Naturalized Epistemology

views updated


Naturalized epistemology is the proposal that the theory of knowledge bears a close relation to empirical studies of cognition. The proposal was first made by W. V. O. Quine in his influential article, "Epistemology Naturalized" (1969). Quine is usually interpreted as subscribing to replacement naturalism : epistemology is to be replaced with empirical psychology. But his proposal may well fall short of full replacement.

In his article, Quine distinguishes conceptual studies, which seek to clarify concepts by defining some of them in terms of others, from doctrinal studies, which attempt to establish laws by proving them. The conceptual studies, were they successful, would facilitate the doctrinal ones because clarifying concepts increases the chance that truths that would otherwise go unrecognized will come to be obvious or come to be perceived as "derivable from obvious truths" (p. 70). Quine allows that progress was made in conceptual studies when Jeremy Bentham suggested paraphrasing sentences about bodies in terms of sentences about sensory experience.

Unfortunately, the project of reducing talk of bodies to talk of sensory experience together with set theory, pursued by Rudolf Carnap in The Logical Structure of the World (1928/1967), did not come to fruition. Carnap's later attempts at a rational reconstruction of science abandoned the aim of providing equivalences that would enable us to eliminate the terms of science in favor of sensory terms, and so they did not legitimate science. Carnap's reduction failed, according to Quine, because scientific theories do not have observational consequences except in the presence of collateral scientific theories. In view of the failure of the reduction, Quine proposes that conceptual studies seeking to clarify terms be replaced by an empirical psychology that describes how science is related to experience: "If all we hope for is a reconstruction that links science to experience in explicit ways short of translation, then it would seem more sensible to settle for psychology" (p. 78). This is the first part of Quine's proposal that empirical psychology is to enter into epistemology: Empirical studies of the cognitive development of science are to succeed the earlier reductive conceptual studies.

Regarding the doctrinal studies, Quine notes that it has been clear since Hume's treatment of induction (1739/1978) that we cannot derive scientific theories from sensory observations. Moreover, Quine claims that scientific theories have consequences for sensory experience only in the presence of collateral science (the Duhem-Quine Thesis). So scientific theories are not supported by observation alone. Since support for any scientific theory depends in this sense on further science, there is no reason to persist in the Cartesian stricture that any reliance on empirical science to understand how science is related to observation is circular. And so, for Quine, there is no point in excluding empirical psychology from such an understanding. This is the second part of Quine's proposal that empirical psychology is to enter into epistemology.

But Quine's reasoning here can be challenged on two grounds. He infers from the permissibility of relying on collateral scientific theories to support a given scientific theory that it is permissible to rely on a specific scientific theory, psychology, to understand how the given scientific theory is related to observation. But psychology is generally not the collateral scientific theory on which, according to the Duhem-Quine Thesis, we are allowed to rely for support of a given scientific theory; and the argument from the permissibility of relying on a collateral theory to support a given scientific theory to the permissibility of relying on psychology to understand how the theory is related to observation is not clearly valid. The latter challenge raises the worry that, in moving from the issue of the support of the theory by observation to the issue of understanding how the theory is related to observation, Quine makes room for psychology, but only by changing the subject from the support of the theory to understanding the relation between theory and observation. This challenge does not, however, undermine Quine's argument if he does not propose a full replacement thesis but rather the idea that empirical psychology is to figure in the project of supporting scientific theory.

The naturalized epistemology that results from Quine's proposals thus has two parts. The conceptual studies that attempted to clarify concepts by reduction are to be replaced by a psychology that understands how science is related to observations. The doctrinal studies are also to be altered. Regarding the latter, most commentators have assumed that Quine intends that we replace normative epistemology with a descriptive psychology of the cognitive development of theories. Quine's summary, however, leaves room for a normative as well as a descriptive enterprise:

Epistemology, or something like it, simply falls into place as a chapter of psychology and hence of natural science. It studies a natural phenomenon, viz., a physical human subject. This human subject is accorded a certain experimentally controlled inputcertain patterns of irradiation in assorted frequencies, for instanceand in the fullness of time the subject delivers as output a description of the three-dimensional external world and its history. The relation between the meager input and the torrential output is a relation that we are prompted to study for somewhat the same reasons that always prompted epistemology; namely, in order to see how evidence relates to theory, and in what ways one's theory of nature transcends any available evidence. (p. 83)

This passage could be taken to propose the replacement of epistemology with psychology: "evidence" could have the descriptive meaning of observation rather than a normative meaning, and seeing how theory transcends evidence might be a descriptive enterprise.

