WHY WE CAN'T UNDERSTAND THOUGHT FROM THE OUTSIDE 3 page
The subjectivist cannot save his position by conceding the unavoidable appearance of certain objective forms of thought in our actual procedures of reasoning--conceding the appearance as a psychological fact--while at the same time insisting that this doesn't mean that the aim of those procedures is to lead us to what is true independent of our beliefs. He cannot do this, because such a "phenomenological reduction" would again be to try to get outside of these thoughts and regard them merely as appearances--which is precisely what can't be done.
Attempts to relativize objectivity to a conceptual scheme fail for the same reason. Suppose someone concedes that in a sense not everything can be subjective--that in any system of thought something must play the role of that which is objectively or nonrelatively valid--and that we necessarily run up against it when we attempt to class some of our other responses as subjective. And suppose he then suggests that this might be something different in different conceptual schemes or different types of minds, that in any case it seems to belong to the contingent cognitive psychology of reasoning. The reply is that since reasoning produces belief, and belief is always belief in the truth of what is believed, the distinction between
the mere phenomenological acknowledgment of reason and the recognition of its objective validity is not intelligible. We can't, for example, just observe from the sidelines that logic provides an unconditional frame for our thoughts. We may of course decide, for good reasons, to abandon as erroneous forms of argument that we once found persuasive. But if reflection and argument actually do persuade us of something, it is not going to be possible for us at the same time to regard that as just a deep fact about the phenomenology of thought. This is merely an instance of the impossibility of thinking "It is true that I believe that p; but that is just a psychological fact about me; about the truth of p itself, I remain uncommitted."
The idea of alternative conceptual schemes is no help here. There are types of thought we cannot do without, even when we try to think of ourselves, from outside, as thinking creatures. We are no more able to get outside of those thoughts when thinking of other possible thinking creatures. So the idea of an alternative mind or conceptual scheme is useless in distancing ourselves from such thoughts: Their content defeats all attempts to relativize it. 11. In the end, I believe, there is no position in intellectual space for the perspectivalist to occupy.
However much one may try to construe one's concepts and thoughts naturalistically, as the expression of contingent
This is not essentially different from Davidson's attack on the idea of alternative conceptual schemes. Though Davidson's result emerges from the conditions of interpretation, it is not merely a matter of having to see other minds in terms of my own--which sounds much too subjective. Rather, it is the actual content of certain thoughts about the world and forms of reasoning that sets the conditions of interpretation: Nothing could qualify as thought which did not meet those conditions. See Donald Davidson, "On the Very Idea of a Conceptual Scheme," in his Inquiries into Truth and Interpretation ( Oxford University Press, 1984).
forms of life, the logic of the effort will always generate thoughts too fundamental for this, thoughts which one cannot get outside of in this way and which one must employ in trying to view from the outside anything else that one does. These thoughts cannot in any way be given a first-person interpretation or qualification: They bob to the surface again in their unqualified form whenever we try to subordinate them to psychology, sociology, or natural history. I don't mean that they all have irreversible finality; sometimes they are refuted or displaced by superior reasons. But they form an outer boundary whose interest is nonrelative. My aim is to argue this in more detail with respect to particular forms of reasoning.
There is a further general response that I want to mention at this point, one that haunts all rationalist and realist philosophies. The question may now be asked: Even granting your description of how we think, why doesn't that show only that we cannot say that logic, for example, or ethics, is rooted in our natural, unquestioned practices, but that this nevertheless shows itself in the way in which arguments and justifications come to an end, in judgments on which we naturally agree? We cannot say that logic depends on such practices, because that would itself violate those practices, in which logic itself has the last word. But doesn't the ultimate authority of those practices show itself in the fact that this last word is the last word in our arguments, our thoughts, our reasonings?
This proposal derives from Wittgenstein's doctrine that the truth of solipsism cannot be stated but nevertheless shows itself in the fact that however impersonally I describe the world, it is still described in my language. I cannot truly say in this language that the world is my world, because in my language that is false: The world existed before I did, and would have existed even if I had never been born, for example. But all this is being said in my language, and that shows that in a
deeper way the world is my world, even though it cannot be said. 12.
