HOW TO READ "EPISTEMOLOGY NATURALIZED"*

W.V. Quine is well known for urging the abandonment of epistemology, as traditionally pursued, in favor of the scientific project he calls "naturalized epistemology."¹

Epistemology, on Quine's view, is a branch of natural science; ... [it] should not concern itself with how we should form beliefs but with how we do in fact form them.²

[Quine] is generaIly credited with having started the move toward naturalism in epistemology .... Quine's view appears to leave no room for nonnative evaluations in epistemology.³

No one, to my knowledge, would dispute these characterizations of the burden of Quine's "Epistemology Naturalized"⁴ and its role in "the move toward naturalism in epistemology." Yet every student of Quine knows that as early as "Two Dogmas of Empiricism"⁵ and as late as Pursuit ofTruth⁶ he pursued normative epistemological investigations that cannot sensibly be seen as falling within the purview of natural science or, more particularly, psychology. Hilary Putnam, the only philosopher I know to have considered how this glaring conflict might be resolved, was forced to throw up his hands; although Quine repeatedly assured him that he did not mean to "rule out the normative," Putnam could find no way of reading "Epistemology Naturalized" except as a proposal to do exactly that.⁷ Eventually,

" I dedicate this essay to the memory of Willard van Orman Quine.

I am indebted to Roger Wertheimer, as always, for much discussion, and for giving me the benefit of his unerring critical eye; I thank Paul Saka, as well, for perceptive comments on two late drafts.  My work was supported by a Faculty Development Leave from the University of Houston's College of Liberal Arts and Social Sciences.

Quine gave the world at large the same assurance, and even sketched his views of the fundamental norms of epistemology and how they are rightly arrived at,⁸ but no one has yet done better than Putnam in seeing how the essay can be squared with that assurance or with Quine's undiminished interest in normative questions.

I take it as a given that the universal reading of "Epistemology Naturalized" as a proposal to "surrender the epistemological burden to psychology" is wholly mistaken, since the alternative--that Quine, in mid-career and without so much as a "by your leave," delivered himself of a proposal whose implementation would have required him to abandon some of his deepest lifelong concerns, and that he failed for decades to detect the conflict--is utterly absurd. Here I shall offer a reading of the essay on which its central proposal is consistent with his fundamental views, and briefly discuss a few related matters. On my reading, not only does Quine not urge the abandonment of epistemology as traditionally pursued, but, whatever inspirational role "Epistemology Naturalized" may have played in the move toward "naturalism in epistemology," its central proposal is actually antithetical to a good deal of what is nowadays done under that heading.

The universal misreading of Quine's essay is unquestionably a bizarre phenomenon, and so, fittingly, is its explanation: the fault lies not in the universals incompetence of the essay's readers, but rather in a disastrous failure of its author's outsized gifts as an expositor of his own views.  No one familiar with Quine's work could arrive lightly at such an assessment of any part of it, nor have I done so in this case; in trying to understand how things went wrong I have, however, been buoyed by one truth I take to be beyond peradventure: better expository failure than perdurable blindness to blatant conflict between principle and practice.

We begin at ground zero. After noting some milestones in the history of attempts to "account for the external world as a logical construct of sense data," including Rudolf Carnap's efforts at rational reconstruction, which, if successful, "would have enabled us to translate all sentences about the world into terms of sense data, or observation, plus logic and set theory," Quine asked his now famous questions:

[W]hy all this creative reconstruction, all this make-believe? The stimulation of his sensory receptors is all the evidence anybody has had to go

p.80

on, ultimately, in arriving at his picture of the world. Why not just see how this construction really proceeds? Why not settle for psychology (EN 75)?

