VI
Naturalized Epistemology
Barry Stroud
The Significance of Philosophical Scepticism
Oxford University Press, 1984
The traditional Cartesian examination aims at an assessment of all our knowledge of the world all at once, and it takes the form of a judgement on that knowledge made from what looks like a detached 'external' position. I have tried to show that on the traditional conception of the philosophical enterprise and its relation to the knowledge-claims we make in science and in everyday life there is no substance in the familiar charge that scepticism violates or distorts the meanings of the very words used to express it. And I have tried to suggest how, once we become familiar with the prospect from that lofty, detached philosophical standpoint, it becomes difficult to see how anything but scepticism could be the proper verdict on our putative knowledge of the world. Scepticism can come to seem inevitable, not just invulnerable against a certain line of attack.
When we try to explain all our knowledge of the world as Descartes does we try to understand how the things we believe in science and in everyday life are connected with and warranted by the bases or grounds on which we come to believe them. All possible evidence is ultimately sensory; our knowledge of the world is empirical. But it cannot be denied that any particular course of sensory experience could fail to give us reliable information as to how things are; the world can be different from the way it is perceived to be. Within the special context of the traditional epistemological project this otherwise apparently harmless truism seems to have disastrous consequences. If all our knowledge of the world around us is in question all at once we cannot then help ourselves to some independently reliable information about the world, as we usually do, to settle the question whether our present course of experience is or is not on this occasion a reliable guide to the way things are. Once we have granted that the grounds or bases of all our beliefs about the
p.210
world are restricted to what we can get through the senses, and we have distinguished in general between everything we get through the senses and what is or is not true of the external world around us, there will be no eliminating the possibility that the external world is completely different from what we perceive it and believe it to be. The dream-possibility as it is deployed in Descartes's argument is a dramatic illustration of the point. If it must be eliminated for knowledge of the world to be possible, and it cannot be eliminated on the basis of sensory experience alone, sensory knowledge of the world is impossible.
We then appear to ourselves to be in the position of someone limited to the television screens in a locked room. If we really imagine him fully restricted to the images he can see, with no independent information about whether or not those images are generated in the normal way by the states of affairs they unquestionably represent, I think we must conclude that he knows nothing of the world outside the room. Of course it would be difficult if not impossible to contrive an actual case like this. Unless the victim had been raised in the room since birth he would already possess at least some reliable information about the outside world when he began his confinement. But from those television screens alone, with no such independent information, he could never know. If our information in everyday life were similarly restricted to what we get through the senses, so that it was always an additional step to any conclusion about the world beyond them, our knowledge would be similarly restricted to what is directly available to us and could never extend to the external world beyond.
The apparent inevitability of this conclusion can make it look as if what must be avoided is the idea that our sense experience can be understood in that Cartesian way--as providing us only with information that leaves it open whether the world around us is this way rather than that. That is how it looked to Kant; a completely general distinction between what we get through the senses and what is or is not true of the external world would cut us off forever from knowledge of the world around us. But perhaps that general distinction has fatal sceptical consequences only
p. 211
within the context of the traditional philosopher's conception of the epistemological task. Perhaps it is only when it is put to work from a detached 'external' viewpoint that the distinction between what we are given through the sense and what is true of the external world can be seen to make knowledge impossible. If so, scepticism might be avoided and our knowledge of the world made intelligible without abandoning a general distinction between 'the sensory given' and what is true of the external world. It would be a matter of avoiding or exposing as illusory that detached 'external' standpoint from which our epistemic position has traditionally seemed so impoverished.
W. V. Quine's 'naturalized epistemology' rests on the denial of any such 'external' position. Science and everyday knowledge and the languages and thought processes in which they are pursued and expressed are to be seen as natural phenomena and studied and described and explained scientifically like any other part of the natural world. That is just the empirical study of how knowledge is possible from which Kant's special a priori investigation was distinguished. But epistemology or the theory of knowledge is nothing more than the study of what knowledge is and how it comes to be. And for Quine there is no reason to suppose that the study of human knowledge or language or thought requires a fundamentally different sort of investigation from the study of physics or animal behaviour or mathematics. All attempts to find out about ourselves and the world must be made from within the conceptual and scientific resources we have already developed for finding out about anything. Even those questions traditionally regarded as especially philosophical can only be pursued from within what we now take to be our knowledge or our best hypotheses as to how things are . We have no alternative. Whatever the proper role of the philosopher might be, then, it cannot require an investigation of the world or of science or of our conceptual resources by someone who even momentarily stands outside them.
There is no such cosmic exile. He cannot study and revise the fundamental conceptual scheme of science and commonsense without
p.212
having some conceptual scheme, the same or another no less in need of philosophical scrutiny, in which to work.l
There is no special detached position from which a philosopher might conduct such inquiries.
Science, even at the most abstract reaches of theoretical physics, proceeds always 'from within'. Hypotheses and theories are evaluated and accepted or rejected in the light of what is already known or can somehow be discovered. Scientists, then, are like sailors who must repair or rebuild their ship while staying afloat on it in the open sea. There is no drydock in which they can lay a new keel and start again from new foundations; nor can they simply abandon ship and choose another of more efficient design. There is no other. This metaphor of Neurath's is Quine's favourite image for the scientific enterprise, and for him “the philosopher and the scientist are in the same boat” (WO, 3). The philosopher too is concerned with reality, with how things are, but his investigations are simply more general than geography or physics or mathematics. To determine what particular kinds of physical objects there are is the task of the natural scientist. Whether there are even prime numbers, for example, is a question for the mathematician. But the acceptance of the realm of physical objects itself, or of numbers or classes, is a question that typically falls to the philosopher. But such questions differ from the others only in “breadth of categories” (WO, 275). Philosophy is simply a more general attempt to discover the truth and advance our understanding of the world and our place in it “from within”.
It is Carnap's view that philosophical questions that appear to be about reality or what there is are really 'practical' questions to be resolved by the adoption of this or that linguistic framework for thinking about reality. Philosophy deals with words or linguistic frameworks for understanding reality, not with reality itself. It is a 'second-order' or 'meta-scientific' investigation. We saw that any such view would apparently have to countenance an inquiry or activity that takes place 'outside' mathematics or physics or the framework of ordinary spatial things; its questions would be
1 W. V.Quine, Word and Object, pp. 275-6. (Hereafter cited as WO.)
p. 213
“external” to all such frameworks. We found difficulty in understanding the “externality” of Carnap's questions and in accepting the associated idealist thesis that no facts hold or fail to hold independently of our adopting this or that linguistic framework. And there was the further difficulty of identifying the linguistic framework, if any, to which that thesis itself belongs. Quine's conception of philosophy as continuous with the rest of science would avoid all those obscurities.
He does not deny that ontological questions traditionally regarded as philosophical seem to be more about words or about our conceptual framework than about extra-linguistic reality. They also seem to be governed more by pragmatic considerations of convenience, simplicity and overall conceptual economy than by current matters of observable fact. As we move from asking whether there are mountains in Africa or unicorns anywhere to asking whether there are numbers or propositions or physical objects we seem to move to a different sort of question. Carnap thought it was the kind of question that cannot be settled by observation or theoretical considerations at all.
Carnap's conclusion ought to be resisted, according to Quine. It is true that in philosophy it is usually more profitable to talk about the terms and frameworks we use to understand reality than to talk directly about reality itself. For Quine that is simply because progress on ontological philosophical issues is more likely if the participants engage in “semantic ascent” and discuss the theoretical efficacy of their terms by mentioning them in the “formal mode” rather than simply using them in the “material mode” to talk directly about what does or does not exist. Such direct use obscures the theoretical character of ontological disputes. But semantic ascent from the “material” to the “formal” mode is possible everywhere; it does not apply uniquely to philosophy or even to the most abstract levels of discourse. Nor does the possibility of “semantic ascent” show that assertions or questions that mention words rather than using them to talk directly about reality are really only about words or linguistic frameworks and not about the reality they are used to describe. For example, what we regard as the
p. 214
empirically known contingent truth that there are wombats in Tasmania can be paraphrased in the “formal mode” as , ‘“Wombat” is true of some creatures in Tasmania’, but that does not transform it into an assertion solely about language and not about extra-linguistic reality (WO, 272).
On Quine's view, Carnap's conception of philosophical questions as linguistic and as resolvable only on pragmatic and never on factual grounds arises partly from a misreading of the significance of semantic ascent.
For it is not as though considerations of systematic efficacy, broadly pragmatic considerations, were operative only when we make a semantic ascent and talk of theory, and factual considerations of the behavior of objects in the world were operative only when we avoid semantic ascent and talk within the theory. Considerations of systematic efficacy are equally essential in both cases; it is just that in the one case we voice them and in the other we are tacitly guided by them. (WO, 274.)
This is not a rejection of Carnap's stress on pragmatic considerations such as convenience or simplicity of theory to philosophical questions; it is rather a reminder of the importance of such factors to all investigations into what is the case, and hence a denial of the distinction Carnap would draw between philosophy and everything else.
Kant also would distinguish between philosophy and everything else, and Quine's “naturalizing” of philosophy obviously stands equally opposed to the Kantian idea of a special a priori philosophical investigation, and indeed to the belief that there is any a priori knowledge at all. But Kant's and Carnap's views, for all their obscurities, were meant to explain how traditional scepticism about the external world is to be avoided. They both acknowledge what called the conditional correctness of scepticism:' if the traditional philosopher had managed to raise a meaningful theoretical question about the external world his sceptical answer to it would be correct. Quine rejects the Kantian and Carnapian accounts of philosophical problems and insists on the 'scientific' or 'theoretical' character of the question about external physical objects and our knowledge of them. Does he thereby avoid the scepticism that Kant and Carnap (and Descartes for that matter) would argue is
p. 215
then inevitable? Does Quine's naturalized or scientific epistemology give a satisfactory answer to the very question Kant and Carnap despaired of answering directly and so developed a special philosophical theory to explain away? Does Quine even try to answer that very question? It is not
easy to say.