Why is this not an appropriate comment on my claim that logical, arithmetical, and even ethical thoughts are devoid of any first-person element and admit of no first-person qualification? Why not say that their being expressions of our most deep-seated responses, practices, or habits shows itself in the actual process of reasoning, though it cannot be said?
I wish it were enough to reply, as Frank Ramsey is said to have done to Wittgenstein, "What can't be said can't be said, and it can't be whistled either." But I want to reply more strongly that the truth of solipsism is not shown by the fact that the language in which I describe the world is my language, and the truth of some other form of subjectivism is not shown by the fact that justification comes to an end at certain points at which there is natural agreement in judgments. Nothing about the framework of thought is shown by these facts, because the thoughts themselves dominate them.
Everything depends on the outcome of this peculiar contest over the last word. The subjectivist wants to give it to the recognition that justifications come to an end within our language and our practices. I want to give it to the justifications themselves, including some that are involved or implicated in that recognition, which is subordinate to them, just as the recognition that a notation is essential for thinking about arithmetic is subordinate to arithmetic itself. A certain extra step that some people try to take offers only the illusion of a thought, a path leading nowhere: When one adds, "This is simply what I do," or "This is my form of life," or "This is what I happen to care about," to what is in itself not a first-person statement, one adds nothing--not even something that can be "shown" but not said. That this is so can be seen from the fact
Ludwig Wittgenstein, Tractatus Logico-Philosophicus ( Routledge, 1922), 5.62, and Notebooks, 1914- 1916 ( Blackwell, 1961), p. 85.
that one can add such an empty qualifier to absolutely anything. If there were nonsubjective thoughts, someone would still have to think them. So the formula that simply notes this cannot be used to demonstrate that everything is based on our responses. A tautology with which all parties to a dispute must agree cannot show that one of them is right.
It is hard to be satisfied with giving the last word to certain ordinary statements or forms of reasoning. If one rejects all relativizing qualifications, it is terribly tempting to add something else in their place: "2 + 2 = 4 and cruelty is evil, not just for us, but absolutely." But if this attempts to go beyond the denial of the qualification, it may, in the immortal words of Bernard Williams, be one thought too many--with the unfortunate implication that unless something positive can be put in that space, we will be left with subjectivism after all. It would be better if we could just come to a stop with certain kinds of judgments and arguments, which neither admit nor require further qualification. But that seems to demand a level of philosophical will-power that is beyond most of us.
The impulse to qualify is very difficult to suppress. The only way to resist the constant temptation to give the last word--even the unsayable last word--to the first person, singular or plural, is to see how first-order reasoning about the world inevitably dominates these ideas if we take them seriously.
One factor that has contributed to the devaluation of reason is a misconception of the importance of language for philosophy. Since languages are human practices, cultural products that differ from one another and have complex histories, the idea that the deepest level of analysis of our knowledge, thought, and understanding must be through the analysis of language has gradually given rise to a kind of psychologism about what is most fundamental, which in turn often leads to relativism. 1 This is a kind of decadence of analytic philosophy, a falling away from its origins in Frege's insistence on the fundamental importance of logic, conceived as the examination of mind-independent concepts and the development of a purer understanding and clearer expression of them.
Language is in itself an important subject matter for philosophy, and the investigation of language is often the best place to begin when clarifying our most important concepts. The same could be said of confirmation and verification conditions. But the real subject then is not language as a contingent practice but, in a broad sense, logic: the system of concepts that makes thought possible and to which any language
One interesting form of resistance to this is Jerrold Katz's claim that it misunderstands language, which is not a mere psychological contingency but rather a Platonic abstract object. See Language and Other Abstract Objects ( Rowman and Littlefield, 1981) and The Metaphysics of Meaning ( MIT Press, 1990).
usable by thinking beings must conform. The particular contingent language that one happens to speak is essentially a tool of thought: In relation to basic questions, it functions like physical diagrams in geometry, or numerals in arithmetic; it is a perceptible aid to the formulation, recollection, and transmission of thoughts. Understanding it is a form of thought, but it is not the material of which thoughts are made. For many types of thought it is indispensable, just as diagrams are indispensable for geometry; but its relation to the content of our thoughts is often rather rough. Anyone can verify this from his own experience, but it is particularly evident in philosophy, where thought is often nonlinguistic and expression comes much later.