Here Quine muddies the water at the very outset.  The reader could easily take him to be contrasting two projects with identical starting points--a subject's experiences--and equivalent endpoints: two versions of a single picture of the world.  That is, he might seem to be suggesting that there is a way, not necessarily "rational," in which subjects actually construct their pictures of the world from their experiences, and a way, sketched by Camap, in which those pictures could be rationally (re)constructed from those same experiences, and proposing that we settle for the former.  But Carnap's project was not to construct rational routes from a subject's data to his picture of the world; rather, it was to devise (re)constructions of our ordinary physicalistic concepts that would allow us to "translate all sentences about the world into terms of sense data, or observation, plus logic and set theory."  Even if his project had succeeded, no subject would have been in a position to infer the Carnapian translation of a typical physicalistic sentence from his experiences by tracing some Camapian path from the one to the other. One reason is elementary: the translation of a typical sentence about the world would have referred to vastly more observations than any subject could possibly make, a point Quine illustrated by noting that "The most modest of generalizations about observable traits will cover more cases than its utterer can have had occasion actually to obselve" (EN 74). A deeper and more theoretically illuminating reason is that there could not in principle be any such translations of most sentences, since "the typical statement about bodies has no fund of experiential implications it cart call its own" (EN 79).

In the terms of Quine's broad framework in "Epistemology Naturalized�," subjects' constructions of their world-pictures and Carnap's efforts at rational reconstruction fall on opposite sides of his division of epistemology into doctrinal and conceptual studies.  Subjects engaged in constructing pictures of the world are developing "doctrines" that are the objects of doctrinal studies, which concern, inter alia, the justifiability of those doctrines; Carnap's efforts at reconstruction are conceptual studies aimed at showing that the everyday concepts employed in formulating those doctrines are theoretically dispensable in favor of sense data, or observations, plus logic and set theory.

But if Quine is not contrasting the individual subject's actual construction of his picture with a hypothetical Camapian parallel, then what is he contrasting it with? What, exactly, is the study he proposes to abandon in favor of the study of psychology? The answer to these

p.81

questions should have been unmistakable, though everyone seems in fact to have mistaken it: "all this creative reconstruction, all this make�believe," that is, the conceptual studies pursued by Carnap and others.  This means that the question--"Why not settle for psychology [rather than pursue the project of creative reconstruction]?"--is quite peculiar; it would make clear rhetorical sense only if the two projects were at least roughly parallel, or comparable. That Quine should have posed this odd question is itself mildly unfortunate, but in doing so he also set the stage for something much worse: the principal expository blunder responsible for the misreading of his essay.  Having proposed to settle for psychology in place of "creative (rational) reconstruction," he went on to describe such a settling as "a surrender of the epistemological burden to psychology" (EN 75; my emphasis); given that description, his readers can hardly be faulted for having read him as proposing to abandon epistemology itself in favor of psychology.

Nevertheless, that reading is as mistaken as the description is misleading. Note first that the description, as it would be understood by any well-informed reader, is absurd; while epistemology includes efforts at rational reconstruction, it certainly does not consist of them, so that abandoning them would hardly amount to abandoning the entire discipline. The extraordinary description is explained by an equally extraordinary fact: as Quine uses the term 'epistemology' in "Epistemology Naturalized," that discipline does consist of efforts at rational reconstruction--or, more broadly, attempts to discover or establish logical, translational or rational-reconstructive links between observation and theory. This is readily confirmed simply by rereading the essay; I shall cite only two passages in whose conjunction it is starkly evident:

 

 

In short, discovering the impossibility of reduction, or translational reconstruction, would have meant the bankruptcy of epistemology were it not that it could be reconceived as a chapter of psychology.

Quine's idiosyncratic use of "epistemology" explains another oddity as well. In briefly sketching a few salient points in the history of modern epistemology, he restricts his attention entirely to efforts of

p.82

a logical or quasi-logical sort by philosophers from Hume to Carnap; one will search in vain, in "Epistemology Naturalized" or elsewhere in his writings, for any substantial discussion of what might be called "mainstream" efforts by numerous epistemologists to show that belief in the external world is, or can be, justified in the absence of such logical or quasi-logical links.  Indeed, he appears at times to be entirely oblivious of those efforts:

The suggestion that awareness is not demanded by any conceivable effort other than rational reconstruction to justify our belief in the external world would have come as news to those many epistemologists who thought (as many still do) both that some role for awareness is demanded for epistemic justification and that such justification is possible in the absence of rational-reconstructive links between evidence and theory.⁹