Many things he says about his conception of epistemology make it sound as if it is meant to answer the very question the traditional philosopher found himself faced with. “Given only the evidence of our senses”, Quine asks, “how do we arrive at our theory of the world?”2 The problem arise because “we know external things only mediately through our senses”; “physical things generally, however remote, become known to us only through the effects which they help to induce at our sensory surfaces” (WO, 1). Here we have what looks like a completely general problem--how do we come to know anything at all about external physical things?-- which is to be answered by an explanation of how what we get through the senses provides us with the knowledge we want to explain. Relative to what we claim to know about the world around us, Quine says, our sensory 'input' is 'meager'. That is what gives rise to the problem--to explain how the human animal could have arrived at “a description of the three-dimensional external world and its history”3 from “the sensory information that could reach him” (RR, 2) at his sensory surfaces. This sounds in many respects just like the problem Descartes leaves us with at the end of his first Meditation: how, on the basis of what we are immediately aware of in perception, can we ever come to know things about the world around us?
The object of Quine's study is the relation between those sensory stimuli and the knowledge to which they eventually give rise, or “the relation between the meager input and the torrential output” (EN, 83). But since that relation is itself part of the world around us it is to be studied like any other natural phenomenon. We can observe and experiment with
2 W. V. Quine, The Roots of Reference (LaSalle, lllinois, 1974), p. 1. (Hereafter cited as RR.)
3 W. V. Quine, 'Epistemology Naturalized', in his Ontological Relativity and Other Essays (New York, 1969), p. 83. (Hereafter cited as EN.)
p.216
human beings, while making use of any parts of current natural science that we think might be helpful. The question is how our science or our knowledge of the world has come to be, and the answer is to be found by pursuing that very science whose origins we seek to understand. There is no alternative. Epistemology for Quine must be seen as part of natural science-'naturalized epistemology' is the only epistemology there can be.
One respect in which Quine's conception of his philosophical task seems to resemble that of the traditional epistemologist is his completely general distinction between everything we can get through the senses on the one hand and what is or is not true of the external world on the other. That distinction is essential to the formulation of the problem Quine thinks a naturalized epistemology should answer, just as it is to the traditional problem of the external world. But for Quine the distinction is itself derived from scientific investigation and reflection.
we can investigate the world, and man as paxt of it, and thus find out what cues he could have of what goes on around him. Subtracting his cues from his world view, we get man's net contribution as the difference. This difference marks the extent of man's conceptual sovereignty-the domain within which he can revise theory while saving the data. (WO, 5.)
We find by investigation that the 'data' that can be 'saved' are 'meager' relative to the scope of man's 'conceptual sovereignty'. We thereby discover the extent to which all of science is man's 'free creation' or even, in Eddington's phrase, 'a put-up job' (RR, 3-4). By 'science' here Quine means everything we take to be true, including all truths about the external world. What we find when we study man's position in the world is how all those things he believes about the world go far beyond the 'information' or 'data' he gets through his senses.
It is a matter of scientific fact, or theory, that our only avenue of information about external objects is through the irritation of our, sensory surfaces by forces emanating from those objects. There is thus a wide gap between our data and our knowledge of the external world, and it takes bold inference to bridge it.4
4 W. V. Quine, "The Natural Theory off Knowledge', unpublished, 1979, p. 2.
p.217
The same gap between 'our data' and 'our knowledge of the external world' that gives, rise to ,the traditional problem seems to be present here, but for Quine the fact that our theory of the world far outstrips its 'sensory or stimulatory background' (WO, 3) is itself a deliverance of science, part of that very theory of the world. So much the better, it would seem, for that distinction, and for the epistemological enterprise itself. The problem a naturalized epistemology must answer is thrown up by the very science whose origins it is meant to explain.
Once we see the epistemological problem in this way, a certain kind of answer to it naturally suggests itself. Of course we do not have all the details of a solution, or even a clear idea in every case of what to do in order to discover them. The scientific study of perception, learning, language acquisition, and the development and transmission of human knowledge is bound to continue in directions and with methods we cannot at the moment clearly foresee. But for Quine the general outlines of the story are clear enough to give us a very abstract and schematic, but nevertheless illuminating, account of human knowledge. At the level of generality appropriate to philosophy we can explain how human knowledge of the external world is possible.
Our knowledge of external things in general is to be understood in just the way any piece of theoretical knowledge is to be understood relative to the 'data' on which it is based. In order to explain some of the things that happen to ordinary perceivable things, a physicist might invent or appeal to a theory committed to unperceived or even unperceivable objects as a way of introducing greater simplicity and economy of basic principles into his total account of the physical world. The truths he introduces and accepts about molecules, for example, or other extraordinary objects, will not be uniquely determined by all the truths he knows or even can imagine about the ordinary perceivable objects whose behaviour his wider theory is meant to explain. There could be many different possible theories, even perhaps equally simple theories, that could be used to imply and therefore to account for the same set of truths about, ordinary perceivable things. He is not forced by those 'data'
p.218
alone to introduce molecules as a simplifying explanatory device. But the theory he does introduce or appeal to is accepted and retained because it does its job. Relative to those truths about ordinary things, the existence of the molecules is therefore a 'hypothesis' or 'posit'; it does not follow from those truths alone, but it is asserted by a theory which, with other facts and theories, can be shown to imply those ordinary observable truths.
For Quine the existence of physical objects in general is to be understood as a 'hypothesis' or 'posit' in just the same way. It is not a 'hypothesis' or theory relative to a set of truths about ordinary perceivable things, of course--such truths already imply the existence of physical objects. Rather all statements even about ordinary physical things are to be understood as 'hypotheses' or theoretical statements relative to what we get through the senses. They are all 'far in excess of any available data' (WO, 22). Science tells us that the only information that reaches us through the senses is provided by irritations at our sensory surfaces, but all the truths we believe about the physical world around us 'are less than determined by our sensory irritations' (WO, 22), just as the physicist's assertions about molecules are less than determined by observable truths about ordinary physical things. And in the case of our beliefs about the physical world in general it is not simply a matter of our limited experience. The underdetermination would remain 'even if we include all past, present, and future irritations of all the far-flung surfaces of mankind' (WO, 22). The theory or 'hypothesis' of physical objects remains far in excess of all' such data. Seen in relation to our sensory surface irritations it is therefore just like the theory or hypothesis of molecules seen in relation to truths about ordinary perceivable things. The only important difference for Quine is that:
the physicist audibly posits [molecules] for recognized reasons, whereas the hypothesis of ordinary things is shrouded in prehistory. Though for the archaic and unconscious hypothesis of ordinary physical objects we can no more speak of a--motive than of motives for being human or mammalian, yet in point of function and survival value it and the hypothesis of molecules are alike. (WO, 22.)
p.219If we can understand how a physicist can come to know there are molecules we can also understand in the same way how we can come to know there are any physical objects at all.
What is the 'function and survival value' that the two sorts of hypotheses are said to share? Precisely that of providing a simpler and more economical total 'theory' while accounting for the 'data' on which that 'theory' is based. Since it is 'underdetermined' by all the 'data', the choice of theory is not uniquely determined, but Quine has long found that the 'hypothesis' of physical objects has succeeded in the task for which, however unconsciously, it was designed. It 'has proved more efficacious than other myths as a device for working a manageable structure into the flux of experience';5 it gives us 'the smoothest and most adequate overall account of the world' (WO, 4). The origin of the physical object theory is 'shrouded in prehistory' in that it is embodied in the languages spoken by human beings from time immemorial. In acquiring the language of our community each of us gradually becomes master of the mechanisms of objective reference by means of which external physical things can be spoken of, and the irritations we inevitably undergo at our sensory surfaces then dispose us to believe and assert things about an objective physical world. Thus do we come to know of external things.
The scientific character of this sketch of an explanation--or the scientific character of the study of the relation between our sensory impacts and our subsequent theory of the world--is the key to understanding a fully naturalized epistemology. Not only does the problem or task of the enterprise arise from within science, its solution is to be sought and found there too. Given his quite general philosophical purposes, Quine himself is more concerned with recommending and sketching the outlines of a naturalized epistemology than with carrying it out in detail. Beyond mentioning the 'two dimensional optical projections and various impacts of air waves on the eardrums and some gaseous reactions in the nasal passages and a few kindred
5 w. V. Quine, 'Two Dogmas of Empiricism', in his From a Logical Point of View (Cambridge, Mass., 1953), p. 44.
p.220
odds and ends' (RR, 2) we get through the senses, he scarcely goes into the physiological or psychological facts. But the project he recommends is nevertheless a scientific project which 'may be pursued at one or more removes from the laboratory, one or another level of speculativity' (RR, 3). His own speculations on the subject tend to concentrate on language-acquisition, and that is an observable phenomenon in the world, however abstractly we might think about it. We know that a child acquiring a language from his elders has nothing but the bombardments of his sensory surfaces to go on, and he ends up speaking, as they do, of objective physical things. A genetic understanding of how that competence comes about could then provide us with an understanding of the relation between our sensory impacts and our theory of the world. That, for Quine, would be to understand the relation between 'observation' and 'scientific theory'.6 That is why he regards the theory of language as vital to the theory of knowledge.