But there are other interpretations that make better sense of Quine's argument in "Epistemology Naturalized." He might mean that we are to use psychology to judge the amount of support the observations provide for given scientific theories, but only in light of an assumed epistemology (that is, an account of what support amounts to) distinct from psychology. On this interpretation, the epistemology tells us how far beyond the observations a scientific theory may go before the observations no longer support the theory; the psychology measures how far beyond our observations the theory actually goes; and the epistemology and psychology combine to tell us whether the theory enjoys support. Alternatively, Quine might mean that we are to use psychology to judge how any suggested epistemology fares in light of whether our actual achievement meets its demands. On this alternative interpretation, the results of psychology constrain epistemology. The psychology measures how far beyond our observations our scientific theories go; and a suggested epistemology is rejected if the measured distance between our scientific theories and our observations exceeds the distance the epistemology sets as the threshold for support.

This second interpretation makes the best sense of the text. Psychology contributes to an account not merely of the causal but also of the support relation between observation and theory. Thus, there is continuity between the old task of supporting science by observations and the new task of accounting for the support relation between observation and theory. The interpretation responds to the charge that Quine's argument from the Duhem-Quine Thesis to the permissibility of relying on psychology changes the subject. And the interpretation is suggested by the fact that Quine assumes that the failure of the conceptual reduction of science to observations, or of the doctrinal derivation of science from observations, does not count decisively against a positive epistemic status for science. Without this assumption, Quine would have no reason to propose that epistemology should abandon reduction and derivation for psychology, rather than that we should terminate epistemology with the judgment that science lacks support because reductions and derivations fail despite being necessary for support. On the preferred interpretation, psychology enters after the failed conceptual reductions and doctrinal derivations, but the use of psychology is warranted only by the separate epistemological claim that the success of science is jeopardized by overshooting the observations, though the jeopardy is not so rigid that the failure of reduction and derivation entails skepticism.

Quine's "Epistemology Naturalized" led many philosophers to take seriously the relevance of psychology to epistemology. Although almost all interpreters read Quine as proposing to replace epistemology with psychology, few epistemologists follow Quine in embracing replacement. Many endorse instead a conceptual naturalism : our everyday epistemic concepts of knowledge, justified belief, or rational belief can be defined or clarified in naturalistic terms, where naturalistic terms are usually taken to be the terms of some respectable science, notably psychology. (On a more liberal view, naturalistic terms are simply those not patently normative.)

Most proponents of conceptual naturalism hold that our epistemic concepts are both normative and naturalistic. The motivations for conceptual naturalism are not often articulated, but they presumably include these: the concepts employed by our respectable sciences are our best-understood concepts and thus the best candidates for definitions in terms we can understand and also the best candidates for clarifying definitions; these are concepts with which we cannot now dispense in our intellectual lives, so for now these concepts are clearly available to provide definitions; and these are the concepts we have the best reason to believe succeed in referring to properties that are actually exemplified, so that knowledge defined in terms of them does not turn out inadvertently to fail to obtain.

Perhaps the most popular version of conceptual naturalism has been reliabilism, according to which knowledge is true belief that results from a reliable belief-forming processa process that tends to yield true beliefs (and similarly, justified belief is belief that results from a reliable process). Whether reliabilism is fully naturalistic depends on whether the notion of truth is naturalistic. Reliabilism has been most extensively developed by Alvin Goldman. In Epistemology and Cognition (1986), Goldman divides his epistemology into two parts. The first part is an analysis of epistemic concepts. Both knowledge and justified belief are defined in terms of reliability. Justified beliefs are (roughly) beliefs permitted by some right J-rule system. J-rules license certain cognitive processes, and "A J-rule system R is right if and only if R permits certain (basic) psychological processes, and the instantiation of these processes would result in a truth ratio of beliefs that meets some specified high threshold (greater than .50)" (p. 106).

