WHY WE CAN'T UNDERSTAND THOUGHT FROM THE OUTSIDE 6 page
The most basic and simple forms of reasoning, in logic and arithmetic, certainly pose deep philosophical problems, but it is impossible to take seriously the idea that they are merely manifestations of contingent and local practices. We cannot think of them as less than universally valid, because not only can we not conceive of their being invalid, but we cannot conceive of a being capable of understanding them who did not also find them self-evidently valid: Nothing would permit us to attribute to anyone a disbelief in modus ponens, or in the proposition that 2 + 2 = 4.
However, most of the interesting questions which we look to reason to answer are much more difficult. We do not find the answers self-evident, or if we do, we acknowledge that the appearance of certainty may be deceptive. As I have said, this acknowledgment is often appropriate even where we are reasoning about necessary truths. A priori knowledge or belief need not be certain, even though it has greater resources for
certainty than a posteriori knowledge. But outside of mathematics and logic, uncertainty is the norm. In most cases, reasoning provides us not with proof but only with reasons for believing a conclusion likely, or for preferring it somewhat to the alternatives.
That is true in science, in other empirical disciplines, and in ethics. The reasons that support a conclusion do not typically rule out the possibility of its falsity, even if they are very strong. And there are sometimes enough reasons in support of consistent alternatives that we may be unsure that the conclusion we have to draw is actually supported by the preponderance of them: We acknowledge that reasonable persons can disagree, and often we can imagine ourselves drawing a conclusion different from the one we have actually drawn. Often reasoning in the strict sense does not support our conclusions directly but only justifies us in trusting or distrusting the more particular judgments and intuitions that occur to us naturally, or as the result of experience. Yet in such cases we believe that what we are thinking about has an answer that is not relative or subjective and that our procedures of reasoning attempt, fallibly, to capture the reasons that bear on that answer in one direction or another.
It is this type of appeal to reason that is most vulnerable to alternative diagnoses and to the charges of self-deception and false universalization. Such charges are sometimes true, and awareness of the possibility should temper our confidence-which must in any case be modest because of the straightforward possibility of mistakes in reasoning and limitations of the available evidence. But they are not inevitably true, and the problem is to give an account of the process that explains this. How is it that in logical, empirical, or practical reasoning that is not incontestable we can nevertheless claim to be using, in a possibly incomplete or inaccurate version, methods whose ideal validity is universal and not relative to anything more contingent and particular about us or our community?
In thinking of this kind, the search for what is universal is itself a regulative principle. That is made explicit in the Kantian conception of moral reasoning, but it is also true elsewhere, for we test our reasons partly by asking whether they are applications of principles that are generally valid, looking for counterexamples, and using both actual and imagined cases in the process. At least this much is true: Unless we think that anyone should draw the same conclusion from the same premises, we cannot regard the conclusion as justified by reason. Reasons are by definition general, and we aim always to extend their generality. So part of the question is whether an attachment to this method is itself something we cannot get outside of, as the form of final assessment of our beliefs-including beliefs about what is and what is not a legitimate subject for reasoning, and beliefs about the boundary between the universal and the nonuniversal.
The process gets its start from the bare conception of an objective reality, within which more subjective points of view, including our own, are embedded. At least as a possibility this is inescapable. While filling it out is extremely difficult and in some respects incompletable, it drives us to seek some nonlocally valid methods in pursuing it, for that is the only way to subject our personal starting points to any kind of testing to determine just how subjective they are. Further, and more riskily, we proceed by taking ourselves and our experiences as samples of a world that we hope to find, at some level, the same everywhere (in time as well as in space), so that the order we discover in trying to explain what we observe aims toward something broader. There is nothing special about us, in other words: Each of us is just a piece of the universe. A vindication of this type of reason would require that we make it credible that the search for order, and some of the methods for identifying that order, will survive every attempt to interpret them as merely subjective--because all such interpretations are defeated by the first-order judgments whose au-
thority they are trying to undermine. That would be structurally analogous to the situation with logic, but without the same kind of necessity in the results.
This extremely general point is so far compatible with the position that the rational base that cannot be explained away is very small--perhaps even limited to logic--and that everything else can be understood as a feature of some more particular type of viewpoint. It is also compatible with the position that reason has a fundamental role in logic, mathematics, and empirical science but that all ostensible examples of practical or ethical reason are better understood as manifestations of specific psychological dispositions. Definite conclusions on these matters depend on more substantive investigation of whether in each domain pursuit of the universal makes sense and, if so, whether it is reasonable to believe that our actual uncertain efforts in that direction are reflections of something that might be further perfected. In this chapter I shall discuss factual and scientific reasoning, but only in the most general terms. The title of the chapter (like that of the one before) may be slightly misleading. I have nothing to say about formal theories of induction and confirmation, or about their relation to the practice of empirical and scientific thinking. My interest is in what kind of thing these theories are theories of.
Reliance on reason can coexist with very substantial doubt about the results, and even with radical skepticism. In fact, traditional epistemological skepticism depends on the objectivity of reason: It is always the product of reasoning to the conclusion that various mutually incompatible alternative possibilities are all equally compatible with one's actual epistemic situation, and that it is therefore impossible to decide among them on rational grounds. Radical skepticism therefore has to rely on some thoughts that are not put in doubt and that are assumed to have objective content. But the same must be true
of less radical forms of uncertainty--the ordinary limited confidence one has in most of one's beliefs, including qualified belief in scientific theories that are accepted as the best candidates for the moment, even though we know they will be superseded. The reasoning that supports such beliefs must be at some level unconditional also, otherwise it could not show us what might, objectively, be the case.
The general aim of such reasoning is to make sense of the world in which we find ourselves and of how it appears to us and others. We proceed by generating, comparing, and ranking possible versions, and it is these comparisons that are the substance of the process. But we begin from the idea that there is some way the world is, and this, I believe, is an idea to which there is no intelligible alternative and which cannot be subordinated to or derived from anything else. My aim is to argue that even a subjectivist cannot escape from or rise above this idea. Even if he wishes to offer an analysis of it in subjective or community-relative terms, his proposal has to be understood as an account of how the world is and therefore as inconsistent with alternative accounts, with which it can be compared for plausibility.
We do not get to the idea of how the world is from the appearances; rather, we begin with that idea, since the appearances from which we start are ways in which the world appears to be. We may decide after reflection and further observation that some of these are mere appearances, that the world is not like that after all. But this always represents a modification in our view of the world, based on alternative possibilities and reasons for preferring some of them to others. What we cannot avoid is the idea that something is the case, even if we don't know what it is. Doubts about the reliability or objectivity of our perceptions and judgments have to be based on revisions of our view of the world; they cannot escape it completely. We start from certain impressions about how things are, cast doubt on the objectivity of some of them by further thoughts (including thoughts about our own na-
ture and our interaction with the rest of the world), and reject some of the appearances in favor of other beliefs about how things really are. All of it, including the observations about ourselves, is firmly embedded in a non-first-person framework of thought about the world.
But how does this generate specific methods in which one can feel some confidence? After all, the mere recognition of a distinction between appearance and reality does not supply a method of discovering reality.
The actual procedure is characterized by a high degree of cognitive inertia, and that implies that our actual worldview is to a considerable extent the expression of a perspective. We begin from a natural view of the world and are led to retreat from it by discovering that in one way or another it is inconsistent with our observations. This creates a gap that we try to fill by imagining alternative possible worlds which would, if they existed, be more consistent with what we observe. The judgments of consistency themselves involve logic, but they cannot produce logical proof of the truth of any such picture, unless it can be shown that it is the only picture consistent with the observations--which, given the fragmentary nature of the data, is probably never the case. Our attitude has to be "This could be how things are, given the evidence." The rest depends on whether there are other candidates and, if so, how we compare them.
The driving force behind all empirical reasoning is the search for order. This can take very simple forms, as when I conclude from a dog's markings that he is the same one I saw yesterday. But it leads to higher and higher levels, as we look for wider regularities behind the more specific ones we infer from observation. We don't always find order among the phenomena, but looking for it is the only way to extend our world picture and fill in the gaps between the observational data. The search for order can often lead people astray--sometimes radically so, as with astrology and other superstitions. But the remedy consists in rigorously testing faulty systems by
reference to the standard of uniformity in nature, not in giving that standard up.
I believe it is possible to understand the demand for order as a direct consequence of the idea of an objective reality, independent of particular observations and observers. Their observations may be different, but the events observed and the laws governing them must be the same. Even the idea of a single object being seen on two occasions by a single observer implies some form of natural regularity; two observers require further regularity; and the idea of an unobserved but similar event implies still more. And the process does not stop there. A thoroughly realistic conception of natural laws will have to try to interpret them also in ways that are independent of any particular point of view or observational standpoint-otherwise they might be merely ways of systematizing our observations. To be truly mind-independent, the laws--and not just the events they govern--must be perspective-free and must explain why things appear as they do from different vantage points within the world.
In modern physics, this idea constrains the development of theories through a requirement of symmetry--that the real natural order should be identified with what is invariant from the points of view of all observers, so that whatever their situation, they can all arrive at the same description of the common reality in which they are situated. The requirement applies not just to particular states of affairs but to general laws. It was this demand for symmetry or invariance in the description of nature that led Einstein first to the special and then to the general theory of relativity and that has apparently had a major role in shaping quantum theory. 1. The search for order
See Steven Weinberg, Dreams of a Final Theory ( Pantheon Books, 1992), pp. 136-47. Bas van Fraassen in Laws and Symmetry ( Oxford University Press, 1989) also emphasizes the role of symmetry in the formation of scientific theories, but his view is that their aim is only empirical adequacy, not the statement of objectively true laws of nature. Still, I take it he believes the empirically adequate theories we aim at are ones that could be objectively true.
and laws of nature seems from my amateur perspective to be driven by the broader idea that our local experiences and observations and the regularities we detect in them are manifestations of something else, something which includes us but on which none of us has a privileged perspective. Each of us is to think of our experiences as presenting us with an arbitrary or random sample of the universe. 2.
There are two potential charges of subjectivism with regard to this method. First, the demand for order cannot itself be rationally justified, nor does it correspond to a self-evident necessity, like arithmetic or logic. On a subjectivist view, the assumption of the uniformity of nature, on which science and ordinary empirical reasoning both depend, is simply the projection of our psychological need for a certain kind of world picture, rather than an intrinsically reliable tool for getting at the "mind-independent" truth.
Second, even the definition of what constitutes order seems to depend on us. For it means that at some level of description, similar causes will have similar effects on different occasions, and so forth--but the only measure of similarity we have available to us is what we count as similar, either by perception or by more technical methods of detection and measurement. 3. If that is so, then the method of arriving at factual conclusions by finding the best overall explanation or theory to account for the evidence is doubly subjective--first in its aim and second in what counts as success.
A fascinating discussion of these issues is found in Gerald Holton, "Mach, Einstein, and the Search for Reality," in his Thematic Origins of Scientific Thought ( Harvard University Press, 1988). Einstein eventually rejected Mach's phenomenalism in favor of Planck's realism. Planck described the aim of science as "the complete liberation of the physical picture from the individuality of the separate intellects" (quoted in Holton, p. 245, from Die Einheit des physicalischen Weltbildes).
This is Nelson Goodman's "new" problem of induction. See Fact, Fiction, and Forecast ( Harvard University Press, 1955; rprnt. Bobbs-Merrill, 1965).
I believe the only way to resist this charge is to argue that if those psychological analyses are taken seriously as hypotheses, they are themselves discredited by the very standards they purport to challenge. But can we make this argument without begging the question?
I think that we can and that there is an interesting difference in this respect between epistemological skepticism and the kind of subjectivism I wish to dispute. When G. E. Moore rebuts skepticism about the external world on the ground that he has two hands, he is begging the question; because if there are no material objects, then he doesn't have two hands, and he has done nothing to dispute the skeptic's argument either for the possibility that there are no material objects or for the impossibility of any evidence against its truth. A non-questionbegging refutation would have to resist the skeptic en route to his conclusion.
In arguing against subjectivism, on the other hand, one is dealing not with a proposal of mere possibilities that cannot be excluded but with a positive interpretation of our thoughts. To gain acceptance, any such interpretation must survive in competition with other claims, and that includes the thoughts being interpreted, so long as they have not been displaced. If the subjectivist does not succeed in persuading us to suspend thinking the objective content of those thoughts, he has failed--just as the skeptic has failed if he does not cause us to doubt that we have hands. That is why I believe resistance to subjectivism can come from the content of objective thoughts themselves without necessarily begging the question. It is not question-begging, provided we rely on the thoughts themselves, rather than on the second-order claim that they must be interpreted objectively.
The subjectivist proposal is not that we don't know whether our beliefs about the world are correct but that it is a mistake to interpret them as beliefs about a mind-independent natural order. Rather, they should be understood as
general features of our perspective or linguistic practice or point of view. My claim is that this is an alternative world picture--in which the central element is a set of human perspectives--and that it is in direct competition with the objective judgments it is meant to displace. Merely to propose this interpretation does not automatically make those judgments change their character. It produces instead a confrontation between two hypotheses: for example, the hypothesis that objects attract one another with a force directly proportional to the product of their masses and inversely proportional to the square of the distance between them, versus the hypothesis that it is a property of objects only as they appear to us (or in our language-game) that they attract each other with a force . . ., and so on. Unless the first hypothesis can be ruled out on some other grounds, it remains considerably more plausible than the second.
Confrontations between unqualified first-order claims and relativizing reinterpretations need not always result in victory for the former. Someone who has been brought up to believe that it is wrong for women to expose their breasts can come to realize at a certain point that this is a convention of his culture, and not an unqualified moral truth. Of course he might continue to insist, after examining the anthropological, historical, and sociological evidence, that it is wrong in itself for women to expose their breasts, and that cultures that fail to recognize the fact are in error. But this response is unlikely to survive the confrontation; it just doesn't have enough behind it (without, for example, a religious explanation of why exposure is wrong).
Unqualified judgments about astronomy, by contrast, are part of a world picture that is very robust in comparison with the Kantian alternative. Unless, as Kant thought, it is a picture that can be ruled out a priori, there is no reason why those judgments should not themselves weigh against a Kantian in-
terpretation of them. In the same way, certain first-order moral judgments can resist emotivist interpretations by their own weight.
In each case we are presented with a conflict between two conceptions of the world and our place in it. Both conceptions are incomplete in various respects. There is no neutral standpoint from which they can be evaluated, so they have to compete with one another directly. The result may sometimes be a standoff, but it is not question-begging to regard the firstorder credibility of a familiar proposition as a reason to reject a relativist or subjectivist interpretation of it. Of course one may be mistaken, but such mistakes are possible anywhere. (If two witnesses contradict each other, each maintaining that the other is lying, you can nevertheless conclude that the first is lying, on the basis of the testimony of the second; even if you are mistaken, you will not have begged the question.) There is no alternative to considering the alternatives and trying to make up one's mind.
One of the attractions of a subjectivist interpretation of empirical claims has always been that it would make radical skepticism impossible, because skepticism depends on interpreting the content of empirical claims--scientific or more ordinary-objectively, and then perceiving a logical gap between them and their empirical grounds. A recent example of subjectivism, usually presented as a way of transcending the outmoded subjective-objective distinction, is the view known as "internal realism," according to which our apparently objective world picture should be understood as essentially a creative product of our language and point of view, and the truth of our beliefs should be understood as their survival in an ideal development of that point of view. If, as Hilary Putnam has claimed,
truth is nothing but "idealized rational acceptability," 4. and if "acceptability" means "acceptability to us," then the logical gap between reasoning and the world disappears.
This position adds a qualification to our empirical claims that I believe is inconsistent with their content, in the same way that subjectivism about logic is inconsistent with its content. Furthermore, the only way to make literal sense of the qualification is in terms of a conception of the world and our place in it which is not itself subjective but according to which our entire system of substantive beliefs, by contrast, is. If we wish to adopt a view of the world that places our own thoughts within it and also answers to the demand for a natural order, it will have to be a view without such qualifications, subject to the same kind of reasoning about how things are that applies elsewhere--not a merely "internal" view.
Internal realism fails its own test of rational acceptability. What we in fact find rationally acceptable is a view of the world according to which we are located in it and arrive at beliefs about it that are confirmed and disconfirmed by our observations of what happens. Even if we concluded, as some physicists do about the quantum theory, that the best systematic account of what we observe cannot be given a realistic interpretation, that would still be a belief about how the world is, period--not a belief that it would be correct to qualify with an "internalist" reading. Reason is used to arrive at it, and the reasoning is not merely a development of our point of view, but objective thought about how things are.
More accurately, our point of view--what we accept on the basis of reason--is a set of beliefs about how things really are, together with copious acknowledgment that there is a lot
See Reason, Truth and History ( Cambridge University Press, 1981). Putnam apparently held this view about mathematics and logic as well, but here I will restrict the discussion to empirical reason.
we don't know and perhaps a lot we can never know about how they really are. Here, just as in the case of logic and arithmetic, we can't get outside of our thoughts about what is the case and think of them merely as the expression of a point of view, within which their content must be situated. Their content, including the idea of a mind-independent reality, dominates any such self-conscious psychological or social image. One might put it as follows: There is no way of determining that a belief is rationally acceptable except by thinking about whether it is true--thinking about the evidence and the arguments and being open to consideration of whatever anyone brings up as relevant. To say that its truth is its rational acceptability deprives both the notion of truth and the notion of acceptability of all content. 5.
The belief that the world is orderly, and that our sense of
In Representation and Reality ( MIT Press, 1988), Putnam asserts that internal realism is not supposed to be a reduction of truth to epistemic notions--that truth and rational acceptability are supposed to be interdependent (p. 115 ). But he doesn't make the position any clearer. On the other hand, still more recently, he seems to have edged away from the position, without actually saying so. Consider the following explanation of why Wittgenstein is not a relativist: "To say something is true in a language game is to stand outside of that language game and make a comment; that is not what it is to play a language game. Whatever it is that makes us want to replace moves like saying 'it's true' or 'it's reasonable' or 'it's warranted' by 'it's true in my language game' or 'it's reasonable in my language game' or 'it's warranted in my language game' (or makes us want to do this when we see that the language game itself is not grounded on Reason) is something that makes us want to distance ourselves from our own language game. It is as if the recognition that our language game does not have a transcendental justification made us want to handle it with kid gloves, or to handle it from a metalanguage. But why is the metalanguage any more secure?" ( Renewing Philosophy [ Harvard University Press, 1992], p. 176).
More recently still, in his Dewey Lectures, Putnam says, "Whether I am still, to some extent, an 'internal realist' is, I guess, as unclear as how much I was including under that unhappy label" ( Journal of Philosophy 91 [ 1994], p. 463, n. 41).
what constitutes order (what properties are usable in the formulation of laws and inferences) is an indication of how the world is organized, is well confirmed in some areas, where we have discovered that the hypotheses to which we are led-theories about unobservables and the laws governing them-predict observations that are not themselves explainable by our belief in those hypotheses. The fact that observation is "theory-laden" seems to me an insignificant point which in no way tends to show that the process of confirming theories by observation is circular or nonobjective. It may require some theory, of telescopes or of photography, to interpret the astronomical photographs that show the bending of light rays by the sun's gravitational field, but the crucial observation--that the images of the stars near the sun are displaced outward--is not dependent on the theory which it confirms--namely, the general theory of relativity.
The possibility of noncircular confirmation is also, I think, the answer to doubts about the role of our natural sense of similarity in determining what counts for us as a regularity or law. The fact is that we can demote a similarity or a kind to the status of mere appearance, or similarity for us, only if it is shown to be not systematically connected with other observed regularities. But if some of the regularities we observe, including those revealed by measurement, turn out to be systematically correlated with others that emerge from different types of observation or measurement, then the most plausible hypothesis is that these are not artefacts of our perspective on the world but, rather, products of the world's systematic interaction with us. The scientific image of the physical world has in this way replaced the more associative and meaning-laden picture characteristic of earlier stages in the development of our culture. As a way of understanding inanimate nature, the latter method turns out to be circular, since the only "theories" it is capable of yielding are either mere summaries of the
appearances or else delusional systems that give rise to appearances corresponding to them. 6.
Yet it has to be granted that the empirical confirmation of the supposition that the world is orderly and that particular phenomena can be explained by general laws has something inevitably circular about it. For when we formulate a law of some kind on the basis of our observations, and then confirm it by experiment, the confirmation, like the original formulation, depends on the judgment that the best systematic explanation of the relation between the original observations and the new experimental results is the one that relates them systematically--one according to which this is no accident. Someone who said at every point that the apparently lawconfirming experimental results were just coincidence would be crazy, but he would not be contradicting himself. The idea of a law-governed world is not just the idea that there is a certain system among our actual observations but that this system can be explained by an order that governs the possibilities as well as the actualities and is not directly observable. We need to rely on the same general idea both to arrive at initial hypotheses about this order and to determine whether the hypotheses have been confirmed or disconfirmed.