Quine's emphasis on the empirical, scientific character of the problem of our knowledge of the external world , makes' it look as if his project or question is not the same as the problem that exercised the traditional philosopher. It is true that for Quine the problem arises because there are many possibilities compatible with everything the senses provide. Since we could adopt anyone of several different theories based on the same 'data', the problem is to explain how we know that there are external objects, how we know that the 'physical object theory' is the right one. Quine says we arrive at that theory, or at any rate that we continue to believe it, on grounds analogous to those on which any scientist accepts a theory that goes beyond his evidence. ‘The last arbiter is so-called scientific method, however amorphous’ (WO, 23). But could we hope to answer the traditional philosopher's question about the external world simply by emulating even the best scientific procedures? Scientists explicitly engaged in theory construction do not normally even consider, let alone justifiably rule out, the kinds of possibilities brought up in generating
6 W. V. Quine, ‘The Nature of Natural Knowledge’, in S. Guttenplan (ed.), Mind and Language (Oxford,1975), p. 74. (Hereafter cited as NNK.)
p.221
the traditional problem. An experimenter does not establish that he is not dreaming when he puts a certain chemical theory to a test. If we ordinary non-scientific mortals arrive at our view of the external world by inference of the same general kind, we will not consider or try to eliminate such bizarre possibilities either. So much is borne out by Quine's account of the genesis of our theory of the world. Nowhere in his story does he explain how we eliminate the possibility that our sensory data are merely the products of a dream or of an evil demon or of some other source incompatible with the physical object 'hypothesis'. That suggests, that whatever Quine's naturalized epistemology is meant to do it could not answer the very question that proved so difficult to the traditional epistemologist. The justified elimination of possibilities incompatible with knowledge of the physical world is precisely what was in question in the traditional problem.
Another apparent difference is that Quine's question about our knowledge is to be answered by making use of any scientific information we happen to possess or can discover, whereas the traditional epistemologist's question was meant to put all that alleged information into jeopardy and hence to render it unavailable for such explanatory purposes. Any question empirical science can answer could not be the traditional philosopher's question. That is not to say there can be no such thing as a science of human knowledge, but only that any such 'internal' investigation, however feasible, could never be expected to answer the traditional question. That is precisely Carnap's reaction. He agrees that there can be an empirical scientific study of human beings and of how they come to know the things they do, and he would have no objection to Quine's naturalized epistemology so conceived. But precisely because of its empirical, resolvable character, the question could not be the same as the traditional philosophical question. That question is philosophical and meaningless for Carnap because no evidence could help to settle it. No one 'theory' is made more credible than any other on the information said to be available to us.
Quine sometimes seems to acknowledge the traditional question and to agree that the sceptical answer to it is correct.
p.222
He finds empiricist philosophers of the past concerned with the relation between sensory data and beliefs about the external world in two different ways. There is the 'conceptual'. question of whether statements about external physical objects can be fully expressed or reformulated without loss in purely 'sensory' terms, and there is the 'doctrinal' question of whether our knowledge of external physical things can be adequately justified on the basis of purely 'sensory' knowledge (EN, 69-70). On this 'doctrinal' question of justification he finds us no farther along today than where Hume left us. 'The Humean predicament is the human predicament' (EN, 72). By that he presumably means that our beliefs about bodies are not justified by our sensory data .
There are apparently two reasons for this despairing verdict. The first is that all general statements or statements about the future (and also presumably the past), even if they could be expressed in purely 'sensory' terms, could not be known with certainty on the basis of present sense experiences. They go beyond what is true of actual impressions. The other reason, at least for Hume, was that any justification our sense experiences could give to statements that go beyond them would have to come from the reliability of inductive or non-demonstrative inference. Any such inference would take us from what has been experienced to what has not, and the principle of an inference of that kind could not itself be a report only of actual sense experiences. Even if we could explain 'scientifically' why we make the kinds of inferences we do, it would not follow that we had answered the traditional 'doctrinal' question of whether the conclusions of our inferences are justified by their premisses. That is because the scientific information we appeal to in our explanation would itself have been arrived at by just such an inference.
Quine at times seems to grant the vicious circularity in any 'scientific' attempt to justify inductive inference in that way. He thinks the Darwinian theory of natural selection, for example, might help explain why induction works so well for those of us survivors who reason inductively; we are descendants of those with well-adapted 'similarity standards'. But
p.223
that biological explanation could not justify induction. 'This would be circular,' Quine concedes, 'Since biological knowledge depends upon induction' (NNK, 70). The same strictures against circularity determined the traditional project. If all our knowledge of the external world is in question all at once, no part of that putative knowledge can be appealed to help explain how we know the rest; all of our knowledge is to be justified on the 'sensory' basis alone. Quine grants that:
a surrender of the epistemological burden to psychology is a move that was disallowed in earlier times as circular reasoning. If the epistemologist's goal is validation of the grounds of empirical science, he defeats his purpose by using psychology or other empirical science in the validation. (EN, 75-6.)
Quine's own naturalized epistemology is precisely 'a surrender of the epistemological burden to psychology' (and to any other sciences that might help us understand human knowledge). If that does not defeat Quine's purpose, although it would defeat the traditional epistemologist's purpose, is that because the question Quine asks and answers is different from the traditional question? What then is the relation between them?
At times he suggests that the questions are different. The 'doctrinal' issue of justifying our knowledge of physical bodies in purely 'sensory' terms, he suggests, should be abandoned as a vain hope. We can then concentrate on the manageable scientific project of understanding the relation between 'observation' and 'science', between the 'meager input' at our sensory surfaces and the 'torrential output' that embodies our theory of the world. Because the traditional epistemologist was bent on validating or justifying our knowledge of the world, he insisted on isolating certain objects of awareness in sense-perception. He wanted to identify the indubitable information we could be said to acquire in perception so that he could pose more sharply and more precisely the question of how that information could ever justify our richer beliefs about an external world. But once the project of justification is abandoned, Quine thinks, we can sidestep the issue of awareness and simply try to
p.224
explain how our torrential theoretical output arises from those events that take place at our sensory surfaces. And in seeking or providing that explanation we are obviously free to use any scientific information we happen to have or are lucky enough to get. An understanding of the relation between two sorts of events in the observable physical world is now our scientific goal, not the hopeless extra-scientific project of somehow supporting our rich theory of nature on the basis of some strange entities that we find we are, strictly speaking, aware of in perception.
This now makes it look as if Quine is simply changing the subject, or recommending a different subject from the one that interested the traditional epistemologist. That would leave open the possibility, sometimes apparently endorsed by Quine, that scepticism is and remains the only answer to the traditional question, and that nothing he says in his naturalized epistemology affects that answer one way or the other. But this accommodation Quine also wants to resist.
In The Roots of Reference he denies that the 'liberated epistemologist' who now marches under the banner of empirical psychology has changed the traditional subject; his 'is an enlightened persistence rather in the original epistemological problem' (RR, 3). The enlightenment comes from recognizing something the traditional philosopher missed. He thought the appeal to any or all of empirical science to explain how our empirical science of the world is possible would be circular reasoning, so no validation or justification of our knowledge of the world could come out of such an appeal. But we can now see, according to Quine, that 'this fear of circularity is a case of needless logical timidity, even granted the project of substantiating our knowledge of the external world' (RR, 2). A liberated naturalized epistemology appears capable of giving us what the timid traditional philosophy despaired of finding. This new understanding of epistemology is 'enlightened':
in recognizing that the skeptical challenge springs from science itself, and that in coping with it we are free to use scientific knowledge. The old epistemologist failed to recognize the strength of his position. (RR,3.)
p.225
What the traditional epistemologist failed to recognize, according to Quine, was that the challenge he raised against our knowledge of the world came from that very knowledge itself. The reasons he had for finding knowledge problematic or doubting its reliability were scientific reasons. If he had recognized the real source of his doubts he would have recognized the strength of his position and the fact that he can use science in answering the doubts he has raised.
Here we return to the important point, mentioned earlier, that the problem the theory of knowledge must answer is thrown up by that very knowledge whose origins or possibility it is meant to explain. The theory of knowledge for Quine as for the tradition has its origin in doubt and the threat of scepticism.
Doubt prompts the theory of knowledge, yes; but knowledge, also, was what prompted the doubt. Scepticism is an offshoot of science. (NNK, 67.)
The 'strength of his position' that Quine is pointing out to the traditional epistemologist is the availability of scientific knowledge for answering his question. However difficult or complicated the investigation might prove to be, it is a scientific pursuit like any other, so we are in no worse position for explaining our knowledge of the world than for explaining any other natural phenomenon.
Quine thinks the confusion centered on the issue of awareness. Older epistemologists thought the facts about the meagreness of our sensory data were discovered by direct introspection or perhaps by simply attending carefully to what is given in perception. But in fact the reasons for finding our data meagre and hence knowledge problematic came from science itself. Knowledge of the world and of how what we perceive can deviate from it were needed as a springboard for the scepticism epistemology then tried to avoid. The suggestion is that because the traditional epistemologist failed to recognize this important fact that 'sceptical doubts' are really 'scientific doubts '-he failed to recognize that he was in the strong position of being able to use his science to answer those doubts and explain how scientific knowledge is possible.
p.226
This is both a diagnosis of the traditional philosopher's quandary and a defense of Quine's scientific epistemology as 'an enlightened persistence in the old epistemological problem'. Not only is naturalized epistemology all we can have; it is all we need or ever needed. Having stressed the fact that the posing of the epistemological problem depends on accepting certain results of natural science, Quine immediately concludes, 'Epistemology is best looked upon, then, as an enterprise within natural science' (NNK, 68). He is not simply saying here that epistemology is best looked upon as an enterprise within natural science, as he might do if he were tired of the interminable disputes of philosophers and thought it better (or 'best') to concentrate on psychology or physiology instead. His 'then' indicates that his scientific conception of epistemology is meant to be supported by what has just been said. What has just been said is that 'sceptical doubts are scientific doubts', that the epistemological problem arises within science itself.
The inference is even more strongly suggested by Quine's talk of the 'crucial logical point' he thinks the traditional epistemologist missed.