This analysis is supported by intuitions in narrow reflective equilibrium. The second part of Goldman's enterprise is an attempt to discover which sorts of beliefs are justified given his analysis of justified belief. This would ideally lead to discovering a right system of J-rules, but Goldman regards such an effort as premature, since "Cognitive science is still groping its way toward the identification of basic processes" (p. 181). Instead, Goldman considers candidates for basic processes individually and attempts to discern their reliability or contribution to a high truth ratio. He examines perception, memory, deduction, probability judgments, judgments under uncertainty, and belief revision in light of the findings of cognitive science. It is fair to say that his review of the reliability of cognitive processes is the most detailed and comprehensive yet undertaken. This second part of Goldman's enterprise exemplifies methodological naturalism : that a significant part of epistemology is an inquiry into whether conditions of epistemic status are satisfied in light of empirical cognitive science. Quine, on the interpretation of his views suggested above, is a methodological naturalist in this sense.

Reliabilism is not the only proposed version of conceptual naturalism. Alvin Plantinga (1993) offers a proper function theory of knowledge that is conditionally naturalistic. According to the theory, knowledge is belief that results from the proper functioning of our cognitive faculties. This theory is naturalistic if proper functioning is naturalistic, although Plantinga denies that it is. More than one writer has noted, however, that most analyses of knowledge and justified belief that eschew the label "naturalism" nevertheless meet the requirements of conceptual naturalism just as well as reliabilism does (Foley 1994, Goldman 1994). For example, some coherence theories define justified belief in nonnormative and even naturalistic terms, such as consistency, mutual entailment, and the like. Despite this, versions of reliabilism differ from coherence theories in usually resulting from an inquiry motivated by the desire to define knowledge and justified belief in natural terms. Coherence theories do not usually result from a naturalistically motivated inquiry.

A view consistent with conceptual naturalism is property naturalism : the property of knowledge or justified belief is identical with certain natural propertiesproperties to which respectable science refers. Ruth Millikan (1984) offers a proper function account of knowledge along these lines, for which the relevant science is evolutionary biology. A view entailed by both conceptual and property naturalism is supervenience naturalism : epistemic properties supervene on natural properties. However, it has been noted (Foley 1994) that few epistemologists have wished to deny supervenience naturalism: Roderick Chisholm (1989) allows that justified belief supervenes on nonnormative properties, despite defining it in normative terms. Keith Lehrer (1997) is rare among epistemologists in denying that justified belief supervenes on nonnormative properties.

Within the category of conceptual and property naturalism, certain views are versions of what Goldman (1994) calls substantive naturalism, according to which the defining terms refer to natural processes or to relations between the subject's belief and the environment. Reliabilism would fit this label. But so would John Pollock's (1989) internalist version of naturalism based on classical artificial intelligence, and Paul Thagard's (1992) coherence theory, which understands the acceptability of a scientific theory as involving a connectionist mechanism. In an influential article, Philip Kitcher (1992) emphasizes a version of psychologism as central to naturalism: knowledge turns on the character of psychological belief-forming processes, as opposed to logical or statistical relations between evidence and belief.

Accounts of knowledge as involving or constituted by psychological processes like intuition and demonstration, and of justified belief as involving causal inference, were common in early modern philosophy. Louis Loeb (2002) argues that, in A Treatise of Human Nature (1739/1978), David Hume held a stability theory of justification: justified belief is the result of a belief-forming operation that tends to produce stable beliefs. C. S. Peirce is also commonly regarded as holding a stability theory in "The Fixation of Belief" (Schmitt 2002). But this psychologism was rejected in the late nineteenth and early twentieth centuries in tandem with the rejection of psychologism in semantical theory by Gottlob Frege. Recent naturalism, such as reliabilism and proper function theory, has brought psychological processes back to the fore in accounts of knowledge and justified belief. The chief ground for giving psychological processes a role in justification has been an attack on the "arguments on paper" thesis, which sees justification as turning merely on an evidential relation between the proposition believed and the evidence possessed by the subject (Goldman 1986, Kaplan 1994).