Given, then, that what Quine proposes to abandon is only the project of rational reconstruction, we can reformulate his stage-setting questions as follows: "why all this searching for logical or quasi-logical connections between evidence and theory? Why not just see how we construct our theories from our evidence? Why not settle for psychology?" An apparent lacuna in "Epistemology Naturalized," one that prompted a second question from Putnam, is made even more glaring by this reformulation: Why did Quine not consider the possibility of finding an inductive evidence/theory connection, once he became convinced that no deductive, or quasi-deductive, one was to be found? But philosophers puzzled by this (as I certainly was) have presumably failed to register the fact that, as early as "Two Dogmas of Empiricism," he had rejected the idea that any such connection existed:

[Physical objects are myths on the same footing with abstract entities]

p.83

 

It follows from this--and I take it that only Quine's distaste for redundancy kept him from making the point explicitly--that the myth of physical objects is no more probable relative to our evidence (whether experiences, sense data, observations, or stimulations), is no more justified by that evidence, than is the myth of the Homeric gods; the only relevant difference between the two is that the former better expedites our dealings with sense experiences.

In Word and Object he made the point explicitly:

 

To indulge again in redundancy, observations alone cannot justify theories.

His clearest and most emphatic statement of the point came in Pursuit of Truth, in one of the few places where he can sensibly be read as at least alluding to foundationalists' contentions that our experience provides support for our belief in an external world:

 

 

Quine's position, then, is that there are neither deductive, nor rational�reconstructive nor inductive links to be found between evidence and theory, and that if we want to understand how the two are related we should settle for psychology--we should study how we relate the two, that is, how we actually construct our theories from our evidence.¹³

Before considering where he thinks such an investigation would lead, we need to address an obvious preliminary question about its viability.  Quine is clearly proposing that we study the way subjects construct their pictures of the world, not countless idiosyncratic ways in which individual subjects may do so.  But how could he sensibly suppose that there is a single such way, given that subjects with comparable stimulation histories construct widely varying, even conflicting, world pictures?

Quine could do so because he thought he had long since identified that way.  In his view, our constructions have two stages, each characterized by a uniform method: in the first stage we construct our proprietary versions of "the immemorial doctrine of ordinary enduring middle-sized physical objects"¹⁴ (our "doctrines of an external world"); in

writings did he initiate any new line of inquiry, psychological or otherwise, that could conceivably be seen as an attempt to identify that method. What are we to make of this? How are we to understand his apparent failure to take up his own proposal? These questions are wrong headed, and simply evaporate once we realize that "Epistemology Naturalized" does not contain any such proposal. Quine's aim in that essay was not to launch a new line of investigation, but rather to persuade his readers to join in an old one in which he had long been engaged.  If we want to locate the origins of his efforts to naturalize epistemology, we have to look backward from "Epistemology Naturalized."

At least as early as Word and Object, Quine had carefully considered the relation between subjects' inputs and outputs, and discovered a good deal about how those subjects relate evidence to theory:

Reflection on the most elementary facts about the nature of subjects' inputs and outputs revealed that even the least inquisitive among us conform in our theorizing to some version, or proto-version, of scientific method; it revealed that our second-stage method is that of constructing theory from evidence in conformity with a norm of maximizing simplicity.

Is this, then, at least a broad-brush, preliminary version of Quine's proposed project and the knowledge to be acquired by pursuing it?  On the universal reading of "Epistemology Naturalized," it is; in the

language of the quotations with which this paper began, the project described is certainly a "scientific" one, it does "not concern itself with how we should form beliefs, but with how we do in fact form them," and it seems to have no place for "nonnative evaluations"--it issues simply in a description of how we do in fact construct our theories from our evidence.  But if Quine's idea were simply that we should abandon normative questions in favor of scientific ones, then why psychology in particular?  Why not, say, geology?  There must be some significant relation between rational reconstruction and psychology if the suggestion that we "settle for" the latter in place of the former is to be at all sensible, and indeed there is: both promise to illuminate the relation between evidence and theory.  It is Quine's revolutionary conviction that by learning the psychological truth that we relate evidence to theory in conformity with scientific method, we learn the philosophical truth that evidence is related to theory by scientific method. This is a point he had long since advertised:

To put his point even more starkly, the only connections there are between our evidence and our theories are those we make in accordance with scientific method. (To avoid any suggestion that this method is independent of us--something we have discovered, rather than devised--it might better have been called "our" scientific method, but for the latter's suggestion that it is one among others.)

