On Refutations of Skepticism

Citation preview

On Refutations of Skepticism Paul Horwich Noûs, Vol. 16, No. 1, 1982 A. P. A. Western Division Meetings. (Mar., 1982), pp. 56-61. Stable URL: http://links.jstor.org/sici?sici=0029-4624%28198203%2916%3A1%3C56%3AOROS%3E2.0.CO%3B2-C Noûs is currently published by Blackwell Publishing.

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/about/terms.html. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/journals/black.html. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.

The JSTOR Archive is a trusted digital repository providing for long-term preservation and access to leading academic journals and scholarly literature from around the world. The Archive is supported by libraries, scholarly societies, publishers, and foundations. It is an initiative of JSTOR, a not-for-profit organization with a mission to help the scholarly community take advantage of advances in technology. For more information regarding JSTOR, please contact [email protected].

http://www.jstor.org Tue Oct 16 04:56:49 2007

O n Refutations of Skepticism MASSACHUSE'IX INSTIlVIT OF TECHNOLOGY

Perception and realism are intimately related topics, for opponents of realism typically base their argument upon certain assumptions drawn from the philosophy of perception. First, a distinction is drawn between what can and cannot be directly perceived; second, the entities and facts whose reality is in question are located beyond the realm of immediate observation; third, all possible ways of inferring the existence of the entities are considered and found wanting; and fourth, it is concluded that we have no reason to believe in their reality. This structure is exemplified by anti-realist arguments in various domains. In this paper I would like to focus mainly on the debate surrounding scientific theories, and then attempt to extract from that discussion some morals for the issue of realism concerning the external world. T o begin with, since "rea1ism"means so many different things to different people, let me be quite explicit about the senses that I shall attach to it. I think there are three forms of realism: epistemological, semantic, and metaphysical. By epistemologzcal realism, I mean the doctrine that some specified class of entities exists. Thus one will normally be an epistemological realist about certain things (e.g. electrons, stars) and not about others (e.g. Martian Canals, the Loch Ness monster) depending on whether one does or does not believe in them, and regardless of any view about what their existence would consist in. Epistemological realism concerning X's is opposed both to general philosophical scepticism and also to any non-philosophical reason to question their existence. By semantic realism, I mean the antireductionist, anti-verificationist view to the effect that statements in a certain domain-are to be construed literally, at 'face-value', and not to be understood as shorthand for complexes of observation reports or assertions of verifiability. Finally, by metaphysical realism, I have in mind the doctrine that the concept of truth is not entirely captured by Tarski's disquotation schema ("P" is trueqf P), but involves a primitive, non-epistemic idea of correspondence with rea1ity.l

ON REFUTATIONS OF SCEPTICISM

57

In terms of this nomenclature, a phenomenalist is an epistemological realist about tables and chairs, for he does not deny that there are such things; but he is not a semantic realist, since he construes statements concerning them as covert assertions about sense-data. On the other hand, a Cartesian sceptic is a semantic realist, understanding claims about the external world at face-value; his anti-realism is purely epistemological. The general form of anti-realist argument that I sketched initially is supposed to show that beliefs in some domain are notjustified. Thus it is designed to undermine epistemological realism. Since my concern in this paper is with that type of argument and with ways of responding to it, I shall not give much more attention to the other two forms of realism; and when I use the term "realism" without qualification I shall have in mind the epistemological variety. An anti-realist with respect to scientific theories (i.e. an instrumentalist) does not believe in the underlying causal structures that are allegedly responsible for observed experimental results. He is prepared to believe facts concerning medium-sized physical objects; these facts fall in the realm of what may be known by observation; but he denies that we have good reason to assign credibility to the scientific hypotheses about unobservable phenomena that are postulated to explain what we can see.2 An anti-realist concerning the whole external world operates with an even more restrictive notion of observability. What we directly perceive are said to be only elements of our own consciousness-variously called sense-data, percepts, a n d appearances-and from these we must, but cannot, infer our beliefs about the physical environment. In response to either of these sceptical positions there are two main strategies: first, challenge the legitimacy of the presumed observation1 theory distinction; or, second, recognize the distinction, concede that the entities in question are not directly observed, but argue that their existence may be legitimately inferred. For the scientific issue, I would like to pursue an approach of the second kind, giving a prominent role to the concept of subjective probability. I shall proceed, first, by describing this approach and second by indicating why I think the alternative strategy (the one that attacks the observatiodtheory distinction) is misguided. Finally I will try to show the relevance of these remarks to the issue of ordinary realism about the external world. In a nutshell, scientific realism (taken to be the view that current scientific theories should be accepted as true) is embarassed by the almost invariable falsification of previous theories; whereas instumentalism (the idea that only observation statements are credible and that theories are just devices for their systemization) is embarassed by the difficulty of drawing a sharp principled line between what may and may not be believed. I shall suggest that the recognition of degrees of