The crucial logical point is that the epistemologist is confronting a challenge to natural science that arises from within natural science. The challenge runs as follows ... [Here is a description of our meagre sensory irritations] ... How, the challenge proceeds, could one hope to find out about that external world from such meager traces? In short, if our science were true, how could we know it? Clearly, in confronting this challenge, the epistemologist may make free use of all scientific theory. (RR, 2.)
The traditional epistemologist did not think it was clear that he could make free use of scientific theory; he thought it was clear that he could not. But if the 'logical point' about the scientific origins of his challenge to science is the only thing he missed, the implication is that ('clearly') scientific knowledge can be used in meeting that challenge precisely because the challenge arises within science itself.
Many philosophers have been tempted by an even stronger conclusion from the scientific origins of the epistemologist's doubts. Those doubts are typically based on the possibility of illusion. But if, as Quine says, 'the concept of illusion
p.227itself rested on science, since the quality of illusion consisted simply in deviation from external scientific reality' (RR, 3), it might look as if we can conclude immediately that no completely general sceptical doubt about all our knowledge can ever be reached. For Quine, 'illusions are illusions only relative to a prior acceptance of genuine bodies with which to contrast them', and 'bodies have to be posited before there can be a motive, however tenuous, for acquiescing in a noncommittal world of the immediate given' (NNK, 67). This easily seems to lead to the conclusion that since some knowledge of science is needed even to understand what an illusion is, it is impossible for an appeal to the possibility of illusion to undermine all our scientific knowledge all at once. It would pull out from under us the very support we originally needed to get that undermining project going in the first place. Whatever got undermined or thrown into doubt, it would seem, could therefore not be all of our science all at once. Or so the argument would run.
I have mentioned arguments of this general type earlier. They are one species of the criticism that the sceptical epistemologist could reach his general conclusion only by distorting the meanings of its terms or by violating the conditions necessary for those terms to mean what they do. The argument would see the dependence of what Quine here calls 'the concept of illusion' on some unquestioned knowledge of external reality as part of the meaning, or a condition of the meaningfulness, of the notion of illusion. In Chapter Two I expressed some general doubts about the prospects of success along these lines. But whatever the argument based on meaning or meaningfulness might be, it is clear that Quine is not making it. His view of language and his rejection of the philosophical use of synonymy or analyticity leave him in no position to appeal to what is or is not included in the meaning of a particular term. It is one of the merits of Quine's views about language that they do not support such dubious argumentation. But if he does not think the scientific origins of the epistemologist's doubts lead the sceptic to outright contradiction or self-refutation, why does he think that because 'sceptical doubts are scientific doubts' the
p.228
epistemologist is 'clearly' free to use empirical science in answering them?
The question is made more difficult by Quine's explicit disavowal:
I am not accusing the sceptic of begging the question; he is quite within his rights in assuming science in order to refute science; this, if carried out, would be a straightforward argument by reductio ad absurdum. I am only making the point that sceptical doubts are scientific doubts. (NNK, 68.)
This is an important concession, and amounts to a very powerful point in the traditional philosopher's defence. If there is nothing logically peculiar or self-defeating in starting with some scientific knowledge and ending up by rejecting or doubting it all, what becomes of 'the crucial logical point' that the traditional epistemologist is said to have missed? If the 'only' point Quine is making is that 'sceptical doubts are scientific doubts', does it follow that epistemology, 'then', is part of natural science, and that 'clearly' the epistemologist may make free use of all scientific theory? Once it is granted that the sceptic might be arguing by reductio ad absurdum, I think it does not follow.
The reductio ad absurdum would presumably run something like this. Either science is true and gives us knowledge or it does not. If it is not true, nothing we believe about the physical world amounts to knowledge. But if it does give us knowledge; we can see from what it tells us about the meagre impacts at our sensory surfaces during perception that we can never tell whether the external world really is the way we perceive it to be. But if that is so, we can know nothing about the physical world. So once again nothing we believe about the physical world amounts to knowledge. On either possibility we know nothing about the physical world.
I do not suggest that this is itself the sceptical argument. The discovery of the meagreness of our sensory data and the impossibility of their supporting our beliefs about the world would be reached on the second horn of the reductio's dilemma only after fairly elaborate reflections of the kind I outlined in Chapter One. The present question is only whether someone whose sceptical reasoning is understood
p.229
as falling within this general reductio pattern would then be in a position to use part or all of his scientific knowledge of the world to show how knowledge is possible after all. It seems to me clear that he would not. That is why Quine's, concession that the sceptic can be understood as arguing by reductio ad absurdum seems to me to count so strongly in favour of the traditional philosopher's understanding of his question. If I am right, the fact that 'sceptical doubts are scientific doubts' does not put the epistemologist who raises such doubts in the stronger position of being free to use scientific knowledge of the world in his effort to answer those doubts and explain how knowledge is possible.
Suppose we ask, as Descartes does, whether we know anything about the world around us, and how any such knowledge is possible. And suppose we ask this question and find an answer to it difficult because of certain things we take at the outset to be true about the physical world and about the processes of perception which give us the only access we have to it. If we then reasoned as Descartes reasons and arrived by reductio ad absurdum at the conclusion that we know nothing of the physical world, and we found ourselves dissatisfied with that conclusion, clearly we could not go blithely on to satisfy ourselves and explain how knowledge is nevertheless possible by appealing to those very beliefs about the physical world that we have just consigned to the realm of what is not known. By our own arguments, despite their scientific origin, we would find ourselves precluded from using as independently reliable any part of what we had previously accepted as knowledge of the world around us. The scientific origin of our original question or doubts would therefore do nothing to show that the answer to our question or the resolution of our doubts can be found in an empirical study of human knowledge as an observable phenomenon in the physical world.
This says nothing about the independent desirability or feasibility of an empirical study of the psychology and physiology of perception, learning, and language-acquisition. Nothing I have said about traditional epistemology is meant to cast any aspersions on that. What is in question is only the relation between Quine's project, and the traditional
p.230epistemological enterprise. Nor would I cast aspersions on the confident everyday assertions of G. E. Moore. In fact I tried to remove philosophically-motivated aspersions cast by others. In that case too the only question was the relevance of Moore's assertions to what we recognize to be the traditional project. On views like those of Carnap and Kant, what Moore says is perfectly legitimate and unassailable, but it does nothing to settle the philosophical issue one way or the other. The results of an independently-pursued scientific explanation of knowledge would be in the same boat. They would be 'scientific' versions of Moore's 'common sense' remarks. But if we feel that the philosophical question is not and could not be answered directly in Moore's simple way (as I think we do), we should also find that it cannot be answered by apparently more scientific assertions to the same effect. The scientific story is not more true or more highly confirmed or more clearly based on experience than what Moore says; it is just more complicated. For Quine 'science is self-conscious common sense' (WO, 3).
Quine does not accept the Kantian or Carnapian conception of the philosophical enterprise, so their appeal to the isolation of the philosophical from the scientific is not open to him. He rejects the possibility of an 'external' position from which all of our knowledge of the world can be seen whole; for him philosophy is continuous with science. Descartes and other traditional philosophers could accept that continuity--at least they understand their philosophical assessment of our knowledge to have the sceptical consequence that what we thought was scientific knowledge of the world is not really knowledge at all. We have no more reason to believe it than to disbelieve it. That is something no philosophical investigation could show, according to Carnap and Kant. But for Descartes science and our commonsense view of the world can come under general philosophical attack, and if the argument can be seen in Quine's way as a reductio ad absurdum, that attack will be seen as coming 'from within'. Even if the doubts that lead to the eventual rejection of our science of the world are themselves 'scientific doubts', we reach a final position in which nothing about the external world counts as knowledge or reliable belief, including
p.231
the very beliefs that helped generate the doubts in the first place. Given the possibility of a reductio, Quine's repudiation of an 'external' detached position for the assessment of knowledge would not in itself guarantee the impossibility of epistemological scepticism. A sceptical challenge 'from within' would be possible, and the knowledge thereby repudiated could not be appealed to to meet the challenge.
What then is wrong with scepticism according to Quine? How can it be avoided? It once looked as if recognition of 'the crucial logical point' that 'sceptical doubts are scientific doubts' would be enough to 'liberate' the traditional philosopher from his gloomy sceptical conclusion. But if that alone does not legitimize a 'scientific' answer to what remains a real question, the problem is still with us.
In the face of that apparent problem Quine is content to stress his 'naturalism', the idea that in his reflections on knowledge he is:
reasoning within the overall scientific system rather than somehow above or beyond it. The same applies to my statement ... that 'I am not accusing the sceptic of begging the question; he is quite within his rights in assuming science in order to refute science.' The skeptic repudiates science because it is vulnerable to illusion on its own showing; and my only criticism of the skeptic is that he is overreacting.7
Scepticism about the external world is not incoherent on Quine's view. He defends the sceptic's right to put it forward ('quite within his rights'). He criticizes it only as a form of extremism.
What is the sceptic's over-reaction, according to Quine? It might look as if it is the panicky response of rejecting science completely and never making use of it again simply because it is vulnerable to illusion on its own showing. That would be an over-reaction, like getting rid of my car or never trusting it to start again simply because it failed once in freezing weather on a high mountain. But I argued that the sceptical reasoning does not turn directly on the simple fact that illusions sometimes occur. That alone does not imply that we know nothing about the world around us. The
7 W. V. Quine, 'Reply to Stroud', Midwest Studies in Philosophy, vol. VI, '1981, p. 475.
p.232
sceptical conclusion comes only with the realization that everything we get through the senses is compatible with countless different 'hypotheses' about what is the case beyond those sensory data, so there is no way of telling which of the many different possibilities actually obtains. If that is the position we are in, it is no over-reaction to conclude that we can know nothing about the world around us. That would be the only reasonable reaction to such a plight. If on the basis of what I see I cannot tell whether the bird in the garden is a goldfinch or a goldcrest or a canary it is far from an over-reaction to conclude that I do not know it is a goldfinch.
