reality, “things-in-themselves,” which is independent of our cognitive capacities, seems to them idle. But there is a very good reason for (4), namely, that (2) implies a sort of cosmic humanism, human creationism, the proposition that the world depends on certain members of one of its planets’ fauna. Metaphysical antirealism comes in many varieties, as different as Berkeley’s subjective idealism, Kant’s transcendental idealism, Hegel’s absolute idealism, Wittgenstein’s logical antirealism, Heidegger’s phenomenology of being-in-theworld, as well as, in more recent philosophy, Michael Dummett’s antirealism and Nelson Goodman’s “irrealism.” Berkeley held that the existence of the things we perceive is dependent on our perception of them, Kant held that their nature is dependent also on our understanding, and Wittgenstein, Heidegger, and Goodman stressed their dependence on our language. In one form or another, metaphysical antirealism has dominated modern philosophy at least since Berkeley, even though it is metaphysical realism that remains the bedrock of everyday and scientific thinking. But, as we have seen, numerous though seldom made distinctions within it are needed. And the reasons for accepting it have seldom been stated in detail. Usually they have consisted in abstract, obscure generalities such as “Nothing can be conceived that cannot be perceived” and “For knowledge to be possible, objects must conform to knowledge,” which are no less controversial than what they are reasons for. In the Introduction I distinguished antirealism from skepticism and relativism. I also distinguished the several varieties of antirealism: ontological, cosmological, perceptual, conceptual, linguistic, and logical. One can be an antirealist but not a skeptic, and one can be a skeptic but not an antirealist. One can be an antirealist but not a relativist. One can be a cosmological but not an ontological antirealist. One can question the reality of logical objects such as facts, and thus of the world conceived as the totality of facts, but not the reality of things, e.g., animals, vegetables, and minerals. And one may be neither a realist nor an antirealist regarding some items, e.g., those that subsist (Russell) or those that can only be shown (Wittgenstein), hence the need to acknowledge what I called semirealism. Within both cosmological and ontological antirealism, we need also the distinction between perceptual and conceptual antirealism. Since a concept may be understood as either a purely mental, psychological, and therefore subjective item, or as a linguistic, public, and therefore objective one, we must distinguish

2 Antirealism: ontological, cosmological, and logical

113

between Kant’s psychological version and the 20th century linguistic versions of conceptual antirealism. Further distinctions are needed within the variety of conceptual antirealism that I called logical. As we shall see, one may accept a realist interpretation of singular statements, a semirealist interpretation of compound and universal statements, and an antirealist interpretation of generic statements. The distinction between ontological and cosmological realism/antirealism is of special importance. If the world is the totality of facts, not of things, then one who rejects the category of fact also rejects cosmological realism, realism with respect to the world, but not necessarily ontological realism, realism with respect to things. Even if the world is not the totality of facts, it is certainly not just the totality of things, a mere collection of unrelated items, a whole without structure. It must still have a logical structure, and thus its reality is subject to question by the logical antirealist. These distinctions may seem fussbudgety, academic fretwork, but they are needed because each variety of metaphysical antirealism calls for different considerations. The realism/antirealism debate for too long has appeared to allow only two alternatives: all and nothing. It has been needlessly vague because those crucial distinctions have not been made. One aim of this book is to show that the issue is far more complex than participants in the debate usually suppose. I argued in the Introduction that a clear and plausible defense of antirealism should shun purely abstract arguments and appeal to specific and readily understandable truths. One such truth is that logical expressions, i. e., the expressions distinctive of logic, play an essential role in all developed talk and thought but almost certainly stand for nothing in the world. It is the truth on which logical antirealism rests. Standard examples of such expressions are the sentential operators: “not” (“~”), “and” (“•”), “or” (“v”), “if…then…” (“⊃”); the quantifiers: “all” (“"”), “some” (“$”); and the verb “to be” in its senses of identity (“=”), existence (“there is,” “exists”), and predication (“is,” “are”), even when the latter is expressed only by syntactic order. “Is” is a logical expression in both “Socrates is human” and “God is,” though standard logic would express the former not with a separate sign but by the juxtaposition of the subject and predicate terms (e. g., “Hs,” “H” standing for “is human” and “s” for “Socrates”), and would either ignore the latter as ill-formed or translate statements in which it occurs by employing the particular quantifier “some.” No abstract argument or obscure philosophical principle is needed to convince nonphilosophers as well as most philosophers that, though essential to any cognition above that of infants, such expressions do not stand for any items, that in the world there is no such object, for example, as not, all, is, or and. Frege used the phrase “logical objects” for the objects of arithmetic in the context of his project of reducing arithmetic to logic, a project continued later

114

Chapter Six: Metaphysical Realism and Logical Antirealism

by Russell and Whitehead.¹⁷⁶ Wittgenstein used it for the entities a logical realist may think are required for understanding the sentential connectives. In the Introduction I applied the phrase also to facts, in Russell’s sense. Clearly, it is applicable to “thoughts” in Frege’s sense. Both Russellian facts and Fregean thoughts are categories of entities accepted because of broadly logical considerations. Both were explicitly introduced as the category of entities that declarative sentences stand for or express. I have accordingly also counted declarative sentences as logical expressions, facts being the logical “objects” to which, if true, they supposedly correspond. One of the central concerns of logic has been the analysis of sentences with respect to their logical form. A basic presupposition of logical analysis is that sentences that are unlike in surface grammar may share the same logical form. In analysis we begin with the surface grammar of the sentence but search for what it must have in common with other sentences, especially those in other languages, if they are to have the same truth-value and the same implications. That common feature is the logical form of the sentence. Its representation would require logical expressions. The logical realist holds that at least some logical expressions correspond to entities. As we saw in the previous section, Frege and Russell (at one central stage of his philosophy), were clearly logical realists. So seemed to be Gustav Bergmann, though in the next chapter we will find that this was not at all clear. Wittgenstein, in Tractatus Logico-Philosophicus and his later works, appeared to be a logical antirealist, but we have seen that his position would be better described as semirealism. Logical antirealism was characteristic, even if only tacitly, of traditional philosophy. For example, Kant wrote that logic abstracts “from all objects of knowledge,” that it has “itself alone and its form” to deal with, and that the concern of logic is to give “an exhaustive exposition and a strict proof of the formal rules of all thought.”¹⁷⁷ In effect, he acknowledged the chief thesis of logical antirealism: there are no logical objects even though logic is present in all thought. All thought, even when it does not have objects, must conform to logic, but logic has no objects. This could have been an argument for transcendental idealism additional to those Kant explicitly offered, but he did not present it as such. Logical antirealism leads directly to cosmological antirealism if “fact” is a logical expression and the world is the totality of facts. Perhaps animals, vegetables, and minerals also have logical structure, but this is not evident as it is

 Gottlob Frege, The Basic Laws of Arithmetic (Los Angeles & Berkeley: University of California Press, 1987), ii, § 147.  Critique of Pure Reason, B ix.

2 Antirealism: ontological, cosmological, and logical

115

in the case of a fact. Thought and talk about individual things do usually employ logical concepts and expressions, but only phenomenalist theories, which analyze statements about material objects as statements about actual and possible experiences, explicitly attribute logical structure to individual things. Logical antirealism does not entail ontological antirealism because individual things have no logical structure, they are not facts and thus involve neither propositional connectives nor quantifiers in the way facts, and the sentences expressing them, do. They can be subjected to chemical analysis but not to logical analysis. The belief that they could be was generated by Wittgenstein’s claim that all complexity is propositional and had been encouraged by Mill’s view that a physical object is nothing but the ‘permanent possibility of experience,’ i. e., nothing but its actual and possible appearances, what one would sense in all relevant circumstances. Much later, Price showed that these would be unlimited in number, and consequently statements about the object could not be translated into a conjunction of statements about its appearances.¹⁷⁸ An antirealism that is only logical and also only cosmological would be so moderate as to be virtually part of common sense, the thoughtful nonphilosophical judgment that all theorizing must respect even if not accept. Common sense would rebel against the postulation of objects corresponding to the sentential connectives. It is satisfied with counting them as syncategorematic, mere ancillaries in assertions, incapable of use as subjects or as predicates. Common sense could be easily convinced that sentences, the expressions Russell took to stand for facts, are also “syncategorematic.” They, too, function as neither names nor predicates. Facts were not on Aristotle’s or any other list of categories before Russell and Wittgenstein included them in theirs. It would occur only to philosophers that the sentential connectives and sentences themselves might stand for anything in the world. The position of semirealism, of course, would be too technical for common sense to judge. Only a few philosophers, most notably Russell, have thought that there are “general facts,” which correspond to universal sentences, as well as “molecular facts,” which correspond to molecular, especially negative, sentences. Frege, who together with Russell was the target of Wittgenstein’s attack on logical realism,¹⁷⁹ abhorred Russell’s category of fact (on the grounds that to say that p is a fact is to say no more than that it is true that p) but he did appeal to the no less questionable category of objective entities he called “thoughts,” which are expressed by senten Price, H. H., Perception, (London: Methuen, 2nd ed., 1950).  “[T]he words ‘complex’, ‘fact’, ‘function’, ‘number’ … all signify formal concepts, and are represented in conceptual notation by variables, not by functions or classes (as Frege and Russell believed)” (Tractatus 4.1272).

116

Chapter Six: Metaphysical Realism and Logical Antirealism

ces. Frege also accepted the existence of general and negative thoughts, as well as of “compound thoughts” such as conjunctions, disjunctions, and conditionals. His vigorous arguments, as well as Russell’s, for the irreducibility of general and negative statements were, in effect, arguments for logical realism. To avoid confusion, in this chapter I shall often follow Frege’s, Russell’s, and Wittgenstein’s practice of using “general” instead of “universal,” even though particular (“existential”) statements and generic statements (to be discussed in chapter 8) are also general. None of them thought that true generic statements stand for generic facts. They did not even consider them.