Related to the role of psychological processes is naturalist opposition to idealized epistemology, which derives norms or standards of justification from logic, probability theory, utility theory, or statistics, without attention to human limitations. For example, epistemologists have endorsed norms of rational belief-revision such as the following: we are to avoid contradictory beliefs, and we are to believe the proposition favored by the total evidence available to us. Hilary Kornblith (2002) cites reasons for doubting that we are generally able to guarantee that our beliefs are consistent. Goldman (1986) criticizes the claim that there must be a failure of rationality if one's belief fails to conform to the total available evidence; the fault may lie in one's access to memory rather than in one's reasoning, which is the focus of the evaluation of rationality.

Of course, the norm of avoiding contradiction or of conforming to the total available evidence could be understood as the qualified requirement that we are to avoid contradiction or conform to the total available evidence when we are able to do so. But if it turns out that we are rarely if ever able to satisfy to these norms, there is little plausibility to the view that rational belief-revision requires satisfying, or even being guided by, such norms. Again, the norm of avoiding contradiction could be understood as the requirement that we are to approximate as nearly as feasible (or as cost-effective) to avoiding contradiction. But if it turns out that we are far short of being able to approximate the goal, then it seems there is no such norm. The question concerns the content of epistemic norms and an associated issue of the methodology of identifying norms: Can we formulate norms in ignorance of contingent facts about our cognitive powers, protecting the norm from empirical disconfirmation by making it merely a requirement to approximate a goal, or must we craft norms under assumptions about human limitations that would best be empirically informed (Schmitt 2004)? The naturalistic methodology finds support in the theoretical point that the approximate idealizing view has no means of suppressing epistemic ideals that are intuitively plausible but so demanding that no norm should require approximating them to any degree.

A final issue within naturalism is methodological. Should we conduct epistemology by defining knowledge so as to explain the functions (biological, social, or cognitive) served by our concept of knowledge and practices of epistemic evaluation? Or should we instead identify knowledge with the real properties involved in states we label "knowledge," studying knowledge on the model of a natural kind like aluminum or frog, opening the possibility that knowledge diverges from the properties represented in our concept, and that it serves primary functions quite different from any suggested by the functions of our use of the concept?

An account of the first sort is offered by Edward Craig (1990), who defines knowledge so as to explain the functions served by our applications of the concept. He proposes that our concept has its content in virtue of serving the social-cognitive function of picking out good informants. Craig rests his conditions of knowledge on everyday observations of the function of our concept, but it would be possible to rely on scientific sociology in such a study. An account of the second sort is offered by Hilary Kornblith (2002). He argues, by appeal to studies of animal cognition, that animals possess knowledge, and he defends the view that human knowledge is no different in kind from animal knowledge. In effect he proposes that we infer the conditions of knowledge from the biological functions of the states we label "knowledge." As it happens, Craig's conditions of knowledge roughly coincide with Kornblith's: both are versions of reliabilism about knowledge. But Craig's methodology is incompatible with Kornblith's. For conditions of knowledge inferred from the social-cognitive functions of applying the concept of knowledge need not be coextensive with conditions inferred from the biological functions of knowledge itself. Nothing guarantees that the properties that humans ascribe in order to pick out good informants must be the properties that enable animals to survive in their habitats.

It is a further question whether, given Kornblith's approach and findings, knowledge turns out to be a natural kindfor example, in the sense of a homeostatic property cluster or a cluster of self-maintaining properties. If instances of knowledge are to be instances of a natural kind and knowledge is to be reliable belief, then every instance of knowledge must be a state of a cognitive system in which a variety of reliable processes (perhaps perceptual, memorial, and inferential processes) routinely support one another in producing knowledge. This would seem to be the weakest sense in which the properties essential to an instance of knowledge could be said to be self-maintaining. If so, the claim that knowledge is a natural kind entails two key assertions: that knowledge requires not merely a reliable process yielding the given belief but also that instances of knowledge are embedded in a nexus of reliable processes. We may wonder, however, whether there is any informative condition of embedding in such a nexus that holds for all instances of knowledge across species; if not, there is no general natural kind of knowledge.

See also Carnap, Rudolf; Chisholm, Roderick; Cognitive Science; Epistemology; Frege, Gottlob; Goldman, Alvin; Hume, David; Lehrer, Keith; Memory; Millikan, Ruth; Peirce, Charles Sanders; Perception; Plantinga, Alvin; Psychologism; Quine, Willard Van Orman; Reliabilism; Underdetermination Thesis, Duhem-Quine Thesis.