But Quine suggests that the sceptic is over-reacting because we do not at the moment actually have good reason to reject science on the sceptic's grounds. He grants that:
Experience might, tomorrow, take a turn that would justify the sceptic's doubts about external objects. Our success in predicting observations might fall off sharply, and concomitantly with this we might begin to be somewhat successful in basing predictions upon dreams or reveries. At that point we might reasonably doubt our theory of nature in even its broadest outlines. But our doubts would still be immanent, and of a piece with the scientific endeavor.8
This suggests that the sceptical 'theory' is not yet as well-confirmed as some other views. Perhaps it will become so, but for the moment it lacks sufficient justification.
What Quine calls scepticism here is a far cry from the position reached at the end of Descartes's First Meditation. In invoking the dream-possibility and arguing that there is no way we can eliminate it, Descartes is not suggesting that we should base our predictions on dreams or reveries rather than on what we are pleased to call scientific observation and experiment. And in repudiating science as a source of knowledge of the world he is not announcing that our success in 'predicting observations' is going to fall off if we keep reasoning and theorizing scientifically as we have been doing, and that we should look to our dreams instead. Quine speaks of future experience as perhaps justifying the sceptic's doubts about external objects, as if those doubts are not sufficiently justified at the moment. But whether scepticism is the correct
8 'Reply to Stroud', p. 475.
p.233
answer to the epistemological question is not something to be settled by further observation or experimentation. If the question is posed correctly--as Quine himself poses it--we already know that whatever future experience might be like, it can only give us more of what will remain laughably meagre sensory data relative to our rich set of beliefs about the world around us. We will always be faced with the question of whether we have any more reason for adopting the physical object 'hypothesis' rather than anyone of a hundred others that equally go beyond all possible data.
The sceptical view that is a response to that question does not itself take a stand on what is actually the case beyond the data--on which one of the many 'theories' we should adopt--so we do not need to wait for further evidence to see whether the sceptical view becomes more, or less, worthy of our acceptance. If we are restricted to data which far underdetermine what we believe, the sceptical doubts are justified today--in fact, they were justified in that same way in the 1630s. It is not a question of more experience. Scepticism does not say that current science is not knowledge of the world but something else (say, reveries) really is instead. It simply says that none of the competing 'hypotheses' about what is true beyond the data can be known to be true; in fact, that we can have no more reason for believing anyone of them rather than others on the basis of the only sensory data we can ever have. If our data are so inevitably restricted in relation to what we claim to know on the basis of them, the conclusion that we can know nothing beyond the data is no over-reaction at all.
Kant and Carnap, in different ways, both see the sceptical potential in the traditional question. That is why they concentrate on the question itself and on how something must go wrong in the way it gets raised. Quine eschews such purely 'philosophical' or diagnostic activities. He finds sufficient reassurance in the idea that the epistemologist, like everyone else, must operate from within the accumulating body of theory we find ourselves constantly assessing, revising, expanding, and trimming as we are carried along by it. We rebuild and repair our ship on the open sea . But if the sceptic can be seen as arguing by reductio to the conclusion
p.234
that all of that science nevertheless provides no knowledge of the world, the consolations of naturalism alone will not be enough. It would leave us with no way of ensuring that the greater part of our ship, insofar as it represents what we know or have reason to believe, cannot be, or perhaps already has been, abandoned. It would provide the traditional epistemologist in our midst with the possibility of sawing all around that meagre portion of the ship that represents our sensory data, and setting the rest of it adrift, rudderless on the open sea. Certainly there is no guarantee against such sabotage merely in the thought that the saboteur must be aboard ship from the beginning of the journey, or even that he would have to stand with at least one foot on that huge, dispensable portion in order to cut it loose from the ship of knowledge in the first place.
Quine's naturalistic study of knowledge proceeds in terms of a general distinction between what we get through the senses and everything we believe about the physical world on the basis of those data. I would now like to argue that that conception of knowledge and of the epistemological task not only tolerates scepticism, as I have just been suggesting, but is actually committed to it. It would make it impossible for us to understand, even on its own terms, how our knowledge of the external world in general is possible. I must emphasize that I do not mean that there is anything wrong with the scientific study of human knowledge. It is the particular conception of the task that Quine relies on that I want to examine. It is not just a matter of finding out whatever we can about human knowledge by any respectable means; the specific task for Quine is to understand how our knowledge is possible by understanding how the 'meager input' at our sensory surfaces gives rise to the 'torrential output' in the form of sentences we accept as true about the external physical world.
Let us begin by asking how, in the most ordinary or even scientific contexts, we explain someone's knowledge, or explain how it is possible. For Quine it is a straightforward matter of scientific observation and explanation. We study:
a natural phenomenon, viz., a physical human subject. This human subject is accorded a certain experimentally controlled input-certain
p.235
patterns of irradiation in assorted frequencies, for instance and in the fullness of time the subject delivers as output a description of the three-dimensional external world and its history. (EN, 82-3.)
There is nothing problematic in this sort of investigation. As an observer or experimenter, I can observe a human being and observe his environment while also observing the 'output' he produces in the form of utterances I understand to be about the world around him. Given what I know about his surroundings and what (according to Quine) science tells me about the processes of perception, I can try to explain how the 'torrential output' to which I have access is related to, or produced by, the 'meager input' I know he is receiving at his sensory surfaces. Because I know how 'meager' those 'inputs' are, I know that what he says about the world is grossly underdetermined by his sensory impacts--indeed, on Quine's view by all the sensory impacts he and everyone else will ever have (WO, 22).9 In that sense I can see his talk of physical objects as a 'hypothesis' relative to his 'data'. Nothing he says about the physical world follows from truths about what is happening at his sensory surfaces. Relative to those sensory impacts, physical objects are for him 'posits', something he 'projects' from-his 'data' (EN, 83).
In calling his conception of the physical world a 'posit' or 'projection' beyond his 'data' I do not in this normal context imply that he does not know anything about the physical world or that his beliefs about it are not true or not reasonable. Since I am in a position to see whether what the person says about the world around him is true, I can determine whether his belief on a particular occasion is a mere posit or projection--something he believes and asserts, but
9 To say that truths about the physical world are 'far in excess of any available data' or are 'less than determined by our surface irritations' could mean that (i) truths about the physical world do not follow from the fact that certain surface irritations are occurring, or that (ii) truths about the physical world do not follow from the 'data' or 'information' supplied by those surface irritations. On the first interpretation, not all truths about the physical world would be underdetermined by the sensory impacts, since the truth 'Impacts I1, I2, ... occurred at sensory surfaces S1, S2,...' is not underdetermined by those impacts, and it is a truth about the physical world. I give more reasons below for thinking that Quine is to be understood in the second way, despite his explicit repudiation of the notion of 'awareness' or 'data'.
p.236
with no basis in fact. If I find, in a particular case or in general, that what he says is true of the world around him, I can still hold that his beliefs go well beyond his sensory impacts and in that sense are 'projections', even though they do not go beyond or in any other way misrepresent what is actually the case right before his eyes. If he says 'There is a tree' in a situation in which I find there is a tree before him, I know that what he says is true. If I were to explain his knowledge of the world in that situation I would at least have to explain how he came to get things right in that situation. In the normal case that is of course not difficult to do. I as an observer of the subject of my study can tell whether he gets things right or not because I can know both what he is saying about the world around him and what is true in the world he is saying it about. That is why my granting that the subject's beliefs about the world are 'hypotheses' or 'posits' relative to the meagre impacts at his sensory surfaces does not in itself imply that he knows nothing of the world or that I cannot explain how his knowledge is possible.
Sometimes when I am observing another human subject I can see that what he says or believes about the world around him is not true; there is no tree or anything else before him, despite his confident assertion that there is. In that case I see immediately that he' does not know there is a tree before him, he only believes there is. I might then say that he is merely projecting, that his belief in the tree is a mere posit, and is not true. I might go on to try to explain how and why he comes to have that false belief, but of course that would not be an explanation of how he knows there is a tree there, or of how his knowledge is possible, or of how he comes to get things right. In that case I would know that he does not get things right, he does not know there is a tree there. Since I see that his belief is false and he does not know, my explanation does not explain his knowledge, nor does it explain how he comes to have a true belief. Not every explanation of a subject's beliefs is therefore an explanation of knowledge, or even of true belief.
What then is required for us to explain someone's knowledge or even true belief in the kind of experimental situation
p.237
Quine envisages? I think the two kinds of cases considered so far show that the truth of what the person believes must play an essential role. But although the truth of the subject's beliefs is always relevant to whether he knows, it is not enough. Even if I can explain how and why the person comes to believe what he does, I will not necessarily thereby have explained how he comes to know it, or how he comes to have a true belief, even if his belief is in fact true and I as the observer know that it is. That is because it is possible for someone to have a true belief and yet lack knowledge--it might have been a coincidence or a lucky guess or a belief held for reasons unconnected with the truth of what is believed. A subject in such a position would lack knowledge, despite the truth of his belief, so no explanation of his belief would be an explanation of his knowledge.