3 The logical structure of the world The subject matter of metaphysics is said to be being, reality, what exists, or just “the world” as in Goodman’s Ways of Worldmaking. Its first question is, what is real, what kind of entities exist? Various answers have been given, e. g., that only material entities exist (Hobbes), that only mental entities exist (Berkeley), that in addition to material and mental entities also abstract entities (Plato) or God (Aquinas) exist. But however we answer the question, we take for granted that the world has a structure, that it is not a mere collection of isolated items. What kind of structure is fundamental, absolutely necessary, to the world? Not a causal structure: Hume denied that there are causal connections, except in the bland sense of spatiotemporal correlations. Not a physical or even just spatial structure: the dualist holds that in addition to material entities there are irreducibly nonphysical mental entities such as thoughts and feelings, and the idealist even holds that everything is mental. Both deny that mental entities are in space: they lack geometrical characteristics and do not enter in relations such as two-miles-from. Nor is even a temporal structure absolutely necessary. The Platonist holds that there are abstract entities, e. g., numbers, which are not in time – they do not enter in relations such as two-years-earlier-than. The theist holds that the spatiotemporal world and thus presumably space and time themselves were created by God. The fundamental structure of the world, which though seldom mentioned or even considered is denied by no one, is logical. That a world must have a logical structure is a requirement even more basic than that a world must be “logically possible,” if the latter means that it must not “involve contradiction,” for even a contradiction has logical structure, often as simple as that of “p and not-p.” Astronomy, physics, chemistry, and biology focus on the causal, spatial, temporal, and physical structure of the world. Philosophy, at least as it was understood by Frege, Russell, and Wittgenstein – the founders of contemporary logic and ana-

3 The logical structure of the world

117

lytic philosophy – focuses on its logical structure. If the first question of metaphysics is what is real or exists, its second question is what kind of structure it must have. And surely the answer is that it must have a logical structure. Aristotle held that the “science of being qua being,” i. e., metaphysics, begins with the study of the principles of the “syllogism,” i. e., logic.¹⁸⁰ What is meant by “logical structure”? The answer lies in what is meant by “logic.” Logic is concerned with the relations between sentences (propositions, statements, judgments) that hold in virtue of their logical form. And the logical form of a sentence is best understood as what is left when we completely abstract from its subject matter, what it is about. According to Russell’s canonical account, in respect to logical form sentences (propositions) are either atomic, compound, or general.¹⁸¹ Essential to discerning the logical form of a sentence are the logical expressions it contains. To say that the world must have a logical structure entails that any adequate description of it must employ such expressions. And we have no conception of a world that is not, at least in principle, describable. Nothing would count as a world if it did not allow for atomic statements, e. g., “This page is white.” Nothing would count as a world if it did not allow for compound statements, e. g., “This page is not blue” and “If this page is white then it is not blue.” And nothing would count as a world if it did not allow for general statements, e. g., “All winters in Minnesota are severe” and “There is water on Mars.” It is natural therefore for the logical realist to see the logical structure of the world as providing room for atomic facts about the properties and relations of individual things, for compound facts, most notably, facts about what is not the case (negative facts), what is the case only if something else is the case (conditional facts), for two facts being both the case (conjunctive facts), for either one or the other of two facts being the case (disjunctive facts), and for general, either universal or particular, facts. The logical realist may go further. Since nothing would be a world if it did not allow for saying of a thing what it is, that is, for predication, the logical realist may hold that there is in the world a relation or nexus of exemplification or instantiation. Since nothing would be a world if it did not allow for distinguishing one thing from another, the logical realist may hold that there is negation in the world. And since nothing would be a world if it did not allow for generalization, the logical realist may hold that there is generality in the world. The logical antirealist disagrees, insisting that, insofar as it is knowable, the world has a logical structure only because we describe it with a language that contains logical expressions.

 Aristotle, Metaphysics, 1005 b 7– 35.  Principia Mathematica, Introduction.

118

Chapter Six: Metaphysical Realism and Logical Antirealism

Logical expressions are parts of sentences. Sentences are the vehicles for the description of the world. I have suggested that we may consider sentences themselves as logical expressions, and the facts supposed to make them true as logical objects. The logical antirealist will argue, however, that a fact is indistinguishable from the sentence it supposedly makes true. A fact is a complex entity with a structure or form that seems to image the structure or form of a sentence. For example, Russell’s requirement that an atomic fact “contain” particulars and properties or relations corresponds to the grammatical requirement that a simple sentence contain a subject and a predicate, and his distinction between atomic and molecular facts corresponds to the grammatical distinction between simple and compound sentences. Indeed, the former correspondence was explicitly acknowledged in Wittgenstein’s doctrine that atomic sentences are logical pictures of atomic facts. It is fairly clear that when describing the ontological characteristics of facts and their logical structure or form Russell relied on the grammatical characteristics of sentences. Of course, he was concerned not with the “surface” grammatical form of a sentence but with its logical form. But one arrives at the latter only by beginning with the former. Logic is concerned with what sometimes is called logical grammar, not with ordinary grammar, but we have little if any conception of logical grammar apart from our conception of ordinary grammar. The central role of sentences was explicitly acknowledged in Frege’s explanation of his technical notion of thought (Gedanke) as the “sense” (Sinn) of a sentence. Even though he held that thoughts belong neither in the physical nor in the mental world, but rather “in a third realm,”¹⁸² Frege wrote, “The world of thoughts has a model in the world of sentences, expressions, words, signs. To the structure of the thought there corresponds the compounding of words into a sentence.”¹⁸³ Russell’s and Wittgenstein’s heavy indebtedness to Frege in logical theory was freely acknowledged by both. Like Fregean thoughts, Russell’s and Wittgenstein’s facts, at least those that are atomic, also were understood as analogous to, even “pictured” by, sentences. Although the German word Gedanke is a synonym of the English “thought,” as used by Frege it has no ordinary translation in English. “Proposition” might be least misleading, as long as we think of a proposition as an objective item distinct both from the sentences in the various languages that express it and from the ideas and judgments we entertain about it. If so, there would be no clear difference between a true Fregean thought and a Russellian fact. This is why Frege

 Beaney, The Frege Reader, 337.  The Frege Reader, 351.

3 The logical structure of the world

119

wrote, “What is a fact? A fact is a thought that is true.”¹⁸⁴ His view that thoughts are neither mental nor physical may seem mysterious, but as we saw earlier so might Russell’s and Wittgenstein’s facts. Neither propositions nor facts are “things” that might enter in spatial relations, or mental states that might enter in temporal relations. Neither the proposition nor the fact that this table is two feet from me at any other distance from me, and neither is a mental image or state that is earlier or later than some other mental image or state. In logic we speak usually of propositions, sometimes of sentences, not of facts. This is a symptom of our ambiguous conception of its subject matter, our uneasy attempt to straddle the perceived chasm between the logic of sentences, which is about language – a certain behavior of some members of our planet’s fauna and thus essentially a zoological subject matter – and the logic of facts, which supposedly is about the world and thus, as Plato might have said, a more appropriate subject matter for “spectators of all time and all existence.” According to logical realism, the chasm is real. According to logical antirealism, it is an illusion. The logic of the world, the antirealist holds, is not distinguishable from the logic of words, and Frege’s thoughts, as well as Russell’s or Wittgenstein’s facts, are merely hypostatized sentences, shadows that sentences cast upon the world. I shall have much more to say about the topic of facts in chapter 9. The general thesis of metaphysical antirealism now appears more plausible. What makes a world a world, rather than a congeries of things, is what requires sentences, not mere lists of names, for our description of it, namely, a logical structure. But the only conception we have of logical structure is that of the logical structure of sentences. This is why in speaking of a world we must appeal to the category of facts. This is why Wittgenstein unhesitatingly wrote that the world is the totality of facts, not of things. Sentences, of course, are parts of language, and their logical structure is a feature of language. But language is something human, “ours.” We have no genuine conception of a language that is both nonhuman and in principle untranslatable into a human language. Therefore, insofar as we can conceive of the logical structure of the world, and thus of the world as a world, they depend on us, they are ours, human. Of course, that the only conception of logical structure we have is that of sentences does not entail that the world does not have logical structure independently of language. But, surely, it is a good reason for reaching that conclusion. For it does entail that our cognition of the world, insofar as it involves logical concepts and employs logical expressions, depends on language.

 The Frege Reader, 34.

120

Chapter Six: Metaphysical Realism and Logical Antirealism

Many twentieth century philosophers have held that all cognition, not just that involving logical concepts, is dependent on language. Earlier I cited Wittgenstein’s assertion that “The limits of my language mean the limits of my world,” Heidegger’s that “language is the house of being” and that “language is the happening in which for man beings first disclose themselves to him each time as beings,” and Quine’s that “thought, as John B. Watson claimed, is primarily incipient speech.” These opinions exemplify the linguistic turn philosophy took in the twentieth century. It was more fundamental than its earlier “turns,” such as the Platonic turn to abstract entities, the theological turn in the Middle Ages, and the idealist turn in the 18th century. For it applies to everything we think is real, including abstract entities, God, and minds. It led to linguistic antirealism. But the linguistic antirealism I have sketched is not as extreme as that implied by the quotations from Wittgenstein, Heidegger, and Quine. It does not hold indiscriminately, as a matter of general principle, that the world as we cognize it, as it is “for us,” depends on language. It restricts the dependence to logical expressions. It avoids the highhandedness, often characteristic of philosophy, which the quotations seem to exemplify. It does not hold that all cognition depends on language. It is only a logical antirealism, even though its implications are far-reaching. It is thus a limited, moderate, and for this reason more plausible antirealism. It does not offer a general and a priori answer to the question whether cognition depends on language. This question, as we saw in Chapter Five, is mainly empirical, to be answered by neuroscientists, psychologists, and linguists. And the proper answer may well vary according to what kind of objects of cognition we have in mind. Rocks are different from headaches, both are different from electrons, and all three are very different from numbers. It would be rash to suppose that what is true of our cognition of some is true of our cognition of all. Rock-climbing is not introspection, and neither is physics. There are also many, very different kinds of cognition. Some may depend on language while others do not. Surely, there is cognition in the form of sense perception, enjoyed by infants and nonhuman animals, which does not involve language. Recognition is a higher form of cognition, but it too is found in infants and nonhuman animals. And driving a car involves specialized cognition that only superficially, if at all, finds or needs expression in language. Creative work in music or painting only minimally involves talking. Worldly people, especially in the law, politics, and diplomacy, rely heavily on “reading” facial expressions, which language notoriously fails to capture. We achieve greater specificity and a sharper focus when we ask, Does logical cognition have a distinctive subject matter, is it, at least in part, about anything distinctively logical? If not, then to that extent logical cognition would appear to be nothing but use of language, since it would lack the feature essential to other

