Classical foundationalism maintains that all knowledge and justified belief rest ultimately on a foundation of knowledge and justified belief that has not been inferred from other knowledge or belief. Because the classical foundationalist typically assumes an account of knowledge in terms of justified or rational true belief, it might be best to focus on the distinction invoked between inferentially and noninferentially justified beliefs. What is written in this entry will apply mutatis mutandis to the distinction between inferential and noninferential knowledge.
The Principle of Inferential Justification
If one thinks about most of the beliefs one takes to be justified and asks what justifies them, it seems natural to answer in terms of other justified beliefs. A person's justification for believing that it will rain, for example, may consist in part of that person's justifiably believing that the barometer is dropping rapidly. But under what conditions can one justifiably infer the truth of one proposition P from another E ? The classic foundationalist typically insists that to be justified in believing P on the basis of E one must be justified in believing E. So, for example, one cannot be justified in believing that the world will end tomorrow by basing that belief on an unsupported hunch that the earth will be hit by a giant meteor. More controversially, many classic foundationalists—at least implicitly—also seemed to presuppose that to be justified in believing P by inferring it from E one must also be justified in believing that E confirms (makes probable) P (where E 's entailing P is the upper limit of E 's making probable P ). Thus, one cannot justifiably infer the arrival of Armageddon from a fortune-teller's prediction that the world will end tomorrow unless one has some good reason to believe that the fortune-teller's predictions make probable the occurrence of the events predicted. Call the principle stating both of the above requirements for justification the principle of inferential justification (PIJ):
To be justified in believing P on the basis of E one must be: (1) justified in believing E ; and (2) justified in believing that E makes probable P.
The principle of inferential justification is a crucial premise in the famous regress argument for foundationalism. If the principle is correct, then to be justified in believing some proposition P on the basis of some other evidence, E1, one would need to be justified in believing E1. But if all justification were inferential, then to be justified in believing E1 one would need to infer it from something else E2, which one justifiably believes, and so on ad infinitum. This first regress is generated invoking only clause (1) of the principle of inferential justification. If the second clause is correct, the potential regresses proliferate endlessly. To be justified in inferring P from E1 one must justifiably believe not only E1 but also that E1 makes likely P, and one must infer this from something else F1, which one must justifiably infer from some other proposition F2, which one justifiably infers …. And so on.
But one must also justifiably believe that F1 makes likely that E1 makes likely P, so one must justifiably infer that from some other proposition G1, which one justifiably infers …. And so on. If all justification were inferential then to justifiably believe any proposition P a person would need to complete not just one but an infinite number of infinitely long chains of reasoning. However, the human mind is finite and cannot complete infinitely long chains of reasoning. To avoid the absurd conclusion that people cannot ever be justified in believing anything whatsoever, we must suppose that some beliefs are justified without inference and that these noninferentially justified beliefs ground the justification of all other justified beliefs.
The principle of inferential justification is also often a critical assumption of classic skeptical arguments, most of which presuppose a strong form of foundationalism. So, for example, Hume seemed to conclude that we have no reason for believing any description of an external world when ultimately all we have to rely on as evidence is our knowledge of fleeting and subjective experience. The problem, Hume argued, is that we have no way of establishing sensations as reliable indicators of the existence of external objects that they take to be their cause. Indeed, the difficulty of avoiding a fairly radical skepticism within the constraints of classical foundationalism is one reason so many philosophers became disillusioned with the view.
Classical foundationalists refer to the foundations of knowledge and justified belief in a variety of ways—for example: noninferentially justified beliefs, self-evident truths, directly evident truths, incorrigible beliefs, infallible beliefs, and so on—but there is no consensus on what confers foundational status on a belief. Some, following Descartes, seek foundations in beliefs that do not admit of the possibility of error. As will be seen, the possibility in question may be interpreted in a number of different ways, but classical foundationalists usually invoked a very strong concept of possibility: If a belief is foundational it must be inconceivable that the belief be false. The having of the belief must somehow entail its truth. Thus Descartes famously purported to find an ideal foundation for knowledge in one's belief that one existed. It seems trivially true that if someone S really does believe that he or she exists, that belief couldn't possibly be false. S has to exist in order to believe that S exists (or to be in any other conscious state).