This important condition of success for the kind of explanation even a fully naturalized epistemology should provide is not a consequence of imposing unreasonable standards of strictness or certainty on the notion of knowledge. Any explanation of the desired kind must at least explain how the subject comes to get things right, how he comes to have a true belief, whatever else (if anything) beyond that minimal condition the notion of knowledge is thought to imply.10 Of course, I might be able to explain how a person comes to believe what he does, and I might also be able to explain how or why the state of affairs he believes in actually came about, but that in itself would not be an explanation of true belief of the kind any theory of knowledge should provide. Suppose someone believes there are exactly one thousand four hundred and seventeen beans in a certain jar, and suppose I as a student of human behaviour can explain how he came to have that belief. He did not count the beans, he did not see them being put into
10 Some philosophers hold that knowledge requires nothing more than a true belief that p that is partly caused or maintained or explained by the truth of p. Even on such a view an explanation of someone's knowledge would have to explain how the person comes to get things right. For some examples of views along these lines see A. Goldman, 'A Causal Theory of Knowing', The Journal of Philosophy, 1967; P. Unger, 'Experience and Factual Knowledge', The Journal of Philosophy, 1968, and 'An Analysis of Factual Knowledge', The Journal of Philosophy, 1967; F. Dretske, 'Conclusive Reasons', Australasian Journal of Philosophy, 1971; R. Nozick, Philosophical Explanations, ch. III.
p. 238
the jar, no one connected with the filling of the jar told him how many beans there are, but he became convinced that there are just that number, and I know how he got that belief. Suppose I also happen to know that there are in fact exactly that number of beans in the jar, perhaps because I put them there myself, and I can explain how and why exactly that number of beans got there. Now those two explanations together--the explanation of the person's belief and the explanation of the jar's containing just that number of beans--might be said to provide me with an explanation of how he comes to have a true belief, or of how what he believes came to be true, but it will not be an explanation of how that person came to get things right. It will not be the kind of explanation of a true belief that must be involved in any explanation of human knowledge.
The kind of explanation that is required for explaining someone's knowledge involves something more. It will not be enough if it does not trace a connection between the truth of what is believed and its being believed. My combined explanation of the truth of the person's belief about the beans in the jar failed to do that. It leaves the truth of the belief an accident or coincidence. It is no accident that the person believes what he does--I have a fully satisfying explanation of that--and it is no accident that there are just that many beans in the jar--that too I can explain. But simply accepting both explanations does not provide me with an intelligible connection between the truth of the belief and its being a belief of his. In the kind of experimental situation Quine is imagining, then, I can explain the subject's knowledge in the right way only if I know that the world around him is as he says it is, and that its being that way is partly responsible for his saying or believing it to be that way. Only then would I be doing more than explaining the origin of a belief that happens to be true. An appeal to its truth would play an essential role in the explanation of the origin of the belief.
I am not raising a difficulty for the scientific, experimental study of human knowledge. It is obvious that we can and do observe human beings in interaction with their environment, that we can and do regard them as knowing
p.239
things about that environment, and that we can and do explain how they come to know those things. I have only been concerned to stress one of the conditions of our understanding how that knowledge comes about. We must be able to establish some connection between the truth of what they believe and their believing it. Knowing only what they believe, or even that what they believe happens to be true, would not be enough.
The point is confirmed by reflection on another position it is possible to find ourselves in when we are observing another person. So far I have imagined a case in which I know what the subject's beliefs are and I know whether or not they are true. If I know they are true, and I can explain how they come to stand in the proper relation to the facts they are about, I can understand how the person knows what he does. If they are not true, I can see that he lacks knowledge. But sometimes I might not be able to tell whether the beliefs I am interested in are true or not. I might be able to observe the subject of my study and determine what his sensory impacts are, and I might know what beliefs are expressed in that part of his 'torrential output' now of interest to me, but for some reason I might be unable to see or get any other information about the states of affairs he believes to hold. Perhaps some barrier obstructs my view, so I cannot at the moment tell, for example, whether there is a tree before him or not. In such a situation I would be restricted to what is happening in the subject himself and to his 'output', but I would know nothing about the world he is describing.
This of course is no position from which to conduct scientific research into this or any other subject's knowledge, and there is no suggestion that it is our normal position. If some barrier prevents me from checking on the truth of his beliefs I should simply remove the barrier, or wait until it goes away, or change my position, or at the very least study those beliefs of his on which I can check without difficulty. I do not mention this possibility to suggest any difficulties for the empirical study of man.
The point is that in this untypical and unusually restricted position I could not establish whether the subject knows something about the world he is describing or not. I would
p.240
know what he believes, and I would perhaps know what impacts at his sensory surfaces led him to believe it, but since I would not have access to the part of the world those beliefs are about there would simply be no telling whether they amount to knowledge or not. I could not compare what he says with the world he says it about, as I can in the normal unobstructed observational position, so I could not explain the relation or lack of relation between them. Given only what I would have access to in this unusually restricted position, I could see his beliefs as 'projections' from his 'data', so I could say 'He projects (or posits or puts it forward) that .. .', but I could not say 'He correctly believes that .. .' or 'He knows that .. .'. I could not see those beliefs as anything more than a mere projection or posit on his part. That is not to say that I would be in a position to say that they are nothing more than a mere projection and that they are not really true. I couldn't tell that either. It is just that I would not be in a position to see them as more than that.
Even if I could somehow explain in that position how the subject's 'meager sensory input' has led him to make and adopt the 'construction' or 'projection' I know he has made (and it is not clear how I could do even that)l1 that explanation would not be an explanation of his knowledge or of how
11 The problem is that if we had no information about the world beyond the subject's sensory surfaces we could 'explain' his 'output' by tracing its genesis only as far as his sensory surfaces. We could not go further and explain why his sensory surfaces are stimulated in just the ways they are. That would be at best a very limited 'explanation' of how our subject 'posits' bodies and 'projects' his physics from his 'data'. Furthermore, we could never explain how he came to speak as he does, and therefore how he came to believe what he does. Language-learning is to be explained by showing how the subject comes to behave linguistically in ways that conform to the general practices prevalent in his linguistic community, and we would know nothing about that community since it is beyond his sensory surfaces. Why he says things in the particular form in which he says them cannot be explained by his sensory impacts alone. Englishmen and Frenchmen often get the same impacts. Worse still, without some independent information about the world the subject's utterances are about, we could not even understand those utterances and thereby identify his beliefs about the world. So even to suppose that we have access to the subject's 'output' presupposes that we know things about the world he is talking about-even if, on a particular occasion of utterance, we might lack such information. This unusually restricted position with respect to another person's knowledge is therefore not one we could find ourselves in quite generally. This point is stressed by Donald Davidson in his theory of 'radical interpretation'.
p.241
he comes to have a true belief. I am simply not in a position to see his beliefs as knowledge, or as true. To explain how his knowledge or true belief is possible I must know what his beliefs are, and I must know what is the case in the world they are about. And I must gain my knowledge about the world independently of knowing simply what the subject's beliefs are; that he believes there is a tree before him is not enough for me to know whether that belief is true. Only if I had that independent information could I compare his belief with the world it is about and ascertain whether or not it is true.
So far I have discussed some conditions of success in the experimental study of particular cases of another person's knowledge in observable circumstances. That study as Quine conceives of it proceeds in terms of a distinction between a person's 'meager' sensory 'data' and everything he believes to be true about the external physical world on the basis of them. Even if there is nothing problematic about that distinction when applied in particular cases, we are still faced with the completely general question of how any human knowledge of the external physical world is possible at all. An explanation in terms of 'meager input' and 'torrential output' would help me explain how anyone at all ever comes to know anything about the external physical world only if I could see that that same kind of explanation can be applied quite generally to all other people and also to myself. We have perhaps by now learned to be at least suspicious of such generalizing moves in philosophy. We need to look carefully at the extension of what seems to work in particular cases to a general conclusion about all of human knowledge. If it turned, out that I could successfully apply that distinction on some occasions to some other people only because I did not simultaneously apply it to all, or if I could apply it to everyone else only because I did not simultaneously apply it to myself, I could not employ that distinction to understand with complete generality how our knowledge of the world is possible. Let us grant for the moment that I can understand in Quine's way how other people's knowledge is possible. That leaves the question whether I can understand in that way how my own knowledge of the external physical world is possible.
p.242
There might seem to be no difficulty here. Quine explains how we are to achieve the required generality:
We are studying how the human subject of our study posits bodies and projects his physics from his data, and we appreciate that our position in the world is just like his. Our very epistemological enterprise, therefore, and the psychology wherein it is a component chapter, and the whole of natural science wherein psychology is a component book all this is our own construction or projection from stimulations like those we were meting out to our epistemological subject. (EN, 83)
The position we find another human subject in, on Quine's view, is that of 'positing' bodies or 'projecting' all of physics from the 'meager' sensory data to which he is restricted in his contact with the world he believes to exist. If each of us, in thinking of himself, must 'appreciate that our position in the world is just like his', each of us will have to appreciate that we too are restricted to 'meager' sensory data, and that all of our beliefs about the physical world around us go far beyond, or are grossly underdetermined by, those data.
I think we cannot perform that act of 'appreciation'--we cannot see all our own beliefs about the world as a 'construction or projection from stimulations'--while still explaining how our own, or anyone else's, knowledge of the world is possible. I do not mean simply that I cannot see all my own beliefs about the physical world as 'posits' or 'projections' which go beyond the 'meager' data at my sensory surfaces. Perhaps I can manage to see my position that way. It certainly seems possible to see another person as in just that position. Nor do I mean that an explanation of how someone else's knowledge is possible cannot be understood to apply to my own knowledge as well. I am a human being like everyone else, so what is true in general of others is also true of me, including the ways we come to know things about the world around us. But I do deny that I can do both things at once. I think I cannot see all my own beliefs about the physical world as a 'construction or projection from stimulations' and at the same time explain how I can know anything about the world around me. Even if I can see others' beliefs in that way and still explain their knowledge of the world in those terms, as it seems I can, what I want to argue is that I cannot explain how my own knowledge is
p.243
possible if I regard all my beliefs about the world around me as 'posits' or 'projections' that go, beyond my 'meager' sensory data. And if I cannot understand in that way how my own knowledge is possible, I cannot understand in that way how any other person's knowledge is possible either. Or, putting it another way, I can understand others' knowledge as a 'projection' from 'meager' sensory 'data' only on the condition that I do not understand all human knowledge of the world in that way. That is what I will now try to show.