Because language grows in response to the demands of thought and its communication, it will reflect the character of what it is used to represent; but the order of explanation here is from the fundamental nature of things to language, even if in some cases the order of understanding can be the reverse. While there are certainly concepts which are just the artefacts of a particular language, with purely local roots, that is not true of the most important concepts with which philosophy is occupied. In particular, it is not true of the most general forms of reasoning. Those do not depend on any particular language, and any language adequate for rational thought must supply a way of expressing them.
All this is heretical, I know. Yet it seems to me much more plausible than the view that the social phenomenon of language is at the bottom of everything, and the negative thesis can be accepted even without a positive theory of what thoughts are. We cannot account for reason by means of a naturalistic description of the practices of language, because the respects in which language is a vehicle for reasoning do not admit of naturalistic or psychological or sociological analysis. To the extent that linguistic practices display princi-
ples of reasoning or show us, for example, something about the nature of arithmetical propositions, it is not because logic is grammar but because grammar obeys logic. 2. No "language" in which modus ponens was not a valid inference or identity was not transitive could be used to express thoughts at all.
Another example of explaining a type of thought in terms of contingent linguistic practice is R. M. Hare's attempt to ground ethics in the analysis of the language of morals: He finds the ultimate basis of the principle of universal prescriptivism in the contingent fact that the word "ought" is used in a certain way. 3. Not only does this take us outside of ethics in search of the ultimate basis of ethics, but it takes us to a much less fundamental level--that of contingent, empirically ascertainable linguistic practices. In this case I think the general response has been that whatever the merits of Hare's substantive moral theory, on the question of foundations he is simply looking in the wrong place. But other forms of the inversion of explanatory values in ethics have gained wider currency-for example the perspectives of sociology, cultural anthropology, or evolutionary biology.
Looking for the ultimate explanation of logical necessity in the practices, however deeply rooted and automatic, of a linguistic community is an important example of the attempt to explain the more fundamental in terms of the less fundamental. It is this pattern of inversion in general that I want to criticize; in its various manifestations, with different social facts in the position of ultimate explanation, it has become something of a cultural norm. I don't mean to deny that language is a system of conventions, or that correctness in the use
Not to mention the fact that the consequences of the rules of grammar are determined by logic. Cf. W. V. Quine, "Truth by Convention" ( 1936), in his The Ways of Paradox ( Harvard University Press, 1976).
R. M. Hare, Moral Thinking ( Oxford University Press, 1981).
of language requires conformity to the usage of the linguistic community. What I deny is that the validity of the thoughts that language enables us to express, or even to have, depends on those conventions and usages.
There is no doubt that mere custom can give rise to a strong sense of objective correctness, and that this can seem to detach itself from the contingent conventions that are its true sources. Anyone who, with comic exasperation, has lived through changes in the English language that began as mistakes and snowballed until they turned into norms will be keenly aware of this. The use of "disinterested" for "uninterested" or "enormity" for "enormous size" will probably continue to strike me as objectively wrong even if I live to an age when almost no one any longer recognizes them as errors. But in these cases of usage, as opposed to validity, one has to recognize that objectivity can't really outstrip community practice.
That is not true, however, of the content of thought, as opposed to the meanings of words. The fact that contingencies of use make "and" the English word for conjunction implies absolutely nothing about the status of the truth that p and q implies p. What is meant by a set of sentences is a matter of convention. What follows from a set of premises is not. This is just another case where relativism is inconsistent with the content of the judgments under analysis.
I also don't wish to deny that consensus sometimes plays a role in determining the extension of a concept--but these are special cases. There is a difference between the instruction "Add two" and the instructions "Pick all the ripe strawberries" or "Don't invite anyone without a sense of humor." There are some concepts like humor whose extension is determined ultimately by agreement in response among some set of persons (not necessarily speakers of a single language)--but it is impossible to think of "Add two" in this way. When we consider the difference between ourselves and the person who, if told to keep adding two, says, "996, 998, 1000, 1004, 1008, 1012
. . . ," the right thing to say is not just that we do it differently, automatically and without reflection, that here our spade is turned, and so forth. The right thing to say is that if this man hasn't simply misunderstood us in the obvious way, so that he can be corrected, then he has a screw loose and is just uttering words rather than expressing thoughts.