Still other foundationalists sought to identify noninferential justification with whatever fact is the truth-maker for the alleged noninferentially justified belief. So, for example, some foundationalists would claim that my justification for believing I am in pain—when I am—is the pain itself. Of course, such a view hardly qualifies as a philosophical theory until its proponent gives a principled account of how some truth-makers justify us in believing the claims they make true, whereas others do not.
Although it was not always spelled out, many other classical foundationalists sought the source of foundational knowledge in some relation (other than belief) obtaining between a believer and the truth conditions of what is believed. One metaphor often invoked is the concept of acquaintance. When one believes that one is in pain when one is in pain, for example, one is directly acquainted or confronted with the pain itself (the very state that makes true the proposition believed). It is the knower's direct confrontation with the relevant aspect of reality to which the truth in question corresponds that obviates the need for any inference. Another variation on the view might insist that noninferential justification consists not just in acquaintance with the fact that is the truth-maker for one's belief but also acquaintance with the correspondence between the truth bearer (sometimes taken to be a thought or "picture" of reality) and the truth-maker.
In addition to direct acquaintance with contingent facts that can yield noninferentially justified beliefs in empirical propositions, there may also be direct acquaintance with logical relations holding between propositions, states of affairs, or properties that yields direct knowledge of necessary truths. So, for example, one might claim that one's noninferential justification for believing that squares have four sides is constituted in part by one's acquaintance with the properties of being a square and having four sides and the way in which the former contains the latter. Or one might hold that one's noninferential justification for believing that nothing can be both red all over and blue all over at the same time is constituted in part by one's acquaintance with the way in which being red excludes being blue.
On the above view, one might locate the source of both a priori and a posteriori foundational knowledge in the same relation of acquaintance. Traditionally, philosophers have made a great deal of the distinction between a priori knowledge (knowledge of necessary truths that is in some sense independent of sense experience) and a posteriori knowledge (knowledge of contingent truth that somehow relies on sense experience). But it is hard to see in what sense knowledge of one's own beliefs, for example, fits neatly into this traditional way of making the distinction. That one believes that it will rain tomorrow is a contingent truth that one knows, but it doesn't seem that one's knowledge of that truth depends on sense experience. On the acquaintance theory, the difference between a priori and a posteriori knowledge might better be thought of as lying more on the side of the relata of the acquaintance relation than on the source of the knowledge.
Classical foundationalism has come under considerable attack from many different directions. The second clause of the principle of inferential justification is particularly controversial. A worry exists that it is far too strong a requirement for inferential justification and may simply invite a vicious regress. In assessing the claim that inferential justification requires access to a probability connection between one's premises and one's conclusion, it is important to make sure that the arguments one considers are not enthymematic. as we ordinarily talk, it is natural to describe the dark clouds overhead as evidence of an approaching storm. But it is doubtful that the the presence of the clouds by itself constitutes the entire body of evidence from which people predict the storm; it is the dark clouds together with one's knowledge of a past association between dark clouds and storms. One might argue that when one considers genuinely non-enthymematic reasoning it is less plausible to suppose that one needs knowledge of connections between premises and conclusion in order to legitimately infer one's conclusion.
Still, even in the case of deductively valid arguments there is a great deal of plausibility to the claim that one cannot get justification for believing the conclusion of the argument unless one not only has reason to believe the premises but also sees the connection between premises and conclusion. To avoid regress, people need noninferential knowledge of connections between premises and conclusions; and whereas it may not be that hard to convince oneself that one can discover without inference that one proposition entails another, it is much harder to convince oneself that one can just "see" probability connections (connections that are lower than entailment).
Without noninferential awareness of probability, however, skepticism looms on the horizon. Of course, in deciding what one can or cannot be noninferentially justified in believing, the question of just what might constitute noninferential justification needs to be addressed.
what constitutes noninferential justification?
Some would argue that the search for infallible beliefs as the foundations of knowledge is both fruitless and misguided—at least if infallibility is understood in terms of a belief's entailing the truth of what is believed. As has been shown, there are trivial examples of beliefs that do entail the truth of what is believed. My beliefs that "I exist; that I am conscious; that I have beliefs" are all trivially infallible in the sense defined. Critics have pointed out, however, that if one believes a necessary truth, one's belief will also trivially entail the truth of what is believed. If one says that P entails Q when it is impossible for P to be true while Q is false, then every proposition will entail a necessary truth—necessary truths cannot be false.