But how, then, can Quine claim that by learning how we relate evidence to theory we learn how evidence is related to theory? He can do so because in his view the former determines the latter, which is a corollary of his more general view that there is no exile from theory.  By the latter he means, inter alia, that there is no vantage point from which our comprehensive method of devising theories to explain past observations and predict future ones can be evaluated in the relevant respect: we have no independent way of identifying true theories that would allow us to consider whether our method of theorizing tends, or tends better than would alternative methods, to yield true theories: "[W]e can never do better than occupy the standpoint of some theory or other, the best we can muster at the time."¹⁷ In terms of Quine's favorite simile, borrowed from Otto Neurath, our comprehensive

theory is like a boat at sea, and we have no alternative to staying on board if we want to stay afloat, that is, to think rationally at all; if we want to find out how to arrive at the truth on the basis of our evidence--if, that is, we want to understand how evidence relates to theory--we have no alternative but to take (our best understanding of) the way we do arrive at theory as the way we are to arrive at theory.  It is, of course, an all-important fact that any part of our comprehensive theory--including any statement about how evidence relates to theory--can be modified or repaired, that "no statement [belief] is immune to revision"¹⁸

We have now arrived at an entirely new picture of Quine's principal efforts in "Epistemology Naturalized"; so far is he from proposing to abandon the normative that he is proposing instead to discover the norms that govern theorizing by discovering the norms that we conform to in our theorizing. This means that his talk of surrendering the burden of epistemology to psychology is unfortunate in a second respect as well. What he is actually proposing is to enlist the aid of psychology in addressing the burden of epistemology: psychology will identify the norms we adhere to, and philosophy will tell us that, by virtue of their being the ones we adhere to, they are the ones we are to adhere to.  He himself later put the point in much the same way:

Several loose ends need tidying up.  First, the question, "Why not settle for psychology?" is at best a bit peculiar if we have no alternative--if, that is, there are no connections between observation and theory other than those we make, hence none for us to discover--and I can see no way of reading it so as to obviate this peculiarity altogether.

p.89

Yet there is an important point to be made about it.  We have seen that for Quine settling for psychology does not mean settling for mere description; getting the relevant psychological descriptions right is only the first step in arriving at an understanding of the norms of epistemology.  Thus I suggest that Quine may have had in mind something like the following question: "Why not settle for the justification that we can see we already enjoy, rather than search for a further justification via rational reconstruction (especially since none is to be had)?" This question captures the central thrust of his essay, to wit, that we should stop beating a dead horse--stop trying to devise some ingenious story about how our beliefs are justified by virtue of some intrinsic relation in which they stand to our evidence--and remind ourselves that the only possible account of epistemic justification, hence the one for which we should "settle," is one framed in terms of the scientific method to which psychology tells us we conform, which simply is "the way to truth."²¹

A second loose end concerns the original apparent obstacle to the idea that there is a single way in which we construct our comprehensive theories: If we all construct them in accordance with scientific method, then what explains the fact that subjects with comparable stimulation histories arrive at conflicting theories?  The explanation is that, although we all rely on the scientific method, we arrive at different judgments about which theores are sanctioned by that method. The relevant principle of scientific method runs something like this: ceteris paribus, Theory T is superior to Theory T' if it is simpler, or more general, or stronger than T' relative to the combination, or some weighted combination, of the various "systematic virtues," or adopting T would require abandoning, or modifying, fewer, or less fundamental, previously held beliefs than would adopting T', or .... Not only is there an enormous amount of "slack" in this "principle," but we differ in our weightings of the various systematic virtues, and these considerations might well suffice to account even for ideally rational agents' arriving at conflicting theories on the basis of comparable bodies of evidence.²²

Finally, I want to consider briefly the passages mentioned earlier in which Quine offered his assurances that he was not proposing to

abandon the normative, since they contain what might seem to be yet another puzzle about his conception of naturalized epistemology:

What is new and puzzling here is Quine's use of the term 'engineering', and his apparent suggestion that while engineering is normative (in that it attempts to construct the best device for a given purpose), science is descriptive (and provides engineers with its "findings").  But compare this with two of many similar antecedents in his writings:

Here scientific theories yield predictions.  Of course, in deriving those predictions we are constrained by logical norms, but this is obviously not what Quine has in mind by the "technology" of prediction. What has happened, I suggest, is simply that he has adopted new terminology for making familiar points. What he here calls engineering is simply theorizing-thought of now as devising/constructing scientific theories designed to maximize predictive efficacy.  As for the scientific "findings" made use of in such projects, they are simply whatever theories we take for the nonce to have been established--in accordance with our normative scientific method--along with our observations, especially those that are currently inexplicable or have falsified earlier theories.  On this reading, normative epistemology does indeed fall into place as a part of science; its distinctive tasks are to identify, formulate, evaluate, and perhaps modify the most general norms that

guide our scientific theorizing, and to evaluate our theories in light of those norms; even in its new clothing and setting, this is unmistakably epistemology as traditionally pursued.

Given my proposed reading of "Epistemology Naturalized," several points about the relationship between epistemology and psychology, and Quine's view of it, deserve brief consideration.

Despite its reputation, the essay contains little that is new; it is largely a sustained argument for the idea that much if not all of what goes on under the name of epistemology can be seen to be part of the scientific enterprise broadly-and rightly-conceived. We have seen, however, that, even so conceived, Quine's proposal about how the nature of the evidence/theory relation should be determined is best described as a matter of enlisting the aid of psychology in that investigation. This is certainly a striking example of the occasional direct relevance of psychology to philosophy, and in this it resembles John Rawls's reliance, in developing his theory of justice, on psychological hypotheses about what the agents in his "original position" would do; but neither Quine nor Rawls is aptly described as having surrendered a philosophical burden to psychology.²⁷

At the conclusion of his essay, Quine mentions several theoretical endeavors that, he suggests, might be furthered by "rubbing out the boundaries" between epistemology and science (psychology and linguistics, in particular), but his suggestions are quite puzzling.  He mentions the possibilities of (i) determining whether there is an "alphabet" of perceptual norms toward which we tend unconsciously to rectify all perceptions, which "could be taken as epistemological building blocks, the working elements of experience," (ii) determining whether various features of our cognitive processes can be explained on the basis of their survival value, and (iii) "clarifying" induction by appeal to evolution.  I simply do not understand what he can have in mind in suggesting that (i) or (ii), as interesting as they are, might be furthered by rubbing out disciplinary boundaries, though (i) might be another case of psychology's being an important aid to an epistemological investigation. As for (iii), Quine himself once attempted a justification of induction by appeal to evolutionary theory, but abandoned it (whether rightly or wrongly does not matter here) in the face of Goodman's new riddle of induction, and later described a different role for that theory:

 

 

There is nothing to dispute here, but neither is there anything to suggest that rubbing out boundaries promises theoretical progress. Of course Quine may have had in mind something else altogether, but I am at a loss to imagine what that might be.  In sum, it is hard to see in his brief remarks any promise of theoretical progress to be made as a result of eliminating this boundary.²⁹

Finally, while many, perhaps most, philosophers have seen in "Epistemology Naturalized" the original impetus toward various views now often grouped under the title "naturalized epistemology," and while various such views were doubtless inspired by that essay as it has hitherto been read, not only do some of them have no discernible relation to the proposals actually contained in it, but Quine would have objected vigorously to them.

Given his enormous influence on epistemology, it can come as something of a surprise to realize that he had virtually no interest in much of the work that held center stage in that discipline during his lifetime. We have already noted his lack of interest in mainstream traditional theories, and two related phenomena deserve highlighting: when he suggested ways in which progress might result from rubbing out the boundary between epistemology and psychology, he did not so much as hint at the possible development of "naturalistic" or "psychologistic" accounts of either knowledge or justification, and in later years he showed no interest at all in the proliferating "externalist" accounts of those concepts.

Concerning knowledge, or what he once referred to as "our so�called knowledge,"³⁰ he had next to nothing to say; such little thought as he gave it may well have been captured in his response to the question, "How can we know that [one complete physical theory] is true and [another] false?":