It is true that the possibility of a language requires at some level automatic agreement in judgments and linguistic practice: Someone whose usage diverges radically enough from that of his fellow speakers just doesn't have the concept. But such agreement is not all the concept consists in, any more than the perceptual experience by which we identify a physical object exhausts the concept of it. Rather, at some point, for people to learn arithmetic they simply have to grasp the concepts "plus" and "two," and this means understanding that correctness here is not grounded in consensus--by contrast with rules of pronunciation, for example. Whether they have grasped the concepts or not will show up in their linguistic practice, but it does not consist in that practice. Meaning, in other words, is not just use--unless we understand "use" in a normative sense that already implies meaning.
One reason for my conviction, oddly enough, is Wittgenstein's argument about rules. Whatever his own view may have been (an issue to which I shall return), I believe his observations show decisively that thinking cannot be identified with putting marks on paper, or making noises, or manipulating objects, or even having images in one's mind--however much contextual detail (including community practice) is added to such an account. Wittgenstein's grocer, with his box marked "apples" and his color chart and series of numerals, 4. is so far an empty
Philosophical Investigations ( Blackwell, 1953), sec. 1.
shell. Such a description cannot possibly explain what it is for words to have meaning. Or to take another canonical example: any reductive account of what thinking "Add two" consists in, behavioristic, anthropological, or otherwise, cannot be right because it could not by itself have the implications of that thought with respect to the difference between what satisfies it and what does not in an infinite number of cases.
Intentionality cannot be naturalistically analyzed, in other words, nor can it be given naturalistically sufficient conditions. It is not to be captured by either physical or phenomenological description. But to say that nothing that happens when I hear the instruction "Add two" determines the correct way to carry it out for any arbitrary integer depends on restricting one's conception of "what happens" to what can be described in abstraction from its intentional content, and then asking for a retrieval of the intentional content from this denuded material--which is of course impossible. The fallacy is that of thinking one can get "outside" of the thought "Add two" and understand it as a naturalistically describable event. But that is impossible. The thought is more fundamental than any facts about mental pictures or how we find it natural to go on. It is a mistake to pose the question by stepping back from the thought "Add two" itself, looking at the words or accompanying mental images apart from their content, and then asking what their content consists in. That is the crucial move in the conjuring trick.
So in my view, Wittgenstein's argument has the force of a reductio, like certain other famous arguments--Zeno's paradoxes, for example, or Hume's argument that no preference can be contrary to reason. However, as with Zeno, it is not immediately clear what it is a reductio of. The problem is to find the fatal assumption that is responsible for the unacceptable conclusion. In Wittgenstein's case, the unacceptable conclusion, as I should put it, is that thought is impossible. The faulty assumption, I suggest, is that to think or speak is simply
to do something, in the right circumstances and against the right background, which can be described without specifying its intentional content.
The conclusion that every naturalistic account of meaning simply contradicts the concept is a consequence of the Wittgensteinian paradox that Saul Kripke has expounded. 5. While I rely on Kripke's argument, I am now doubtful as to the right conclusion about Wittgenstein's positive view about meaning. According to Kripke, Wittgenstein believes not only that no natural fact about me makes it true that I mean something-he further believes that this notion should be explained not in terms of truth conditions at all but in terms of conditions of assertability. I am now inclined to think that this, too, is more reductionist than Wittgenstein would have wanted to be. Perhaps the argument establishes only the negative result that no analysis of the intentional in terms of the nonintentional can succeed--indeed, that no analysis of thought should be attempted.
The argument, in brief, is that my meaning a particular mathematical function by an expression--meaning addition by "plus," for instance--cannot consist in any fact about my behavior, my state of mind, or my brain, since any such fact would have to be finite (I am a finite being) and therefore could not have the infinite normative implications of the mathematical function. Whatever we may add on to the mere word, in the way of further states of a physically and mentally finite being like me, will not be enough to determine the difference between the right and the wrong answer to a request for the sum of any pair of integers.
Kripke expounds the problem initially as one about the past: What was it about me that made it the case, on a past occasion, that I meant addition by "plus"? The conclusion:
See Saul Kripke, Wittgenstein on Rules and Private Language ( Harvard University Press, 1982), pp. 73-87.
nothing. So there was no fact as to what I meant by "plus." But there is nothing special about the past; the conclusion is completely general:
If there was no such thing as my meaning plus rather than quus [an alternative function] in the past, neither can there be any such thing in the present. When we initially presented the paradox, we perforce used language, taking present meanings for granted. Now we see, as we expected, that this provisional concession was indeed fictive. There can be no fact as to what I mean by "plus," or any other word at any time. The ladder must finally be kicked away. 6.