But surely belief in a necessary truth does not constitute knowledge if the person holds the belief as a matter of pure whimsy. If one becomes irrationally convinced that every third sentence of a book expresses a truth, and by employing this decision procedure for belief ends up, by a remarkable coincidence, believing an extraordinarily complex necessary truth (far too complex for one to even recognize as a necessary truth) it hardly seems right to suppose that one would have any justification whatsoever for believing that truth. Once one sees that the entailment relation between belief and the truth of what one believes is not sufficient for knowledge or justified belief, one might begin to wonder whether it is ever getting at the heart of any interesting epistemic concept.
Still other philosophers have pointed out that beliefs that entail their truth are few and far between, and that if knowledge rests on a foundation of these, then that foundation is precarious indeed. Consider a favorite example of a foundational belief offered by classical foundationalists: the belief one has that one is in pain. Believing that one is in pain seems to be a state logically distinct from the pain. As such it seems always at least conceivable that the belief could occur—perhaps produced by some evil demon—without the pain. for all we know, the brain state causally responsible for one's believing that one is in pain is a distinct brain state from the one causally responsible for the pain. If so, then one could presumably induce belief in pain without producing the pain. Yet if one cannot get foundational justification or knowledge for accepting descriptions of one's own psychological states, then an impoverished foundation indeed exists upon which to attempt to build an edifice of knowledge.
externalist approaches to infallibility
Some contemporary philosophers are sympathetic to the idea of direct knowledge understood in terms of beliefs that cannot be false, but have understood the relevant possibility in causal or nomological terms. Thus the circumstances that produce the belief that P may be causally sufficient for the truth of P. It is not easy to spell out in an interesting way how one might specify the relevant circumstances causally responsible for a belief, but this approach does succeed in raising an alternative to the classical foundationalists' emphasis on conceivability or logical possibility as the relevant concept to employ in defining epistemically interesting concepts of infallibility. It also raises the prospects of a much richer array of propositions being contained in the foundations. Such externalist approaches to understanding infallibility, however, are probably anathema to classical foundationalists, who typically wanted the conditions that constitute foundational knowledge or justification to be conditions to which people have a kind of unproblematic direct access (a desire that might itself raise again the specter of vicious regress).
It was pointed out earlier that one cannot very well identify truth-makers as the source of noninferential justification without giving a plausible account of just what gives some truth-makers a critical epistemic role that most others fail to have. But reliance on the concept of acquaintance to define the concept of foundational knowledge has not fared much better when it comes to contemporary philosophical fashion. The standard line most often taken is that there is no such relation and, even if there were, it would be of no epistemic interest. Foundational knowledge must be knowledge of propositions if it is to yield the premises from which people can infer the rest of what they justifiably believe. But acquaintance with a fact seems to be a relation that has nothing to do with anything that has a truth value. Facts are not the kinds of things that can be true or false. How does acquaintance with a fact yield access to truth? Indeed, can one even make sense of reference to facts independently of truth? Some philosophers would argue that to refer to a fact is just another way of referring to a proposition's being true. If facts are reducible to truths, it would clearly be uninformative to locate the source of noninferential knowledge of truths in terms of acquaintance with facts to which truths correspond.
The claim that acquaintance with facts is not by itself constitutive of noninferential knowledge of truths is one that an acquaintance theorist might grant, however. As was noted earlier, one might introduce a critical role for truth-bearers to play in one's acquaintance theory. To be noniferentially justified in believing some proposition P, one might argue, one must be acquainted not only with the fact that P but the truth-making relation of correspondence between the thought that P and the fact that P. When one has acquaintance with the truth-bearer, the truth-maker, and the truth-making relation that holds between them, one is in a complex state that does just constitute the most fundamental kind of propositional knowledge.
To attack various versions of foundationalism is not, of course, to respond to the regress argument for foundationalism. It has already been noted that some contemporary foundationalists accept the fundamental idea that there are foundations to knowledge but reject classical accounts of what those foundations consist in. As was seen in considering alternative conceptions of infallibility, many externalists identify justificatory conditions for belief with the circumstances producing the belief. Reliabilists, for example, count a belief as justified if it is reliably produced and they allow that a belief might be reliably produced even if the input producing the belief involves no other beliefs. Such reliable "belief-independent" processes can end a regress of beliefs justified by reference to other beliefs. Reliabilist conceptions of noninferential justification also divorce noninferential justification from infallible justification. According to some reliabilists, a justified belief might result from nondoxastic input and be just barely more likely to be true than not.