3 The logical structure of the world

121

kinds of cognition: a subject matter, things they are about, which make them, at least in part, the cognitions they are. Seeing a cat differs from seeing a dog partly because cats differ from dogs. Seeing a cat differs from hearing a cat partly because colors and shapes differ from sounds. It would not follow, however, that all there is to logical cognition is language. I used the phrase “distinctive subject matter” precisely because, outside logic, sentences containing logical expressions do have a subject matter, namely, what is denoted by the nonlogical expressions, the “descriptive words,” that they also contain. But our question is whether such sentences have a subject matter insofar as they are logical. The logical structure of the world may be supplied by language, but the world does not have only a logical structure: it also contains the entities that are structured. The tendency to forget this may explain why philosophers seldom see that logical antirealism, which most of them accept, at least tacitly, is a version of metaphysical antirealism. They say, for example, “Of course ‘All humans are mortal’ is about the world – it is about humans.” They overlook the fact that while what the sentence says depends on more than the logical expressions “all” and “are,” it does depend on them. Its truth-value depends on both the nonlogical and the logical expressions in it. In contrast, the statements of logic employ only logical expressions. If logical expressions stand for nothing, then the statements of logic have no extralinguistic significance – a consequence commonly, though rashly, accepted. But all cognition beyond that of babes, not just the cognition expressible in logicians’ statements, must be expressible with statements employing expressions such as the verb “to be” (especially in its sense of predication), the adverb “not,” sentential connectives like “or,” and the quantifiers “all” and “some,” even if, unlike the statements of logic, they employ also nonlogical, descriptive, expressions. All developed human cognition is logical, in this broad sense. All statements, not just the statements of logic, employ logical expressions. A language that lacked them would be either in principle untranslatable or as primitive as baby “talk.” Therefore, no developed human cognition is possible without language. This does not entail but is a good reason for holding that the world, insofar as it is the object at least of developed cognition, cannot have a character fundamentally different from what we humans take it to have. What I mean by developed cognition need not require higher education – it would include the cognition expressed by “All sheep eat grass.” If a shepherd did not know that all sheep eat grass, there would be precious little the shepherd knew. If we could not say that all sheep eat grass, there would be precious little we could say. In chapter 8 we shall see that we are more likely to make the generic statement “Sheep eat grass,” and then the antirealist implications of our reliance on generality would be even more striking because obviously there are no facts that are just generic. For now, however, I shall remain within the fa-

122

Chapter Six: Metaphysical Realism and Logical Antirealism

miliar context of universal statements. In that context, while the words “sheep,” “eat,” and “grass” do stand for things in the world, “all” clearly does not. There is no such entity in the world as all. It would not be an individual thing, nor a property or relation of an individual thing. This is why the phrase “entity in the world as all,” which I just used, is a grammatical outrage. It follows that there are also no distinctive, universal, facts to which statements making essential use of “all” might correspond. The sentence “All sheep eat grass” stands for nothing distinctive in the world. If we take advantage of the familiarity in philosophy of Kant’s term “transcendental,” we may say that logical concepts and expressions are transcendental. They are essential at least to any developed cognition of the world, but stand for nothing in the world. This would be a way of stating the thesis of logical antirealism. It would hardly be news. Philosophers and grammarians have commonly called the logical expressions syncategorematic. But they have offered little explanation or argument, and few have seen the metaphysical implications.

4 Frege and Russell on Negation and Generality I shall take the words “not” and “all” as paradigms of logical expressions. They are fundamental not only to logic but all developed cognition. They express, respectively, the logical concepts of negation and generality (more accurately, universality). In this chapter I shall offer a preliminary discussion of both, mostly in relation to Frege’s and Russell’s classic defense of logical realism regarding them. Wittgenstein’s logical antirealism/semirealism, his “fundamental thought” that the logical constants do not represent, was mainly intended as a rejection of their views. Chapter 7 will be devoted to a fuller discussion of generality, but in relation to Wittgenstein’s and Bergmann’s classic defense of logical antirealism, or at least of logical semirealism. Both the thesis of logical realism and the thesis of logical antirealism would be sufficiently established if made plausible with respect to negation and generality. Regarding generality, Frege wrote, “It is surely clear that when anyone uses the sentence ‘all men are mortal’ he does not want to assert something about some Chief Akpanya, of whom perhaps he has never heard.”¹⁸⁵ The sentence is not equivalent to the conjunction of its singular substitution instances, one of which might be “Chief Akpanya is mortal.” Frege classified what “all” express-

 Peter Geach and Max Black, Translations from the Philosophical Works of Gottlob Frege (Oxford: Blackwell, 1970) 83.

4 Frege and Russell on Negation and Generality

123

es as an “unsaturated” entity that is a second-level function, “saturated” by firstlevel functions like those expressed by “x is a man” and “x is mortal,” which in turn are saturated by individual objects such as Chief Akpanya.¹⁸⁶ A function is unsaturated, or incomplete, in that it requires “attachment” to something else, much as the grammatically incomplete expression “is human” requires attachment to a grammatically complete expression such as “Chief Akpanya.” For example, in the thesis of materialism, “("x)(x is material),” i. e., “Everything is material,” the universal quantifier “"x” represents a second-level function, and “x is material” represents the first-level function that completes it. The latter function is also incomplete, but what completes it is an individual thing, say, this page. Frege’s terminology is awkward, but what matters for our purposes here is his explicit acknowledgement that there is an entity to which the word “all” corresponds. To be sure, it is an “incomplete” entity, but so is also the property of being human. Frege called a first-level function a concept (Begriff), but like many of his contemporaries he did not mean by “concept” anything mental. He meant what “property” means in current philosophy. Frege also held that there are negative thoughts, in his technical sense of “thought” I explained earlier, since “for every thought there is a contradictory thought” which “appears as made up of that thought and negation,” though not as “mutually independent” parts. “The thought does not, by its make-up, stand in any need of completion; it is self-sufficient. Negation on the other hand needs to be completed by a thought. The two components are…quite different in kind….One completes, the other is completed. And it is by this completion that the whole is kept together.”¹⁸⁷ Negation is incomplete, just as the secondlevel function expressed by “all” is incomplete. The difference is that negation is completed by a complete entity, viz. a thought, while a second-level function is completed by an incomplete entity, a first-level function. But negation is not something mental or subjective. It is not “the act of denial.”¹⁸⁸ Nor is it a kind of judging: “Negation does not belong to the act of judging, but is a constituent of a thought.”¹⁸⁹ It is an objective part of a no less objective entity, viz. a negative thought. The sentence “It is not the case that this page is blue” consists of two parts: “it is not the case that” and “this page is blue.” The later could stand by itself, the former could not. But what the whole sentence says is no more psychological or subjective than what “This page is blue” says.

   

“Function and Object” and “Concept and Object,” both in Beaney. Beaney, 358. Beaney, 358. Beaney., 363.

124

Chapter Six: Metaphysical Realism and Logical Antirealism

Russell too was a realist about generality and negation, though he argued for the existence of general and negative facts, not Fregean thoughts. Regarding generality, he wrote, “When you have taken all the particular men that there are, and found each one of them severally to be mortal, it is definitely a new fact that all men are mortal.” For, “In order to arrive [by “complete induction”] at the general statement ‘All men are mortal’, you must already have the general statement ‘All men are among those I have enumerated.’” True general statements, such as “All men are mortal,” stand for “general facts.” Therefore, “there are general facts.” Moreover, because “You cannot ever arrive at a general fact by inference from particular facts, however numerous,” “there must be primitive knowledge” of some general facts.¹⁹⁰ There is therefore “the necessity of admitting general facts.”¹⁹¹ Russell also argued that “there are negative facts” and that “negativeness is an ultimate.”¹⁹² He wrote, “Usually it is said that, when we deny something, we are really asserting something else which is incompatible with what we deny. If we say ‘roses are not blue,’ we mean ‘roses are white or red or yellow.’ But such a view will not bear a moment’s scrutiny… The only reason we can deny ‘the table is square’ by ‘the table is round’ is that what is round is not square. And this has to be a fact, though just as negative as the fact that this table is not square.”¹⁹³ In discussions of our topic one often finds, in addition to logical realism and logical antirealism, the supposedly third option of logical reductionism. For example, in the case of universal statements, the realist holds that the universal quantifier “"x,” or the word “all,” stands for a real entity, whether a Fregean second-level function or some other kind of logical object. The antirealist denies this. The reductionist, also finding the reality of such an entity implausible, “analyzes” universal statements as the conjunctions, and particular statements as the disjunctions, of the singular statements that instantiate them, claiming that this is what they “really” say. Of course, the word “and,” or the symbol “•,” which is essential to a conjunction, and the word “or,” or the symbol “v,” are also logical expressions, and the reality of any logical objects they might stand for is no less dubious. In any case, universal and particular statements are not equivalent to such conjunction and disjunctions. Gustav Bergmann took this to be obvious: “What can be said with the quantifiers cannot be said

 The Philosophy of Logical Atomism, 101– 103. Also in R. C. Marsh, ed., Logic and Knowledge (London: Allen & Unwin, 1956), 235 – 236.  Marsh, 289.  Marsh, 215 – 216.  Marsh, 288. Italics in original.