Carnap, Rudolf. The Logical Structure of the World. Translated by R. George. Berkeley and Los Angeles: University of California Press, 1967. First published in 1928.

Chisholm, Roderick. Theory of Knowledge. 3rd ed. Englewood Cliffs, N. J.: Prentice Hall, 1989.

Craig, Edward. Knowledge and the State of Nature. Oxford: Oxford University Press, 1990.

Dretske, Fred. Knowledge and the Flow of Information. Cambridge, MA: MIT Press, 1981.

Feldman, Richard. "Methodological Naturalism in Epistemology." In The Blackwell Guide to Epistemology, edited by John Greco and Ernest Sosa. Oxford: Basil Blackwell, 1999.

Foley, Richard. "Quine and Naturalized Epistemology." In Midwest Studies in Philosophy. Vol. 19, Philosophical Naturalism, edited by Peter A. French, Theodore E. Uehling, Jr., and Howard K. Wettstein. South Bend, IN: University of Notre Dame Press, 1994.

Goldman, Alvin I. Epistemology and Cognition. Harvard University Press, 1986.

Goldman, Alvin I. "Naturalistic Epistemology and Reliabilism." In Midwest Studies in Philosophy. Vol. 19, Philosophical Naturalism, edited by Peter A. French, Theodore E. Uehling, Jr., and Howard K. Wettstein. South Bend, IN: University of Notre Dame Press, 1994.

Hume, David. A Treatise of Human Nature. 2nd ed, edited by P. H. Nidditch. Oxford: Oxford University Press, 1978. First published in 1739.

Kaplan, Mark. "Epistemology Denatured." In Midwest Studies in Philosophy. Vol. 19, Philosophical Naturalism, edited by Peter A. French, Theodore E. Uehling, Jr., and Howard K. Wettstein. South Bend, IN: University of Notre Dame Press, 1994.

Kitcher, Philip. "The Naturalists Return." Philosophical Review 101 (1992): 53114.

Kornblith, Hilary. Inductive Inference and Its Natural Ground. Cambridge, Mass.: MIT Press, 1993.

Kornblith, Hilary. Knowledge and Its Place in Nature. Oxford: Oxford University Press, 2002.

Lehrer, Keith. Self-Trust: A Study of Reason, Knowledge, and Autonomy. Oxford: Oxford University Press, 1997.

Loeb, Louis. Stability and Justification in Hume's Treatise. Oxford: Oxford University Press, 2002.

Millikan, Ruth Garrett. "Naturalist Reflections on Knowledge." Pacific Philosophical Quarterly 65 (1984): 315334.

Peirce, C. S. "The Fixation of Belief." In The Essential Peirce: Selected Philosophical Writings. Vol. 1 (18671893), edited by Nathan Houser and Christian Kloesel. Bloomington, IN: Indiana University Press, 1992.

Plantinga, Alvin. Warrant and Proper Function. Oxford: Oxford University Press, 1993.

Pollock, John. How to Build a Person: A Prolegomenon. Cambridge, MA: MIT Press, 1989.

Quine, W. V. "Epistemology Naturalized." In Ontological Relativity and Other Essays. New York: Columbia University Press, 1969.

Quine, W. V. "The Nature of Natural Knowledge." In Mind and Language, edited by Samuel L. Guttenplan. Oxford: Oxford University Press, 1975.

Schmitt, Frederick F. "Justification and Consensus: The Peircean Approach," Protosociology: An International Journal of Interdisciplinary Research 16 (2002): 241286.

Schmitt, Frederick F. "Epistemology and Cognitive Science." In Handbook of Epistemology, edited by Ilkka Niiniluoto, Matti Sintonen, and Jan Woleński. Dordrecht: Kluwer, 2004.

Shatz, David. "Skepticism and Naturalized Epistemology." In Naturalism: A Critical Appraisal. South Bend, IN: University of Notre Dame Press, 1993.

Thagard, Paul. Conceptual Revolutions. Princeton: Princeton University Press, 1992.

Frederick F. Schmitt (2005)