What happens when I try to take up the view that all my I beliefs about the external physical world amount to a 'construction or projection' from 'meager' sensory 'data'? I know what all my beliefs about the world are, but I do not have any independent access to the world those beliefs are about on the basis of which I could determine whether or not they are true. In the normal case in which I am studying another person in interaction with the world, I can do that. I know what his beliefs are, and I can know, independently of the fact that he has those beliefs, what is the case in the world those beliefs are about. That is what enables me to explain how his knowledge is possible in that situation. In my own case, if I regard all my beliefs about the world as 'posits' or 'projections' from sensory data, I would not be in that position. I would find myself with a set of beliefs or dispositions to assert things about a physical world, but I would have no independent access to the world those beliefs are about.
Of course, I could try to do what is normally regarded as finding out whether my beliefs about the world are true. I could do what we call looking at the world around me, or perhaps listening or reaching out and touching, or even measuring or doing experiments in order to see whether my beliefs are true or not. But as long as I retained the idea that all my beliefs about the physical world are a 'construction or projection from stimulations' which they far transcend (as I must do if I am to perform Quine's act of 'appreciation'), I would have to regard myself as getting no closer to knowing whether or not my beliefs about the world are true. After 'looking at the world' (or experimenting on it) I would at best find myself with a set of stronger, or perhaps altered, beliefs or dispositions to assert things
p.244
about a physical world. I would know what those reinforced or newly-acquired beliefs are, but I would have to regard them in turn as simply further elements of my 'construction or projection' from my recently-increased, but still extremely meagre, 'input'. Those 'confirmations' or 'verifications' or 'experiments' could not be seen as giving me any independent information about the world against which the truth of the earlier beliefs had been checked. They would just give me more of the same. And what I would still not know is whether any part of my 'construction or projection' is true.
I am not simply making the point that it is impossible to check our beliefs against the world they are about. In fact I think that in the normal case that is possible--at any rate, I see nothing wrong with describing our verification or testing procedures in that way. If I say or believe that a certain book is in a certain position in the next room, and I then go into that room to find out whether I am right in what I say or believe, I see nothing wrong with saying that I checked my belief against the facts, or even that I compared my assertion or belief with the way things are. I think such things happen every day, and that they can be described in those ways.
My present point is that I could not check my beliefs about the physical world against the facts of the world in that way if I at the same time regarded all my beliefs about the physical world as nothing more than a 'construction or projection from stimulations' in the way Quine intends. I would have no independent information about that world that I could use as a test or a check.
We saw that in studying another person it is possible to find ourselves at least temporarily barred from information about the world that person's beliefs are about. In that rather unusual if nevertheless possible position, independent access to the facts is denied me, so I cannot regard the person's beliefs as knowledge or explain how his knowledge or even true belief is possible. I would know enough to enable me to say ‘He projects (or posits or puts it forward) that ...’, but I would have no way of going on to the stronger verdict ‘He correctly believes that ...’ or ‘He knows that ...’. I would be in no better position with respect to my own beliefs about the physical world if I followed Quine's proposal
p.245
of regarding all my beliefs as a ‘construction or projection from stimulations’. I could see myself as believing or ‘projecting’ or ‘positing’ various things about a physical world, but I could not see myself as having knowledge or even true beliefs about such a world, so I could not give an explanation that explains my knowledge or explains how I came to get things right.
This is not to deny that I could think it is possible that I have true beliefs, in the sense that nothing prevents my beliefs from being true. And if I think true belief amounts to knowledge if the state of affairs believed in is connected in the right way with my belief, I might also think it is possible that my beliefs amount to knowledge, since I find no contradiction in the thought that my beliefs are connected in that way with the world they are about. It is to be expected that I would have these thoughts, since if I believe something I will also think it is not impossible that it is true, and if I claim to know something, I will think it is not impossible that I know it. But even if my beliefs did happen to be true or did happen to be connected with the states of affairs they are about, I still could not explain or understand my true belief or knowledge in the way a theory of knowledge should explain it. I could never show that what I believe is in fact true or explain how or why it does amount to knowledge, as long as I retain the idea that all my beliefs about the world are a ‘construction or projection’ from ‘meager’ impacts at my sensory surfaces.
In fact, if we take completely seriously this talk of sensory surfaces, we can see that applying Quine's proposal to oneself would leave each of us in an even worse position than the one I have described so far, and worse with respect to ourselves than even the unusually restricted situation we can find ourselves in when we lack knowledge about another person's environment. Quine's epistemological problem is to explain the relation between the ‘meager input’ at one's sensory surfaces and one's ‘torrential output’ in the form of a body of beliefs or assertions about a physical world. But strictly speaking, my belief that I suffer impacts at my sensory surfaces, and indeed that I even have sensory surfaces at all, are themselves beliefs of mine about an external physical
p.246
world. Even my ‘scientific’ belief that my beliefs about the physical world are ‘projections’ from impacts at my sensory surfaces is itself a belief about the physical world. If I am to see that ‘discovery’ too as nothing more than a ‘projection’ from my ‘data’, what attitude do I now take to the very problem a naturalized epistemology is supposed to answer? In trying to study another person while lacking access to his environment, I at least can know what is happening at his sensory surfaces and what his beliefs about the world are. I cannot fully answer the question about the relation between his ‘meager input’ and his ‘torrential output’ in that position, but I do have some independent information about at least part of the physical world; I know what his ‘impacts’ are. But in my own case, following Quine's proposal, I would not even have that. The unquestioned information about part of the physical world that enables me at least to ask the questionabout another person would have to be seen in my own case as nothing more than a further part of an elaborate ‘projection’ of a physical world that I somehow have been led to ‘construct’ and believe. I would see all my beliefs about my sensory surfaces as just more of my ‘torrential output’. I would have lost independent access to anything physical whose role in producing my ‘output’ I can ever hope to investigate or explain.
Trying to follow Quine's proposal and apply his conception of knowledge to myself, then, I would be left in an even worse position than that of an observer barred only from information about the truth of his subject's belief. Perhaps the closest parallel to it in a third-person case would be that of finding myself alone in total darkness and silence and suddenly hearing the words ‘There is a tree’ coming from somewhere. Obviously there is simply no telling in that situation whether those words express knowledge, or even truth, so there would be no possibility of explaining, with only that sort of information, how in that case knowledge or even true belief is possible. I would have nothing but that bit of ‘output’ to work with. But that is the position I would always be in with respect to my own beliefs about the physical world if I ‘appreciated’ that all my beliefs about the physical world are ‘projections’ from ‘meager’ sensory data.
p.247
I would have nothing but my own ‘output’. That for me would be no better than whistling in the dark.
While I continue to regard all my beliefs about the physical world as ‘projections’, and so remain within this very restricted position, I might nevertheless come to wonder how some of the things I believe in are related to or connected with my believing and asserting the things I do. I believe in impacts at my sensory surfaces, for example, and I believe that my sensory ‘input’ is meagre and my scientific ‘output’ torrential. Natural curiosity, not to mention the ‘reasons that always prompted epistemology’ (EN, 83), might then lead me to seek some explanation of how that torrential ‘output’ could be generated on such a slender basis. In an effort to understand this puzzling relation between some of the things I believe I might appeal to other beliefs of mine--for example, about psychology or physiology or language-acquisition, or any other part of science that I think might help. But as long as I remember that all the ‘science’ I am appealing to itself just amounts to more and more ‘projections’ from my ‘data’, I will appreciate that telling myself that complex story about ‘input’ and ‘output’ is just a matter of expressing more and more of my elaborate ‘construction or projection’ of a physical world. I could not see my efforts as providing me with an explanation that itself is something I know or have reason to believe, as opposed to a complicated story I fully accept and find myself disposed to tell myself from time to time.
The requirement that we see all our beliefs about the physical world as ‘projections’ has disastrous consequences for the theory of knowledge. I think we tend to overlook them or deny them because (not surprisingly) we do not usually manage to fulfil the requirement completely, even in our thoughts. We unwittingly take some things as unquestionably true about the physical world, and not merely as ‘projections’, even while we are trying to think of human knowledge in Quine's way. But if we accept with complete generality the idea of human knowledge as a combination of a subjective and an objective factor, and we see the objective contribution of the world as small relative to the
p.248
total set of beliefs we hold about the world, we must see the subjective factor (the contribution of the knowing subject) as largely determining our total set of beliefs about the world. Countless ‘hypotheses’ or ‘theories’ could be ‘projected’ from those same slender ‘data’, so if we happen to accept one such ‘theory’ over others it cannot be because of any objective superiority it enjoys over possible or actual competitors. Every consistent ‘theory’ compatible with those same meagre ‘data’ is in that sense a competitor of the ‘theory’ we now accept, so our continued adherence to our present ‘theory’ could be explained only by appeal to some feature or other of the knowing subjects rather than of the world they claim to know. And that is precisely what the traditional epistemologist has always seen as undermining our knowledge of the external world. The possibility that our belief in an external world is nothing more than a mere ‘projection’ on our part, nothing more than something we accept because of certain things true about us and not about the independent world we believe in, is the very thing that had to be shown not to obtain if we were to understand how our belief in an external world amounts to knowledge or even something we have reason to believe.
It is Quine's idea that depriving the would-be philosopher of a vantage-point outside our knowledge of the world would be enough to eliminate the prospect of a totally sceptical outcome to reflections on knowledge. That would leave naturalized, scientific epistemology as the only epistemology there could be. There is wisdom in that strategy, but I have argued that it will not succeed as long as all our knowledge of the world is seen as a ‘projection’ from meagre sensory data that grossly underdetermine it. I thereby echo Kant's idea that a completely general distinction between everything we get through the senses, on the one hand, and what is or is not true of the external world, on the other, would cut us off forever from knowledge of the world around us. That general epistemic distinction is fatal to the naturalizing project. It has the effect of casting us out of our own knowledge of the world, as it were, and leaving us with no independent reason to suppose that any of our ‘projections’ are true. It appears to provide just the kind of place an
p.249
epistemic exile could resort to when he discovers the poverty of his position with respect to knowledge of the world. If there is to be no such exile, there should be no such place.