belief provides a way out of this dilemma. Perhaps, given the sad historical record, a moderate level of confidence is the appropriate attitude towards even our favorite theories. This position accommodates the history of science (the disastrous meta-induction, as Putnam calls it), but does not require the epistemological dualism of instrumentalism. It is not uncommon in philosophy for a controversy to feed upon some misconception shared by all the major parties in the dispute. It then may seem that the problem is to decide which of the opposing positions is right. But in fact, none of them is; and the correct move is to locate and dissolve the misconception which fuels the dispute and infects all the adversaries. In the case of scientific realism versus instrumentalism, it seems to me that the dialectical energy is provided by a shared tendency to ignore gradations of belief and to presuppose that sceintific theories are simply accepted or not. This crude oversimplification naturally yields just two positions concerning the right doxastic stance towards our 'well-supported' theories: realism in favor of acceptance, and instrumentalism against it. However, both positions are objectionable. Realism comes to grief over the history of science. Few old theories are accepted today; most have turned out wrong. Consider Newtonian physics, phlogiston, the ether, conservation of mass, the wave theory of light: in time virtually every theory seems destined to bite the dust. Would it not be quite arrogant and irrational to assume that this pattern of scientific revolution will not continue, and to think that our favorite current theories will not eventually be rightly abandoned? Thus, it would be quite wrong to embrace realism and believe that our present sceintific theories are true. But if acceptance is not the appropriate attitude, what is? Instrumentalism naturally results from the assumption that our only remaining option, given the disastrous meta-induction, is to recognize that since rational belief is clearly unattainable it could not be the proper object of scientific inquiry. Rather, science may, and does, have no higher aspiration than the economical systemization of data; the constructions with which this organization is to be achieved are not designed for belief, and are not intended to correspond to underlying features of the world. This view certainly avoids the arrogant presumption of realism. But it surely errs in the opposite direction. Moreover it introduces a sharp epistemological distinction between the sentences recording observed facts and the theoretical machinery devised for their systemization; and this line is notoriously difficult to draw. Consider the sequence: ant, flea, amoeba, protein, atom, electron. Clearly any sharp line between those things accepted as real, and the fictional entities

ON REFUTATIONS OF SCEPTICISM

59

invoked for organizational purposes, will seem arbitrary and quite alien to the attitudes of scientists. The recognition of degrees of belief permits us to preserve the elements of truth in both doctrines while avoiding their flaws. There is no reason to deny that theoretical claims possess determinate truth values, no need to turn a blind eye to the history of science, and no need to impose an arbitrary and misleading distinction upon the body of scientific claims. There are theoretical discoveries in which we have a great deal of c0nfidence-e.g. that there is a planet beyond Neptune, that the species of living things evolved through natural selection, that there are units of matter, hydrogen, and oxygen, which may be combined to synthesize water. Despite the disastrous meta-induction we think it unlikely that these ideas will be overthrown. It is evident that we don't distinguish crudely between those theoretical principles we are sure of, and those we expect to be supplanted. Rather our degree of confidence varies, depending upon, among other things, the evidence, the extent to which the principle is bound up with other things we believe, and the frequency with which principles of the same type have been overthrown in the past. This line of thought concedes the observationltheory distinction, concedes the unobservability of the phenomena in question, and responds to the anti-realist by indicating how inference to theoretical beliefs may be reconciled with the disastrous meta-induction. But let me indicate why I think these concessions are in order. Philosophical objections to the observatiodtheory distinction have generally been founded on the contention that observation is theoryladen. For example, one may speak of observing the parabolic path of an electron in a cloud chamber, and in such a case the perceptual belief requires for its justification the assumption of various disputable theories concerning properties of the cloud chamber and the nature of electrons. Insofar as such theory-ladenness is characteristic of so-called observation statements, they may hardly qualify as a basis of neutral, intrinsically credible beliefs against which all other hypotheses are to be tested. I think this objection does establish that if the observatiodtheory distinction is intended to be employed in a foundationalist model of justification, the right notion of observability cannot be the everyday concept that permits up to speak of observing electrons. What the objection does not do is undermine the idea that new theories are not simply observed to be true, but have to gain their credibility by means of inference. The fact that the electron describes a parabolic path may indeed be observed, but only when the theory of electrons is already believed; and that observation cannot legitimately be used tojustify the theoretical assumptions with which it is laden.