4 Frege and Russell on Negation and Generality

125

without them….Consider (1) ‘(x)G(x)’ and (2) ‘G(a1). G(a2)… G(aN).’ (1) implies (2). (2) does not imply (1).”¹⁹⁴ \ Sometimes the reductionist appeals implicitly to the very same concepts that are to be explained, and thus is no longer properly called “reductionist.” This is case with the claim that the sentential connectives are “defined” by the corresponding truth-tables and that this is “all there is” to what they mean. For example, not-p is said to be “merely a truth function” of p, on the grounds that not-p is true if and only if p is false. But we have no grasp of falsity except as the negation of truth, regardless of what theory of truth we hold, and thus the reductionist is implicitly appealing to negation when offering that explanation of negation. As Russell remarked, it is “extremely difficult to say what exactly happens when you make a positive assertion that is false, unless you are going to admit negative facts.”¹⁹⁵ In his earlier article “On Denoting” Russell had claimed that a universal statement such as (x) ϕx means that ϕx is always true, and that ($x)ϕx means that ϕx is sometimes true. He pointed out repeatedly that “always true” and “sometimes true” express primitive, indefinable notions. But, if so, we could say the same about “all” and “some,” and avoid puzzling attempts to say what they “mean.” Reductionism is often little more than a front for antirealism (for example, the “reduction” of mental states to brain states is little more than the denial of the reality of mental states). But, as we just saw, in the case of logic it seems to merely replace some logical statements with other logical statements, e. g., universal statements with conjunctions of singular statements even though a realist view of the sentential operators, including conjunction, is no more plausible than a realist view of the quantifiers. This is often overlooked because of the common assumption that necessary equivalence is identity. If you are more comfortable with one of the sides of the equivalence, you may be tempted to suppose that what the other side says is “really nothing more” than what the side you prefer says. But, clearly, this is a mistake. It is necessarily the case that p if and only if p or p, but it would be silly, or at lest unmotivated, to speak of reducing either one to the other, of saying for example that what p says is really nothing more than what p or p says. In mathematics, to use Kant’s example, it is necessarily the case that 7+5=12, but we would not want to say that one side of the identity sign is reducible to the other.

 Gustav Bergmann, “Generality and Existence,” in Logic and Reality (Madison: University of Wisconsin Press, 1964), 69. Originally published in Theoria, 28 (1962), 1– 26.  Marsh, 214.

126

Chapter Six: Metaphysical Realism and Logical Antirealism

Frege’s and Russell’s views on generality and negation were explicit espousals of both logical realism and logical nonreductionism. According to Russell general and negative facts are objective constituents of the world, not reducible to any other kinds of facts. It would seem that what makes them general or negative must also be in the world, though Russell vacillated about the matter. Frege did not. He was explicit: general thoughts involve a second-level function, negative thoughts include negation as one of their two components. One would think that Russell’s general and negative facts must also contain components represented, respectively, by the quantifier and the negation sign. For, if they did not, how would general facts differ from the corresponding singular facts and how would negative facts differ from the corresponding positive facts? If there is an item in the world represented by “"x (Φx)” as well as an item represented by “Φa,” as Russell held, how would the two differ if not in virtue of “something” in one that is not in the other, presumably the presence in what “"x (Φx)” represents of what “"x” represents? Similarly, if there is an item in the world represented by “~ p,” as well as an item represented by “p,” how would they differ if not in virtue of “something” in one of them that is not in the other, presumably what “~” represents? For, according to Russell, the differences must be in the facts, in the world, not in our language or our minds. According to Frege, the presence of a second-level function in general thoughts is essential to the truth-values of general sentences and the presence of negation in negative thoughts is essential to the truth-values of negative sentences. Surely, the presence in Russell’s general facts of something expressed by the quantifier is essential to the truth-values of general sentences, and the presence of negation in Russell’s negative facts is essential to the truth-values of negative sentences. If, as Russell held, the truth of a statement depends on what is in the world, then what is essential to its truth presumably must also be in the world. The logical realist faces two tasks. The first, perhaps accomplished by Frege and Russell, is to refute logical reductionism, the view that “upon analysis” the question of the reality of logical objects does not arise because the logical constants that appear to stand for them have been “analyzed away.” By showing that at least in the case of negation and generality this is not so Frege and Russell forced antirealists to shun the comfort of slogans like “A negative statement is just the corresponding false affirmative statement,” “Universal statements are just the conjunctions, and particular statements the disjunctions, of the singular statements that instantiate them,” or “Molecular statements are just truth-functions of their components.” The second task the logical realist faces, however, is to convince us that, since logical reductionism is false, the logical constants do represent entities in the world. The plausibility of Frege’s and Russell’s views attached to their nonreductionism, not to their realism, which remained implausible.

4 Frege and Russell on Negation and Generality

127

The idea that “all” and “some” correspond to entities has seldom been entertained. As we shall see in Chapter Seven, Gustav Bergmann, one of the few philosophers who thought deeply about this topic, did write: “Each quantifier represents something which is sometimes presented. Had it never been presented, we would not know what the quantifier meant.”¹⁹⁶ But this is an (abductive) argument for a statement of phenomenological observation, a statement about what is “presented,” not a report of that phenomenological observation, which is what it should have been, if Bergmann were right. And surely such a report would have been false. But Bergmann at least was aware of what would be necessary if logical realism is to be defended. We shall see in Chapter Seven that so was Wittgenstein, and that in his later work Bergmann revised his uncompromisingly realist stand. Even if Frege’s and Russell’s arguments against reductionism in the case of negative sentences are accepted, antirealism about negation would remain compelling.¹⁹⁷ That negation corresponds to nothing in the world seems almost a tautology. Sartre claimed that it is consciousness that “introduces” negation into the world precisely because he believed that it is not “already” in the world.¹⁹⁸ The words “no” and “not” are learned early in childhood to signal the absence or nonexistence of a thing, to refuse an offered object, or to reject what one is told, and later to deny the truth of what one hears or reads. But surely they do not stand for any entity such as absence or nonexistence. Antirealism about generality would also remain compelling, for much the same reasons. In denying that logical expressions stand for entities, the logical antirealist is not just denying an application of the simplistic “Fido”– Fido principle, according to which every word is a name. What is denied is the natural, not at all simplistic even if ultimately mistaken, assumption that if a word is to serve a cognitive role then it must relate to something in the world in a specific, distinct, and comprehensible way, even if it does not name it, that there must be something in what is cognized that grounds its cognitive role even if it is not named by the word, and that this “something” is accessible to us, if not directly in perception, like colors and books, then indirectly in sophisticated thought, like quarks and perhaps God. The logical antirealist claims that in the case of logical expressions none of this is true. This claim should not be confused with the much weaker claim, to which chapter 5 was devoted, that language is causally necessary for cognition. Presumably, a human being who lacks a language cannot have detailed knowledge of astronomy  Gustav Bergmann, Logic and Reality (Madison: University of Wisconsin Press, 1964), 70.  But see E. Peterson, “Real Logic in Philosophy,” The Monist 60, 2, 1986, and “Logic Knowledge,” The Monist 72, 1, 1989.  Jean-Paul Sartre, Being and Nothingness, 21– 45.

128

Chapter Six: Metaphysical Realism and Logical Antirealism

or of medieval history. But surely God can, and for all we know so can intelligent life-forms in outer space. We may not understand what their knowledge would be like, but then neither can we visualize a non-Euclidean space or have auditory images of the high-frequency sounds that dogs but not humans can hear. The same may be said about humdrum cognitions like that expressed in an inventory of the thousands of chairs in a large building. We cannot make the inventory without using language or other symbols. But, surely, God can. Could the thought expressed in a universal statement like “All my toys are upstairs” be entertained by a child before learning the word “all”? Perhaps it could, if it were just a collection of several singular thoughts, e. g., that the doll is upstairs, that the ball is upstairs, and that the whistle is upstairs. But a mere collection of thoughts is not even what a conjunctive statement, a statement requiring the sentential connective “and.” And even if it were, it certainly is not what the universal statement, the statement employing “all,” expresses. As Russell would have pointed out, the conjunction must include as additional conjunct also the universal statement “These are all the toys I have.” A universal statement is “made true” not just by the “atomic facts” corresponding to its singular substitution instances, but also by “the further fact about the world that those are all the [relevant] atomic facts…[which] is just as much an objective fact about the world as any of them are.”¹⁹⁹ The linguisticism of the position we have reached so far regarding negative and universal statements appears obvious. It does fit the usual formulations of linguistic antirealism. But these formulations make no distinction between ontological and cosmological realism/antirealism, much less between metaphysical and logical realism/antirealism or between antirealism and semirealism. For example, Michael Dummett writes, “[U]ntil we have achieved an understanding of our language, in terms of which we apprehend the world, and without which, therefore, there is for us no world, so long will our understanding of everything else be imperfect.”²⁰⁰ To attempt to “strip thought of its linguistic clothing,” he also says, is to confuse it with “its subjective inner accompaniments.”²⁰¹ Dummett may be right, but surely his judgment is rashly abstract. As we have seen, there are important differences in respect of our “apprehending” them between the world and plants, animals, or minerals, between this book and all books, between this page and its being white, and as we shall see in Chap-

 Marsh, 235 – 36.  Michael Dummett, “The Justification of Deduction,” in Truth and Other Enigmas, 311.  Truth and Other Enigmas, 44.

4 Frege and Russell on Negation and Generality

129

ter Eight between winters in Iowa being severe and all winters in Iowa being severe. Cognition of the world involves confrontation with nonlinguistic items, in sense perception, introspection, perhaps also intellectual and even mystical intuition. Such confrontation need not involve language. A neonate’s exercise of normal sight does not. We need not deny the independent reality of the items thus confronted. Thus we need not deny that we have cognition that is unmediated by language. But it would be cognition at its most primitive level. Since it would not involve statements, it could not even appear to have facts as its objects and therefore, if the world is the totality of facts, to count as cognition of a world.

Chapter Seven: Logical Semirealism 1 Ineffability In the previous chapter I offered a preliminary account of logical antirealism with respect to negation and generality. One need not be a metaphysical antirealist to doubt their existence as entities in the world. Distaste for such entities has led to reductionism, the attempt to show, often in the context of the realism/antirealism debate, that negative and general statements do not really say what they seem to say. In the previous chapter we saw how Frege and Russell dealt with this attempt. Reductionism is far less plausible in the case of “not” and “all” than in the case of sentential connectives like “and,” “or,” and “only if.” Later analytic philosophers provided little detailed discussion of negation, leaving that task to phenomenologists like Heidegger and Sartre. Surprisingly, they also had little to say also about generality. An exception in this respect was Wittgenstein. So was Gustav Bergmann. I shall focus in this chapter on generality, because of its obvious, direct, relevance to cosmological antirealism. Wittgenstein usually called universal statements just “general,” as Russell had done. Sometimes, so did Bergmann. In discussing their views, I will too. All three took for granted that particular statements are equivalent to negations of negative universal statements (($x) Fx ≡ ~ ("x) ~ Fx)), and thus to need no separate discussion, though as we shall see Bergmann vacillated on this point in his early work. None considered the status of generic statements, statements like “Winters in Iowa are severe” and “Patients with prior strokes benefit from taking Lipitor,” the ubiquity of which provides the case for antirealism with decisive support. Generic statements will be the topic of chapter 8. The present chapter is devoted to universal statements, statements like “All winters in Iowa are severe” and “All patients with prior strokes benefit from taking Lipitor.” The logical expressions for generality are of special relevance to cosmological antirealism, antirealism about the world. Antirealism regarding the other logical expressions may lead also to ontological antirealism, antirealism about things. But I shall not consider this possibility here. I want to present here a version of antirealism that is measured, discerning, and as cautious and uncontroversial as possible. Cosmological antirealism is such a version. It questions only the reality of the world understood as the totality of facts. It does not question the reality of the world understood as the totality of things. It does not question the reality of “animals, vegetables, or minerals,” and thus it remains safely in accord with common sense, which would find Wittgenstein’s notion of the word as the totality of facts, not of things, too philosophical to fathom.

1 Ineffability

131

A serious discussion of general statements must focus on what Wittgenstein and Bergmann said about them. With the exception of Frege and Russell, considered in the previous chapter, there is little of metaphysical import to be found elsewhere. According to Bergmann, the universal and the particular quantifiers stand for items with which we are sometimes “presented,” i.e., of which we are sometimes aware. Few have agreed with him. Few had agreed even with Russell’s weaker claim that at least there are general facts, of some of which we have primitive knowledge. Bergmann modified his view later, moving closer to the position Wittgenstein had held. Though adamant in opposing Russell’s logical realism, Wittgenstein did not merely deny that there are general facts. He also offered a positive view, which Bergmann’s later view resembled in important respects. Both represented, with unparalleled sophistication, the middle ground between simplistic realism and simplistic antirealism that I have called semirealism. Of course, general statements had been a central subject matter of logic since Aristotle’s syllogistic. But when Frege invented modern quantification theory, they became also a fundamental concern of metaphysics. Contemporary logicians and metaphysicians seldom ask what, if anything, general statements correspond to in the world. But, as we saw in the previous chapter, the founders of analytic philosophy, Frege and Russell, did, and the question was a major theme also in Wittgenstein’s and Bergmann’s works. All five – Aristotle, Frege, Russell, Wittgenstein, and Bergmann – were aware that there could be no genuine world if there were not generality in the world. This is why all five also saw the tie of the topic to what later became known as the realism/antirealism issue, even though for Aristotle generality involved universals, for Frege general “thoughts,” and for Russell, Wittgenstein, and Bergmann general “facts” or “complexes.” Frege held that general statements express the saturation of second-level functions by first-level functions; Russell, that if true they stand for general facts; Wittgenstein, that they involve matters that can only be “shown,” not “said”; and Bergmann, that they involve the entities generality and existence, for which the universal and the particular quantifiers respectively stand. Like Frege and Russell, Bergmann rejected the facile reductionist view of universal statements as the disguised conjunctions, and of particular statements as the disguised disjunctions, of their singular substitution instances. In his early paper “Generality and Existence” he argued that, like “individuality, universality, and exemplification,” generality and existence belong to the “world’s form.” One is “presented” with them, but they do not “exist” – rather, they “subsist.”²⁰²

 Gustav Bergmann, “Generality and Existence,” in Logic and Reality (Madison: University of Wisconsin Press, 1964), 69. Originally published in Theoria, 28 (1962), 1– 26.

132

Chapter Seven: Logical Semirealism

Thus Bergmann already seemed inclined toward only semirealism regarding generality, not realism though also not antirealism. He used “existence” in two senses: the sense of “some,” which the particular quantifier expresses, and the sense of “exist” in which the world’s form does not exist. Later, he expressed regret over the ambiguity. It is absent from the posthumously published New Foundations of Ontology, where Bergmann’s views received their most developed and detailed formulation. “Generality and Existence” had been preceded by “Ineffability, Ontology, and Method.”²⁰³ Bergmann described the two articles as “materially one.” The first topic of “Ineffability, Ontology, and Method” was the “ineffability” of individuality and exemplification. He wrote, “When I know that this is a green spot, I know also that (1) the spot is an individual, (2) the color is a character, and (3) the former exemplifies the latter (and not, perhaps, the latter the former). How could I know all this if it were not, in some sense, presented to me?”²⁰⁴ But what was thus presented could not be also represented, at least not without futility. “Looking at a name…I know…even if I do not know which thing it has been attached to as a label…the kind of thing, whether individual or character, to which it has been or could be attached.”²⁰⁵ Saying that it is an individual or that it is a character would tell one nothing, for it presupposes what it purports to say. Its even having sense depends on its being true. Yet the sentence is not gibberish. Bergmann went on to say that a certain name “is on the lips of every likely reader,” but that he would not mention it because he did not “on this occasion wish to make assertions about the reading of a notoriously difficult text.”²⁰⁶ The name of course was Wittgenstein’s, and the text was Tractatus Logico-Philosophicus. Wittgenstein had written, “In a certain sense we can talk about formal properties of objects and states of affairs… It is impossible, however, to assert by means of propositions that such internal properties and relations obtain: rather, this makes itself manifest in the propositions that represent the relevant states of affairs” (4.122). As we saw in Chapter Four, by “formal” or “internal” property Wittgenstein meant such properties as being an object and being a fact, “external property” being reserved for what Bergmann meant by “character,” e.g., color or shape. Statements about an object say what external properties it has. Formal properties, however, cannot be properly predicated, “said,” though they can be shown. The similarity of Bergmann’s views in “Ineffability, Ontology, and Method” and “Generality and Existence” to Wittgenstein’s views in the Tractatus was obvious, as    

Philosophical Review, 69 (1960), 18 – 40, also included in Logic and Reality. Logic and Reality, 47. Logic and Reality, 49 – 51. Logic and Reality, 50.

2 Wittgenstein on Generality

133

Bergmann readily acknowledged. It involved, however tacitly, Wittgenstein’s distinction between “saying” and “showing,” which Wittgenstein later called the main contention in the Tractatus since it concerned “the cardinal problem of philosophy.” In Chapter Four I noted that some interpreters, for example, Cora Diamond and Warren Goldfarb, deny that according to the Tractatus there is anything that can be shown but cannot be said. I pointed out, however, that Wittgenstein held that what only shows itself is “the higher” (das Höhere). Surely, he did not think that the higher is nothing. To say that Socrates is an individual, rather than, say, a property or a relation, is not to make a remark about Socrates’s properties, but neither is it to say nothing. To speak of the sense of life is not like speaking of the length of human life, but it is hardly speaking of nothing. The reception of Wittgenstein’s and of Bergmann’s views in the philosophical community were similar, perhaps because both referred to the ineffable. Both dealt with metaphysical questions that few philosophers had even considered, and offered answers of which no previous philosophers had been aware. Bergmann’s critics complained that his philosophy is a Meinongian jungle, or they just avowed that they found it “too difficult.” The critics of Wittgenstein’s Tractatus disparaged it as “too metaphysical,” or they just interpreted it in terms of the much later Philosophical Investigations – they found misery in Wittgenstein’s glory, and glory in Wittgenstein’s misery, as Bergmann sarcastically put it.²⁰⁷ The ineffable, what only shows itself, is not nothing. It does show itself. One is inclined to ask, If it is not nothing, then what kind of being or reality does it have, what is its “ontological status”? Natural though this question may be, it presupposes the sharp and unilluminating distinction between what is real and what is not real, which Wittgenstein attempted to bypass with his distinction between saying and showing. I have already discussed the distinction in other contexts, and here I shall consider the details of how it affected his account of generality in the Tractatus and later works.

2 Wittgenstein on Generality In Tractatus 5, Wittgenstein wrote that “A proposition is a truth-function of elementary propositions.” He had explained that “The simplest kind of proposition, an elementary proposition, asserts the existence of a state of affairs” (4.21), and that “It is obvious that the analysis of propositions must bring us to elementary

 Bergmann used these terms in “The Glory and the Misery of Ludwig Wittgenstein,” Rivista di Filosofia, 52, 1961, 587– 606, Italian translation, included in Logic and Reality.

134

Chapter Seven: Logical Semirealism

propositions…” (4.221). (In the Introduction to the Second Edition of Principia Mathematica, Russell wrote that “Atomic and molecular propositions together are ‘elementary propositions’.”²⁰⁸ It seems to follow that a general proposition is also a truth-function, presumably of its singular (“elementary”) substitution instances. And so, in a letter to Wittgenstein written in 1919, Russell objected: “[In an account of general (universal) propositions in terms of elementary propositions,] it is necessary also to be given the proposition that all elementary prop [ositions] are given.”²⁰⁹ As we saw in the previous chapter, Russell had argued that, for example, the universal proposition “All men are mortal” is not equivalent to “Socrates is a man and Plato is a man and Aristotle is a man and…,” unless another universal proposition such as “All men have been enumerated” or “These are all the men there are” is added to the conjunction. If “All men are mortal” is translated as “(x) (if x is a man then x is mortal),” which seems to say that every object is such that if it is a man then it is mortal, Russell’s objection would be that it is not equivalent to the conjunction of all singular statements of the form “if x is a man then x is mortal” unless the proposition that all objects have been enumerated or taken into account is added to the conjunction. The thesis of materialism, “(x) (x is material),” would not be equivalent to the conjunction of all statements of the form “x is material,” unless we added that we have considered all objects, all values of the variable x. Wittgenstein’s reply to Russell’s objection was vehement. He did not deny Russell’s logical point, which amounted to reminding us that the rule of universal instantiation is not a rule of equivalence. Rather, Wittgenstein wrote, “There is no such proposition [as that all elementary propositions are given]! That all elementary propositions are given is shown by there being none having an elementary sense which is not given…” And he added: “I’m afraid you [Russell] haven’t really got hold of my main contention, to which the whole business of logical prop[osition]s is only a corollary. The main point is the theory of what can be expressed [gesagt, said] by prop[osition]s – i.e., by language – (and, which comes to the same, what can be thought) and what can not be expressed by prop[osition]s, but only shown (gezeigt)”; which, I believe, is the cardinal problem of philosophy.”²¹⁰ Wittgenstein in effect pointed out the implications of his distinction between saying and showing for the notion of the totality of elementary propositions. As

 Alfred North Whitehead and Bertrand Russell, Principia Mathematica to *56 (Cambridge: University Press, 1962), xvii.  Russell: the Journal of the Bertrand Russell Archives 10, 2, 107– 109.  Letters to Russell, Keynes and Moore, 71. In Principia Mathematica (Part I, Summary), an elementary proposition is defined as one that “contains no reference, explicit or implicit, to any totality.”

2 Wittgenstein on Generality

135

we saw in chapter 4, the distinction is especially clear when applied to formal or internal properties, such as being an object. These properties can only be shown. This is why we cannot speak of the “totalities” determined by them: it is nonsensical to speak of the total number of objects or facts, because “object” and “fact” signify formal concepts (4.1272). If “fact” is a formal concept, surely so is “proposition,” and therefore also “elementary proposition,” since propositions are logical pictures of facts There is no such proposition as Russell’s “all elementary propositions are given” because “proposition” is a formal concept, like “object” and “fact.” In his objection, Russell seemed to take for granted what later was called the “substitutional” interpretation of quantification, according to which, roughly, general statements refer to all the elementary statements that are their substitution instances. According to the alternative, more intuitive, “objectual” interpretation of quantification, general statements refer to all values of the individual variable they contain, that is, all objects.²¹¹ Whether the two interpretations in fact require such reference is a debatable question, but we need not consider it here. What matters is that even if Russell had taken for granted the objectual interpretation, Wittgenstein’s reply would have been similar. Russell’s objection would have been that the proposition “all objects are given” must be given, and Wittgenstein would have replied that there is no such proposition because “object” signifies a formal concept. This has to be understood with some care. It does not mean that there are no universal propositions. What is meant is that a universal proposition is not about propositions or about objects, and thus does not say anything about all propositions or all objects, because, “proposition” and “object” being formal concepts, no proposition could be about propositions or objects. For example, the logically harmless “(x) (x is material)” must not be confused with “(x) (if x is an object then x is material),” which employs the formal concept “object” and therefore says nothing. Indeed, this is readily appreciated in common discourse. It is how we ordinarily understand universal propositions. We find it natural to say that “All men are mortal” is about all men, not, as logic suggests, that it is about all individual things, including the moon, my pen, your feet, etc., saying about each that if it is a man then it is mortal. Saying the latter belongs only in the classroom. “All men are mortal” does not include the predicate “is an object,” the only predicates in it are “is a man” and “is mortal.” If stated as “(x) (if x is a man then x is mortal),” the only expression in it other than those pred-

 The classic discussion of the two interpretations of quantification is Ruth Barcan Marcus’s, in “Interpreting Quantification,” Inquiry 5, 1962, 252– 259.

136

Chapter Seven: Logical Semirealism

icates would be the variable “x,” which is neither a name nor a predicate. It is, rather, “the proper sign for the pseudo-concept object” (4.1272). Wittgenstein’s reply to Russell appealed, not to some “analysis” of general propositions that was different from Russell’s, but to the formal status of the property of being a proposition, and thus of the property of being an elementary proposition. Like Russell, he refused to analyze a universal proposition as the conjunction, the truth-functional product, of its singular substitution instances: “I dissociate the concept all from truth-functions…” (Tractatus 5.521). On the other hand, he did not deny that, “in a certain sense,” a universal proposition is equivalent to that conjunction: “Indeed the understanding of general propositions palpably depends on the understanding of elementary propositions” (4.411). Are 4.411 and 5.521 consistent? The answer is provided by the three sentences that immediately follow 5.521: “What is peculiar to the generality-sign is first, that it indicates a logical prototype, and secondly, that it gives prominence to constants” (5.522), “The generality-sign occurs as an argument” (5.523), and “If objects are given, then at the same time we are given all objects. If elementary propositions are given, then at the same time all elementary propositions are given…” (5.524). The occurrence of the variable “x” in “(x) Fx” “gives” objects in the sense that the variable ranges over objects – it is objects that it admits as values, rather than, say, properties or propositions. The generality-sign in “(x) Fx” is “(x) (…x),” the form of “(x) Fx” in virtue of which it is a general proposition.²¹² It may indeed be said to indicate a “logical prototype.” And it may be said to occur in the proposition as an argument because it is part of what determines the truth-value of “(x) Fx,” just as a name put in place of the variable x in Fx would be part of what determines the truth-value of the resulting singular proposition. The statement “(x) fx” is a truth-function of its substitution instances, not because it is replaceable by the conjunction “fa . fb . fc ….,” but because its truth depends, in a straightforward, literal, sense, on their truth. But this only shows itself. It is not and cannot be said. If the variable “x” “gives” objects in the sense that it takes objects as its values, the generality-sign “(x) (…x)” may be said to give all objects. A universal proposition is thus, “in a certain sense,” about all objects (all values of “x”), but this is not something it says, it is something it shows. The claim that “There are objects” is a pseudo-proposition does not mean that there are no universal propositions. For such propositions are not about objects, even though the

 G.E.M. Anscombe (Introduction to Wittgenstein’s Tractatus [London: Hutchinson University Library, 1959], 145) and Robert Fogelin (Wittgenstein [London: Routledge, 1987, second edition], 65) think that the generality-sign in “(x) Fx” is the variable x.

2 Wittgenstein on Generality

137

variable in them is understood as ranging over objects. They do not contain the predicate “x is an object.” This is why a general proposition might be said to “refer” to all objects, if we accept the objectual interpretation, or to all elementary propositions, if we accept the substitutional interpretation. But such “reference” would consist in showing, not saying. Since “object” is a formal concept, “(x) fx” does not say that all objects are f. Nor does it say that all elementary propositions of the form “fx” are true, since “elementary proposition” is also a formal concept. All propositions of the form “fx” may be said to be shown by that form, in the straightforward sense that it is their form – this is why “(x) fx” is a general proposition. In a no less straightforward sense, the truth of “(x) fx” depends on there not being a proposition of the form “fx” that is false. But “(x) fx” does not say that there is no proposition of the form “fx” that is false. It is not a proposition about propositions. Since “proposition” is a formal concept, there can be no proposition about all propositions of the form “fx” or of any other form. One of Russell’s complaints in his letter to Wittgenstein was that “it is awkward to be unable to speak of [the negation of all the values of the propositional variable ].” Wittgenstein replied: “This touches the cardinal question of what can be expressed by a prop[osition] and what can’t be expressed, but only shown. I can’t explain it at length here. Just think that, what you want to say by the apparent prop[ositin] ‘there are 2 things’ is shown by there being two names which have different meanings….e.g., φ(a, b)…doesn’t say that there are two things, it says something quite different; but whether it’s true or false, it SHOWS what you want to express by saying: ‘there are 2 things.’” Then Wittgenstein added: “I suppose you [Russell] didn’t understand the way, how I separate in the old notation of generality what is in it truth-function and what is pure generality. A general prop[osition] is a truth-function of all PROP[OSITION]S of a certain form…I suppose you don’t understand the notation [for the values of the propositional variable ]. It does not mean ‘for all values of .”²¹³ I suggest that “what is truth function” in “(x) fx” is what is expressed by “(x) …,” and that “what is pure generality” is what is expressed by “fx.” All propositions of the form “fx” may be said to be shown by that form, in the straightforward sense that it is their form – this is the “pure generality” in “(x) fx.” And in a no less straightforward sense, the truth of “(x) fx” depends on there not being a proposition of the form “fx” that is false – this is the “truth function” in “(x) fx.” But “(x) fx” does not say that there is no proposition of the form “fx” that is false. It is not a proposition about propositions. Since “proposition” is a formal con-

 Letters to Russell, Keynes and Moore, 73.

138

Chapter Seven: Logical Semirealism

cept, there can be no proposition about propositions of the form “fx” or of any other form. While the distinction between saying and showing has a reasonably clear and important application to propositions of the forms “x is an object” and “next-to is a relation,” how it applies to other, more complicated cases is less clear though not less important. This is certainly true of its application to general propositions. To understand it better, we may take advantage of the notion of presupposition that Strawson proposed decades later and we examined in chapter 2, and assume, at least provisionally, that presupposing something includes implicitly reference to it. Then, if we accept the objectual interpretation of quantification, we can hold that, though “(x) Fx” does not refer to objects and thus does not say that all objects are F, it may be taken to presuppose such reference and thus to refer to objects implicitly. It is “(x) (if x is an object then x is F),” not “(x) Fx,” that explicitly mentions and thus refers to objects, and for this reason it is inadmissible. The point may become clearer if we return to the statement “All men are mortal.” Translated as “(x) (if x is a man then x is mortal),” it does not say that all objects are such that if they are men then they are mortal, but it may be taken to presuppose this since the variable x ranges over all objects and thus to refer implicitly to all objects. What “All men are mortal” does say is just that all men are mortal. It is about men, not about objects. Were we to accept the substitutional interpretation of quantification, we could hold that “(x) Fx” does not say that all elementary propositions of the form “Fx” are true, but it can be taken to presuppose this and thus to implicitly refer to all elementary propositions. The actual statement “All men are mortal” does not say that all propositions of the form “if x is a man then x is mortal” are true, but it may be taken to presuppose this. What it does say is just that all men are mortal. It is about men, not about elementary propositions. After his return to Cambridge in 1929, Wittgenstein continued working on the topic of generality. But what, according to the Tractatus, was shown by a general proposition, whether universal or particular, in Philosophical Remarks, written shortly after his return though published posthumously, was now described as possibilities left open by the proposition, like an incomplete or partially indefinite picture: “The general proposition ‘I see a circle on a red background’ appears simply to be a proposition which leaves possibilities open. A sort of incomplete picture. A portrait in which, e. g., the eyes have not been painted in. But what would this generality have to do with the totality of objects?”²¹⁴ Also: “If

 Ludwig Wittgenstein, Philosophical Remarks (Chicago: University of Chicago Press, 1975, ed.

2 Wittgenstein on Generality

139

I give a correct description of a visual field in which three red circles stand on a green ground, it surely won’t take the form of saying ‘($x (x, y, z): x is circular and red and y is circular and red, etc. etc.’… That is, in such a case, you may say: Certainly I know that three things have the property φ, but I don’t know which; and you can’t say this in the case of the three circles.”²¹⁵ A couple of years later, in Philosophical Grammar, also published posthumously, Wittgenstein returned to the topic. Regarding the position he had defended in the Tractatus, he wrote: “My view about general propositions was that ($x). φx is a logical sum and that though its terms are not enumerated here, they are capable of being enumerated….For if they can’t be enumerated we don’t have a logical sum….Of course, the explanation of ($x). φx as a logical sum is indefensible…. [But] it is correct that ($x). φx behaves in some ways like a logical sum and (x). φx like a product; indeed for one use of the words ‘all’ and ‘some’ my old explanation is correct, – for instance for ‘all the primary colours occur in this picture’… But for cases like ‘all men die before they are 200 years old’ my explanation is not correct.”²¹⁶ I take Wittgenstein to mean that the sentence about the primary colors would be an exception because “primary color” is an abbreviation, say, of “red, green, or blue,” and so the sentence would be an abbreviation of “red, green, and blue occur in this picture,” not a general statement. Wittgenstein also wrote, “If I say ‘there is a black circle in the square’, it always seems to me that here again I have something simple in mind, and don’t have to think of different possible positions or sizes of the circle. And yet one may say: if there is a circle in the square, it must be somewhere and have some size. But in any case there cannot be any question of my thinking in advance of all the possible positions and sizes….I would like to say that in the proposition ‘there is a black circle in the square’ the particular positions are not mentioned at all. In the picture I don’t see the position, I disregard it….”²¹⁷ When asserting the general proposition “There is a black circle in the square,” one does not mention the other possible positions and sizes the circle might have had, one does not think of them, they are disregarded. Indeed, one does not mention even the position it does have. Of course, the circle does have a position, one of an indefinite number of possible positions, but none is mentioned

Rush Rhees, tr. Raymond Hargreaves and Roger White), 115. For the origin of the text, see the Editor’s Note.  Philosophical Remarks, 136.  Ludwig Wittgenstein, Philosophical Grammar (Berkeley and Los Angeles: University of California Press, 1974, 268), composed after 1929 but published posthumously (for the origin of this text, see the editor’s Note in Editing).  Philosophical Grammar, 259.

140

Chapter Seven: Logical Semirealism

in “There is a black circle in the square.” Wittgenstein did not explain these remarks in detail. Nevertheless, they do fit what he had said, not in the Tractatus but in his 1919 correspondence with Russell: the possible positions of the circle are not “said” but “show” themselves. In Wittgenstein’s earlier terminology, which he no longer employed, we might say that the possible positions of the circle are not “said” but “show” themselves. According to Russellian logic, the universal statement “(x) Φx” says that all individual objects are Φ, that everything is Φ, and that ($x). Φx says that some individual objects are Φ, whichever they might happen to be. But in the Tractatus Wittgenstein had held that “it is nonsensical to speak of the total number of objects,” indeed of objects at all, since “object” is a formal concept. Now, in Philosophical Grammar, he makes the further claim that general statements are often, perhaps usually, not even intended to be used in accordance with Russellian logic. The statement “There is a circle in this square” is not intended to say anything about all objects, not even about all objects that are in the square. In effect, Wittgenstein suggests that the general particular (“existential”) statement “There is a circle in the square” and the general universal statement “There are only two things that are circles in this square” are better understood as analogous to the singular statement “This circle is in this square,” rather than to quantified statements containing a variable ranging over all individual objects, or even over just all circles in the square. But how then do the general statements differ from the singular statement? We find no answer in Philosophical Grammar. The answer Wittgenstein had provided in his 1919 reply to Russell was that the generality of a general statement consists not in what it says but in what it shows. He insisted that our use or understanding of general statements is far removed from what Russell’s Principia Mathematica tells us. We do not use “There is a circle in the square” to say something about all things, or even about all circles, viz., that some are in the square. We certainly do not use it to say that it is not the case that no circles are in the square. But In Philosophical Grammar Wittgenstein expressed doubts about the very propriety of representing ordinary general propositions in the canonical forms of Principia Mathematica. “[A] proposition like ‘there are two circles in this square’ is rendered as ‘there is no object that has the property of being a circle in this square without being the circle a or the circle b’….[T]he Russellian notation here gives an appearance of exactitude which makes people believe the problems are solved by putting the proposition into the Russellian form.”²¹⁸

 Philosophical Grammar, 265.

3 Bergmann on Generality

141

Wittgenstein’s misgivings in Philosophical Grammar about the Russellian interpretation of general statements were motivated by his growing concern with how we actually talk. In Remarks on the Foundations of Mathematics he wrote: “‘Mathematical logic’ has completely deformed the thinking of mathematicians and of philosophers, by setting up a superficial interpretation of the forms of our everyday language as an analysis of the structure of facts.”²¹⁹ He was on his way to the Philosophical Investigations. In the next chapter I will argue that “mathematical logic” also ignores the fact that in actual talk and thought most generalizations are what linguists call generic statements, i. e., statements of the form “Fs are Gs,” rather than universal statements, of the form “All Fs are Gs,” and that even universal statements are usually intended only as generic.

3 Bergmann on Generality In “Generality and Existence” Bergmann had used an example similar to Wittgenstein’s example in Philosophical Grammar: being “presented” with a square inside a circle. Bergmann asked, With what else must I be presented when I say “This square is the only one inside this circle”? He pointed out that the transcription of the statement would be “F(a, b) . (x) [(x = a) v ~ F (x, b)],” with “(x),” “a,” “b,” and “F” standing respectively for the universal quantifier, this square, this circle, and the relation of being inside.²²⁰ Bergmann answered the question by saying that when presented with a single square inside a circle he was also presented with generality and existence. He was presented with generality in seeing that the square was the only square in the circle, and with existence in seeing that there is a square in the circle. They are the entities that the universal quantifier, “(x)” or the word “all,” and the particular quantifier, “($x),” the word “some,” or the phrase “there is at least one,” respectively represent. Therefore, there are such entities as generality and existence, though they “subsist,” rather than “exist.”²²¹ In New Foundations of Ontology Bergmann continued to hold that he was presented with generality and existence, though now he called them the universal and the particular quantifiers. He went far beyond “Generality and Existence” by offering a much more complex account of quantification, which still

 Ludwig Wittgenstein, Remarks on the Foundations of Mathematics (Cambridge: MIT Press, 1994), 300.  Logic and Reality, 71. Philosophical Grammar was published 22 years after “Generality and Existence.”  Logic and Reality, 68, 70.

142

Chapter Seven: Logical Semirealism

resembled Wittgenstein’s but was a part of a rich, all-encompassing ontology, which Wittgenstein never attempted. Bergmann renounced his earlier distinction between existence and subsistence, subscribing now to the seeming paradox that “whatever is thinkable exists.”²²² Yet he acknowledged that “the differences among some of the several existents…are very great indeed…momentous, or enormous,”²²³ thus acknowledging the rationale for the distinction. Bergmann’s assertion that everything thinkable exists should be no more surprising than Meinong’s assertion that “there are things of which it is true that there are no such things,” but it is less confusing. There is a golden mountain (Meinong’s famous example), it has being, it exists, Bergmann would say, but of course it is fundamentally different from, say, the Rocky Mountains. Like the latter, it is a “complex” of facts, but unlike the Rocky Mountains it is “pervaded” by the “mode of potentiality,” not the “mode of actuality.” Bergmann’s critics, like Meinong’s, seem to attach magic significance to the words “exist” and “being.” But, important though they are, still they are words, conventional signs, and their ordinary use might not be suited for the purposes of ontology. What matters is that we can think and talk about, even describe in some detail a golden mountain east of Denver, just as we can think, talk about, and describe the Rocky Mountains west of Denver. What we must not do, of course, is to think that the former is an actual mountain. Bergmann began his account of generality in New Foundations of Ontology by denying that variables, whether free or bound, stand for anything.²²⁴ This was not surprising in itself, but it did require surprising changes in the analysis of general statements. We find Bergmann now saying that the universal quantifier in a general statement stands for a function, to be represented by the sign “\/” but without attaching a variable such as “x.” The function takes as argument a “2-tuple” that consists of the individual thing in the singular fact asserted by some singular substitution instance of the general statement and that singular fact itself. The value of the function is the general fact asserted by the statement.²²⁵ If, for example, the statement is “All f1’s are f2’s,” the 2-tuple might be . Bergmann used “2-tuple,” instead of “pair,” because in New Foundations he also offered a highly original account of sets that prohibits casual uses of set-theoretical terms. Perhaps most surprisingly, however, he insisted there that the conscious state or awareness of a general fact, which he calls the “referent” of the corresponding gen   

New New New New

Foundations, 61. Foundations, 43. Foundations, 64. Foundations, 167– 168.

3 Bergmann on Generality

143

eral statement, also includes an “auxiliary act” of consciousness, the “intention” (intentional object) of which is the sentence itself, the words used in making the statement. Bergmann called this intention “the text of the awareness.”²²⁶ “I cannot ‘think’ any generality such as, say, all-men-are-mortal, without at the same time ‘thinking’ the words ‘all-men-are-mortal.’….One cannot believe, or doubt, or remember, and so on, any generality without also perceiving the appropriate words.”²²⁷ Indeed, Bergmann went on to say that “all awarenesses, except primary Perceivings and Imaginings (and undoubtedly some ‘Feelings’), are inseparable from their texts. That…not only gives language its due without giving it too much; it also reassuringly recovers the sound core in a large body of recent and contemporary thought…from Watson to Wittgenstein.”²²⁸ The fact that all f1’s are f2’s is “built” by the function \/ “not just from one argument but, indifferently, from an indefinite number of alternative arguments…from , from , and , and so on. In the text of [the awareness], however…there is no cue to this multiplicity.”²²⁹ Bergmann gave “all green (things) are square” as an example. It is the text of an awareness that has as its object the general fact, presumably not actual, that all green (things) are square. The 2-tuples that the function \/ takes as arguments, e. g. , are not mentioned in the general statement, there is no “cue” in it to their “multiplicity.” But they all are essential to the general fact. That fact would not be actual if the singular facts in the 2-tuples were not all actual: “all f1’s are f2’s” would not be true if all its singular substitution instances were not true. From which of them the function \/ builds the general fact is ontologically indifferent. But it might not be psychologically indifferent, since the speaker or hearer of the general sentence must at least in principle be able to perceive or imagine one of them. I shall return to this latter point. Thus the analysis of “all f1’s are f2’s,” i. e. its ontological analysis, “is not, conventionally … (x) [f1(x) ⊃ f2(x)], but, rather, alternatively and indifferently… \/ [ or any of its variants; indifferently because all those variants are one and not many.”²³⁰ The “variants” of \/ [(a, f1(a) ⊃ f2(a)], of course, are \/ [], \/ [], and so on. Each is an alternative analysis of the one and same general fact. We may note that in standard logic it is also indifferent, unless the context requires otherwise, whether we symbol-

    

New New New New New

Foundations, Foundations, Foundations, Foundations, Foundations,

208. 204. 234. 235. 202.

144

Chapter Seven: Logical Semirealism

ize “all f1’s are f2’s” as “(x) [ f1(x) ⊃ f2(x)], “(y) [ f1(y) ⊃ f2(y)], or “(z) [ f1(z) ⊃ f2(z)]. But standard logic uses variables, which according to Bergmann represent nothing and therefore have no place in ontological analysis. It may seem that Bergmann’s insistence that the arguments that the quantifier \/ takes, say, in the case “all f1’s are f2’s,” are 2-tuples is an unnecessary complication, but the reasons for it are compelling. What else could those arguments be? Not f1(x) ⊃ f2(x), because it contains variables. Nor the properties f1 and f2 themselves. According to Bergmann’s “principle of acquaintance,” one cannot be presented with f1 and f2 except when they are exemplified.²³¹ But even if one could, being presented with them as well as with the quantifier \/ would hardly count as being presented with the fact that all f1’s are f2’s, or indeed with any fact: the three presentations might be unrelated. They might be unrelated even if f1 and f2 are exemplified: one might be presented with them as well as with the quantifier by virtue of being presented with the fact, say, that all f2’s are f1’s. Without the singular fact that is one of the terms of the 2-tuple, there would be no relevant conscious state or awareness at all when one makes the general statement, for there would be nothing relevant for one to be aware of. Could the quantifier take as argument the singular fact f1 (a) ⊃ f2 (a), rather than the 2-tuple ? No, because even if \/ could take f1 (a) ⊃ f2 (a) as argument, its value would not be a general fact. The notation must also make explicit which constituent of the singular fact the quantifier operates on, just as in standard logical notation it must be explicit which variable the quantifier binds. If variables are not used, this can be explicit only if the quantifier, so to speak, “brings” the constituent “out of” the singular fact, while also “retaining” the singular fact. The constituent and the singular fact must both be explicitly in the argument that the quantifier takes, and this amounts to saying that the argument must be the 2-tuple of which they are the members. Bergmann expressed the point by saying that the individual is the “target” of the quantifier, while the singular fact is its “scope.” In the case of the statement “all green (things) are square,” the target might be any particular perceived or imagined object, even your hand, which, if the statement were true, would indeed be square if it is green. There are important similarities between Bergmann’s account of generality in New Foundations of Ontology and each of Wittgenstein’s stands on the topic we have considered – in the Tractatus, in his 1919 letter to Russell, in Philosophical Remarks, and in Philosophical Grammar –, though there are also obvious differences. The singular substitution instance of the general statement that stands for the singular fact which Bergmann calls the scope of the quantifier is not asserted, it is not

 New Foundations, 65.

3 Bergmann on Generality

145

“said,” yet it must be, so to speak, in the background, if the statement is to express a relevant conscious state. We may say that it must “show” itself. For it is the singular substitution instance that provides the general statement with its target and scope, both of which must, in some sense, be “presented” or “given,” though of course not as they would be if what was asserted was the singular, rather than the general, statement. We may say that the 2-tuple from which, as its argument, the quantifier “builds” the general fact must also only show itself. Indeed, we may say that the whole indefinite number of alternative arguments from which the quantifier builds indifferently the general fact show themselves. They must be “there,” in the background, presupposed, like the indefinite number of possible positions of the circle on a red background in the example Wittgenstein gave in Philosophical Remarks. Like those positions, the alternative arguments may be thought of as “possibilities left open,” neither enumerated nor capable of being enumerated, but the speaker and hearer knowing that they are there, though not which they are. Thus, the general statement may be said to be “indefinite,” “an incomplete picture, like a portrait in which, e. g., the eyes have not been painted in.” There can be no question of thinking in advance of all the alternative arguments the quantifier may indifferently take, they are not mentioned at all, they are unseen and disregarded. Yet they are there, presupposed, like the different possible positions and sizes of the circle in the square that the statement “there is a black circle in the square” allows for, even though when making the statement the speaker has “something simple in mind.” In Philosophical Remarks Wittgenstein denied that the general proposition “I see a circle on a red background” has anything to do with “the totality of objects,” but in the Tractatus he could have said that it shows that totality. I have suggested that when Bergmann writes that what he calls the scope of the quantifier is not asserted, yet must be in the background if the statement is to express a relevant conscious state, we may say that it must “show” itself, and also that what he calls the alternative arguments from which the quantifier builds the general fact may be said to show themselves. If we did say this, would “show” have the sense that it had in the Tractatus, or in Wittgenstein’s assertion in his 1919 letter to Russell that while a universal statement does not say that all elementary (singular) propositions are given, this is shown by there being none having an elementary sense which is not given? The truth is that Wittgenstein did not explain that sense, just as Bergmann did not explain, for example, the sense of his term “presented.” The reason, in both cases, presumably was that they thought what they meant was too basic to allow for explanation. According to Bergmann, one must be presented with what he called the quantifier, i. e., generality, what “(x)” or “\/” stands for, and Wittgenstein presumably would have disagreed. “There are no ‘logical objects’,” he wrote, thus

146

Chapter Seven: Logical Semirealism

announcing his break with the logical realism of Frege and Russell, though, as we have seen, he did not adopt logical antirealism. But Wittgenstein might have agreed that the thoughtful use of a general sentences involves awareness, even if only peripheral and unfocused, of the sentence itself, what Bergmann called the “text,” whether by seeing, hearing, or imagining it. Bergmann held that this is a phenomenological, or as he also put it, anthropocentric, even anthropological, fact – this is how we humans think and speak. But Bergmann also offered a detailed account of this fact, which Wittgenstein did not. Bergmann explained that the text is needed in order to close the “phenomenological distance” between what is presented to us when we thoughtfully make a general statement and the “assay,” ontological analysis, of what the statement says, “the gap between what the text of an awareness may lead one to expect, on the one hand, and the assay in fact proposed for its referent, on the other.”²³² The text is “fused,” “absorbed,” into “the nontext,” he wrote,²³³ it has “fusing power.”²³⁴ The general fact that all f1’s are f2’s is built by \/ from an indefinite number of alternative arguments, from , , , and so on, but there is no “cue” to this multiplicity in the sentence “all f1’s are f2’s,” nor of course in its transcription, whether the conventional “(x)f1(x) ⊃ f2(x)” or Bergmann’s “\/ .” This goes unnoticed because of the “fusing” power of the sentence. On no account of generality does a true general statement contain explicit reference to the multiplicity of what makes it true. Whatever account we accept, we must rely on the statement to serve, so to speak, as proxy for that multiplicity. A merit of Bergmann’s account is that it makes explicit what all accounts of generality must admit, namely, that when saying, e. g., “all green things are square,” we could, as he pithily puts it, in principle also say regarding any individual thing “if this were green then this would be square.”²³⁵ Saying the latter would differ from saying the former only by making explicit that the assertion is a thoughtful one, not a mere utterance, that one actually has something relevant in mind when making it. In the old empiricist terminology, it would make explicit the presence “before the mind” of an “idea,” whether of “sensation” or “imagination.” In Bergmann’s terminology, it would make explicit the presence of an individual actually perceived or imagined. Bergmann suggests that if our natural expression was, e.g., “generalized for this: if this is green then this is square,” rather than “all green things are square,”    

New New New New

Foundations, Foundations, Foundations, Foundations,

23. 216 – 217. 235. 235.

3 Bergmann on Generality

147

then when we find the truth of “all f1’s are f2’s” obvious we would be presented with the actuality of such complexes as \/