As was true of those who seek foundations in the causal impossibility of mistake, reliabilists offer the prospect of a greatly expanded class of propositions that might be noninferentially justified. Like other versions of externalism, however, it is not clear that reliabilism succeeds in capturing a concept of justification that would interest the classical foundationalist. The classical foundationalist sought justification that would provide a kind of assurance of truth, and it is not clear how the causal origin of a belief by itself (when one has no access to that origin) could satisfy one's intellectual curiosity.
The Coherence Theory of Justification
Historically, the other main alternative to classical foundationalism was the coherence theory of justification. The coherentist rejects the classical foundationalist's assumption that justification is linear in structure. According to the coherentist, there is no escape from the circle of one's beliefs—nothing can justify a belief but other beliefs. But one doesn't justify a belief by reference to other prior justified beliefs. Rather, each belief is justified by reference to its fit in an entire system of beliefs. When each belief does its part in contributing to a clear, coherent picture of the world, each belief is justified. The coherentist, however, faces a serious dilemma. The coherentist must choose between the view that coherence by itself confers positive epistemic status on the beliefs that cohere, and the view that it is one's awareness of the coherence between one's beliefs that confers such status. If the coherentists embrace the first horn of the dilemma, they are left with a view that seems vulnerable to counterexample. Does one really want to allow that if one consults one's astrologer and comes to believe a set of complex propositions that coincidentally cohere beautifully—even though due to their complexity one could never discover that coherence—the beliefs in question are all justified?
However, if one requires that one must be aware of the coherence among one's beliefs in order to acquire justification for those beliefs, one faces once again the regress that drove so many to foundationalism. To be aware of coherence one must be aware of the fact—that is, have a justified belief—that one has the beliefs one has and that they stand in various logical and probabilistic connections. But how does one come to know what one believes? If one answers in terms of coherence to which one has access the problem just arises again. If one gives oneself unproblematic direct access to one's beliefs and the connections that hold between them, one has simply returned to classical foundationalism.
Armstrong, David. "Is Introspective Knowledge Incorrigible?" Philosophical Review 72, (1963): 427–32.
Audi, Robert. The Structure of Justification. Cambridge, U.K.: Cambridge University Press, 1993.
Ayer, A. J. The Foundations of Empirical Knowledge. New York: St. Martin's Press. 1955.
BonJour, Laurence, and Sosa, Ernest. Epistemic Justification. Oxford: Blackwell, 2003.
Butchvarov, Panayot. The Concept of Knowledge. Evanston, IL: Northwestern University Press, 1970.
Chisholm, Roderick. The Foundations of Knowing. Minneapolois, MN: University of Minnestoa Press, 1982.
Conee, Earl, and Feldman, Richard. Evidentialism. Oxford: Oxford University Press, 2004.
De Paul, Michael, ed. Resurrecting Old-Fashioned Foundationalism. Totowa, NJ: Rowman and Littlefield, 2001.
Descartes, Rene. Discourse on Method and Meditations, trans. Laurence LaFleur. Indianapolis: Bobbs-Merrill, 1960.
Fales, Evan. A Defense of the Given. Lanham, MD.: Rowman and Littlefield, 1986.
Foley, Richard. The Theory of Epistemic Rationality. Cambridge, MA: Harvard University Press, 1987.
Fumerton, Richard. Metaepisptemology and Skepticism. Lanham, MD: Rowman and Littlefield, 1996.
Klein, Peter. "Human Knowledge and the Infinite Regress of Reasons." Philosophical Perspectives 13 (1999): 297–325.
McGrew, Timothy. The Foundations of Knowledge. Lanham, MD: Littlefield Adams. 1995.
Plantinga, Alvin. "Justification in the 20th Century." In Philosophical Issues 2: Rationality in Epistemology, edited by Enrique Villanueva, 43–78. Atascadero, CA: Ridgeview Publishing Co., 1992.
Price, H. H. Perception. London: Methuen, 1950.
Russell, Bertrand. "Knowledge by Acquaintance and Knowledge by Description." Proceedings of the Aristotelian Society, 11, 108–128. Reprinted in Russell, Bertrand, Mysticism and Logic, London: Allen and Unwin, 1963, 152–167.
Sellars, Wilfred. Science, Perception and Reality. London: Routledge and Kegan Paul, 1963.
Richard Fumerton (1996, 2005)