But for Quine the very distinction that I say leads to difficulty is itself a deliverance of science. ‘It is a matter of scientific fact, or theory’, he says, ‘that our only avenue of information about external objects is through the irritation of our sensory surfaces by forces emanating from those objects’. And ‘science itself teaches that ... the only information that can reach our sensory surfaces from external objects must be limited to two-dimensional optical projections’ and the like (RR, 2). If that is indeed what ‘science tells us’ (NNK, 68), how could that general distinction have the consequences I draw from it? In arguing that we cannot understand all our knowledge of the world as a ‘torrential’ ‘construction or projection’ from ‘meager’ sensory ‘data’ in that way, am I not simply flying in the face of the scientific facts?
I don't think so. Nothing I have said implies that we cannot observe a person in interaction with his environment and isolate from everything else certain events that are occurring at his sensory surfaces. We know, and he might know, a great many other things about what is going on in the world around him beyond those events. In fact it is extremely unlikely that he himself will have any idea of what those events at his sensory surfaces are like. But there are such events, and if we know enough about physiology, and about him, we can pick them out from all the rest. And of course this holds quite generally. Whenever any of us is in interaction with his environment there are events occurring at his sensory surfaces. No doubt such events should be reckoned as part of what causes us to get the beliefs about the world around us that we do. It seems obvious that if our sensory surfaces were not stimulated we would never come to believe anything about the world around us. Science (in Quine's all-encompassing sense) does ‘tell us’, as Quine puts it, ‘that there is no clairvoyance’ (RR, 2). There are causal chains of events leading from objects around us to events deep in our brains, and the events at our sensory surfaces occur as parts of such chains.
p.250
There could be no objection to studying those events and seeing what they lead to, and what their effects lead to in turn, and so on.
Quine sometimes describes his project of naturalized epistemology in this way, especially when he is emphasizing how it would avoid the traditional worries about circularity and epistemic priority. The traditional epistemologist was concerned to isolate something that we are directly aware of in perception, and that led to disputes between ‘sense-data’ theorists and Gestalt psychologists about what sort of mental item is present to consciousness in perception. Quine would avoid the issue by ‘talking directly of physical input at the sense receptors’ (RR, 4). Since ‘reception is flagrantly physical’ (RR, 4), ‘it is simply the stimulations of our sensory receptors that are best looked upon as the input to our cognitive mechanism’ (EN, 84). There is then no talk of awareness and no need for the notion of epistemic priority that was essential to the traditional question.
Now that we are permitted to appeal to physical stimulation, the problem dissolves; A is epistemologically prior to B if A is causally nearer than B to the sensory receptors. Or, what is in some ways better, just talk explicitly in terms of causal proximity to sensory receptors and drop the talk of epistemological priority. (EN, 85)
A naturalized epistemology so understood would study the relation between our ‘input’ and our eventually coming to believe what we do about the world around us by studying how those events at our sensory surfaces cause other events closer to ‘our cognitive mechanism’ and eventually cause our beliefs about the world around us. More strictly, it would study the ways in which events at our sensory surfaces cause those events which are comings-to-believe-something about the world around us. But whatever such an investigation managed to discover and explain, it would not provide an explanation of how our ‘meager sensory data’ give rise to a ‘torrential output’ about the world that is grossly underdetermined by those data. It would not show by what ‘bold inference’ we manage to ‘bridge’ the ‘wide gap between our data and our knowledge of the external world’. That is because it makes no sense to say that between one event
p.251and another in the same causal chain there is a ‘gap’ that is to be ‘bridged’ by ‘inference’. There is just one event which leads to another, and then to another, and so on. It makes no sense to say of one event (e.g., an impact at a sensory surface) that it ‘underdetermines’ another event (e.g. a coming-to-believe-something) that occurs later in the series. Of course it is true that all the events that occur in the interval between the earlier and the later event are also needed to bring about the later event, so in that sense the impact alone does not cause the believing, but that kind of causal insufficiency is not what Quine means by ‘underdetermination’.
He means that the ‘data’ do not imply the ‘torrential output’; they do not logically determine what the ‘output’ will be; many different ‘outputs’ are logically compatible with those same ‘meager’ ‘data’. Just as ‘truths about molecules’ are underdetermined by ‘truths that can be said in common-sense terms about ordinary things’, so the ‘truths that can be said in common-sense terms about ordinary things’ are ‘less than determined by our surface irritations’ (WO, 22, my italics). It is the truth or falsity of the content of the ‘output’ that Quine says is not ‘determined’ by the data or the sensory impacts; the relation of ‘underdetermination’ holds between one set of truths and another. The sensory impacts or irritations (along with the intervening events) cause the event that is the coming-to-believe or coming-to-be-disposed-to-assert the ‘output’, but that is not the kind of ‘underdetermination’ Quine has in mind in posing his epistemological question. He asks how knowledge is possible, given that ‘the only information that can reach our sensory surfaces from external objects’ (RR, 2, my italics) is ‘meager’ in relation to what we come to believe about those objects as a result of receiving that sensory ‘information’. That gap is just what gives rise to Quine's epistemological problem. That is why in answering it we will learn ‘how evidence relates to theory, and in what ways one's theory of nature transcends any available evidence’ (EN, 83).
The ‘impact’ itself, the event, is not ‘meager’ in relation to another event that is ‘torrential’. Something happens at a sensory surface, and then a coming-to-believe-something-about-the-world occurs. The relation between those two
p. 252
events is simply that the former causes the latter (along with the help of those in between). But ‘underdetermination’ speaks of a relation between something that is ‘meager’ relative to something else that is ‘torrential’; the latter ‘transcends’ the former. If we think only about the events involved the events at the sensory surface and the events closer to ‘our cognitive mechanism’ that result in our believing what we do and we drop all talk of ‘meagerness’, ‘underdetermination’, ‘torrential output’, and so on, what becomes of Quine's question about our knowledge of the world around us? We are left with questions about a series of physical events, and perhaps with questions about how those events bring it about that we believe what we do about the world around us. But in trying to answer those questions we will not be pursuing in an ‘enlightened’ scientific way a study of the relation between ‘observation’ and ‘scientific theory’, or of the ‘ways one's theory of nature transcends any available evidence’, or of ‘the domain within which [man] can revise theory while saving the data’. We will be studying the connection between one kind of event and another.
I think the question Quine poses in terms of the ‘underdetermination’ of the ‘torrential output’ by the ‘meager input’ makes essential use of a notion of epistemic priority. It is because the ‘information’ we get at ‘input’ does not uniquely determine the truth of what we assert as ‘output’ that we must explain how we get from the one to the other. We could know everything included in our ‘evidence’ without knowing any of the things asserted in our ‘theory’. If ‘input’ were not understood as ‘evidence’ or ‘data’ or ‘information’ in this way, to say that it ‘underdetermines’ the ‘output’ would make no sense, or not the kind of sense Quine says it makes. But if the ‘input’ is to be understood in that way, our ‘data’ must be understood as something we are in some sense aware of, after all. They could be described as ‘evidence’ or ‘information’ only if that were so. Quine explicitly denies that he thinks of impacts at our sensory surfaces in that way; he wants to avoid all questions of awareness. But he can do so only by avoiding all talk of the ‘meagerness’ of our ‘input’ relative to our ‘torrential output’ as well.
p.253
It might seem that once again we simply come up against, the deliverances of science. ‘It was science itself’, Quine says, ‘that demonstrated the limitedness of the evidence for science’ (RR, 3). Has science demonstrated that? Quine I does not specify what scientific results he has in mind here. It could not be the simple fact that we would not believe or know anything about the world around us unless we received impacts at our sensory surfaces. That says nothing about the ‘evidence’ we might have for our beliefs about the world, so it says nothing about ‘the limitedness of the evidence’. Still, as things are, it is true that there is no clairvoyance. The indispensability of causal interaction between the world and active sensory surfaces seems undeniable. What I have meant to deny, with Kant, is that we can regard all our beliefs about the world as ‘projections’ or as ‘theoretical’ relative to some ‘data’ or bits of ‘evidence’ epistemically prior to them, while at the same time explaining how our knowledge of the world is possible. I do not see what actual, or even possible, scientific findings I could be in conflict with in saying that. Quine's project of naturalized epistemology has the interest and the apparent connection with traditional epistemology that it has only because it contains and depends on just such a bi-partite conception of human knowledge of the world. That is what I have argued cannot succeed in explaining how knowledge is possible. But without that conception, ‘naturalized epistemology’ as Quine describes it would be nothing but the causal explanation of various physiological events.
The ‘nothing but’ is not a disparaging expression. It would be absurd to disparage the scientific study of human beings, or of anything else. My aim is only to distinguish ‘naturalized epistemology’, understood as physiology, from the ‘enlightened persistence ... in the original epistemological) problem’ (RR, 3) that Quine claims for his own project; It was also absurd to disparage the lecturer in Chapter Three who said that most of us know there is an enduring world, or even to disparage G. E. Moore who said that he knew there are external things. I have wanted only to point out that those remarks do not answer or even address themselves to the philosophical problem of the external world.
p. 254
The same is true of the physiological study of causal chains of events leading inwards from our sensory receptors. It is only Quine's project conceived in terms of ‘data’, ‘evidence’, ‘theory’ and ‘output’ that I want to say (with Kant) can never explain how human knowledge of the world is possible.