In other words, critics of the observatiodtheory distinction may well be right in contending that there is no body of pure observational knowledge. However this does not suffice to refute the anti-realist's argument. His complaint is directed against those forms of inductive inference which are employed to justify our postulation of theoretical facts. Even if those facts may in some contexts be said to be observed. that does not remove the need for inductive inference in order to justify believing them. It is the validity of those inferences that is threatened by the disastrous meta-induction, and so any adequate response to instrumentalism must face that challenge and cannot rest solely upon criticism of the observatiodtheory distinction. Finally I would like to consider briefly the issue of realism concerning the external world and bring to bear one moral from our discussion of scientific realism. The sceptic maintains that we are immediately aware of our own experiences and have no good reason to infer the existence of any reality beyond that domain. His argument parallels that of the instrumentalist about science who also contends that there is a body of observable facts beyond which any knowledge is impossible. Now I have argued that we should not attempt to combat scientific anti-realism by pointing out that the ordinary notion of observation permits us to speak of observing the behavior of electrons and by casting doubt upon the possibility of non-theory-laden observation. This response, I maintained, is no good since it fails to address the anti-realist's concern about how such perceptual beliefs can be justified. That question is not answered simply by invoking their observational character, for this does not preclude the very real possibility of error. Similarly, I think that the Cartesian sceptic cannot be refuted by the argument that we take ourselves to directly perceive the properties of material objects. He may grant David Smith's point that the structure of everyday perceptual experiences involves a committment to independently existing objects of perception standing in certain causal relations to the experiences3; yet he may nevertheless argue that we have no reason to believe that these experiences (or their realistic committments) are veridical. The thesis that our perceptual experience presupposes realism is ambiguous. I n one sense it means that in undergoing perceptual experience we commit ourselves to realism. This is argued persuasively by David Smith; but even if true, it does not constitute evidence in favor of realism, since it doesn't have any immediate implications for the veridicality of our perceptual experience. In the other sense, to say that perceptual experience presupposes realism, is to say that it could not occur unless realism were correct. This thesis would indeed yield an argument for realism, given that the existence of our perceptual experience is not controversial. However, in this second sense, there is no clear reason why we should agree that realism is presupposed. On the

ON REFUTATIONS OF SCEPTICISM

61

contrary, it is precisely the sceptic's point that we can imagine perceptual experience the same and no external world behind it. I don't mean to suggest that the direct observability of physical objects has absolutely nothing to do with realism. Certainly it bears upon what I have called semantic realism, since it entails, contrary to phenomenalism, that statements about the external world are not shorthand for complex reports on our mental states. In addition it bears on epistemological realism, for it denies the sceptic's premiss that justified beliefs about material things would be inferred from knowledge of our experiences. Nevertheless, the direct observability of physical objects does not settle the issue in favor of epistemological realism. From the sceptic's perspective, it was required that the statements about physical reality be derivable from experience in order to be credible. In maintaining that this is not done, we merely reinforce his view that it could not be done, and we do not touch the real ground of his scepticism: namely, the possibility that all such perceptual claims are mistaken.

'See Paul Horwich, "Three Forms of Realism" Synthese, forthcoming, for a discussion of these doctrines and their relationships to other characterizations of realism. %ome instrumentalists not only oppose epistemological realism but also reject semantic realism, holding that theoretical sentences do not express propositions and are neither true nor false. %ee David Woodruff Smith, "The Realism in Perception", this volume. I should emphasize that David Smith does not himself take this point to constitute a refutation of scepticism. A view similar to Smith's was developed by Itzchak Miller in "The Phenomenology of Perception: Husserl's Account of Temporal Awareness" UCLA Doctoral Dissertation 1979.