This reveals the argument as a true paradox--that is, one whose conclusion is simply unacceptable. We cannot kick this particular ladder away, and if we did, we would be left without the possibility of formulating the argument for the paradoxical conclusion.
The problem is already contained in the original argument about the past, in the course of which, as Kripke says, "we perforce used language, taking present meanings for granted." But it is in a sense still present in the conclusion, where we say there can be no fact as to what I mean by "plus" or any other word. For the idea of alternative possible meanings between which no fact about me determines the actual one is still behind that thought. And what about the words "fact," "word," "mean," and so forth? We are still "perforce" using language in the attempt to "state" its impossibility. This is not a coherent position: The paradox is extremely radical.
I would put it by saying that the thought that I mean something by my words is a Cartesian thought--a thought that I cannot attempt to doubt without immediately discovering the doubt to be unintelligible. Just as I cannot doubt
Wittgenstein on Rules and Private Language, p. 21.
whether I exist, I cannot doubt whether any of my words have meaning, because in order for me to doubt that, the words I use in doing so must have meaning. In essence, the argument invites me to conclude that perhaps I'm not thinking--which is clearly the impossible denial of a Cartesian thought.
It is not impossible to discover that some of the words I am accustomed to use don't mean anything; but to think this I must use other words, like "word," which do mean something. Yet the argument for the Wittgensteinian paradox is perfectly general: If it works, it leaves nothing standing, including itself. Therefore it can't work. But that of course does not show us what is wrong with it. That's why there's a paradox.
My response is not a solution to the philosophical problem of meaning. But I conclude that since I mean addition by "plus" now, I certainly could have meant it in the past, and if no fact about me in the past that does not already include the specification of what I meant can be the fact in virtue of which it is the case that I meant addition, it follows that there is no noncircular explanation of what meaning addition by "plus" consists in. Some complex meanings can be analyzed in terms of simpler ones, but there is no noncircular explanation, in naturalistic terms--behavioral, dispositional, psychological, or physiological--of meaning in general.
The crucial problem is not just the disparity between the finiteness of physical or psychological states and the infinite implications of meaning, but, as Kripke points out, the gap between the nonnormative and the normative. 7. Meaning implies the difference between right and wrong answers or applications. Behavioral, dispositional, or experiential facts have no such implications. Therefore the former cannot consist in the latter. It is a straightforward instance of Hume's is-ought gap. 8.
Ibid., pp. 22-3 .
A Treatise of Human Nature, book 3, part I, sec. 1.
A naturalistic account of the normative is not possible in ethics, either, but that topic will be taken up properly later on. Here, the claim is that the Wittgensteinian paradox reveals it to be a mistake to think of someone's meaning something by a word as a natural fact about him that can be analyzed in nonnormative, nonintentional terms. "Meaning it is not a process which accompanies a word. For no process could have the consequences of meaning." 9. I do mean addition by "plus"; it is in a perfectly good sense a fact about me. But in response to the question "What fact?" it is a mistake to try to answer except perhaps by further defining "addition" for someone who may be unfamiliar with the term. It is a mistake to try to escape from the normative, intentional idiom to a plane that is "factual" in a different, reductive sense.
The move from the terrain of truth conditions to the terrain of assertability conditions does not seem to me an advance. So long as these, too, are described naturalistically, in terms of how people find it natural to go on and what they agree in doing "blindly," without need of further justification, I do not see how they can be regarded as giving an adequate account of the phenomenon of meaning. It is patently insufficient to say, in answer to the question how a finite being can grasp a concept like addition, which has infinite implications, that it is simply part of the common usage of the term that we are warranted in ascribing that infinite concept to a person who applies it in accordance with common practice in a finite number of appropriate cases. I can't believe that was Wittgenstein's view; it seems to me just as reductionist as a corresponding theory of finite naturalistic truth-conditions for meaning. Crispin Wright underlines the radical character of this position with respect to the possibility of truth outrunning assertibility: