foundationalism the view that knowledge and epistemic (knowledge-relevant) justification have a two-tier structure: some instances of knowledge and justification are non-inferential, or foundational; and all other instances thereof are inferential, or non-foundational, in that they derive ultimately from foundational knowledge or justification. This structural view originates in Aristotle’s Posterior Analytics (at least regarding knowledge), receives an extreme formulation in Descartes’s Meditations, and flourishes, with varying details, in the works of such twentieth-century philosophers as Russell, C. I. Lewis, and Chisholm. Versions of foundationalism differ on two main projects: (a) the precise explanation of the nature of non-inferential, or foundational, knowledge and justification, and (b) the specific explanation of how foundational knowledge and justification can be transmitted to non-foundational beliefs. Foundationalism allows for differences on these projects, since it is essentially a view about the structure of knowledge and epistemic justification. The question whether knowledge has foundations is essentially the question whether the sort of justification pertinent to knowledge has a twotier structure. Some philosophers have construed the former question as asking whether knowledge depends on beliefs that are certain in some sense (e.g., indubitable or infallible). This construal bears, however, on only one species of foundationalism: radical foundationalism. Such foundationalism, represented primarily by Descartes, requires that foundational beliefs be certain and able to guarantee the certainty of the non-foundational beliefs they support. Radical foundationalism is currently unpopular for two main reasons. First, very few, if any, of our perceptual beliefs are certain (i.e., indubitable); and, second, those of our beliefs that might be candidates for certainty (e.g., the belief that I am thinking) lack sufficient substance to guarantee the certainty of our rich, highly inferential knowledge of the external world (e.g., our knowledge of physics, chemistry, and biology).
Contemporary foundationalists typically endorse modest foundationalism, the view that non-inferentially justified, foundational beliefs need not possess or provide certainty and need not deductively support justified non-foundational beliefs. Foundational beliefs (or statements) are often called basic beliefs (or statements), but the precise understanding of ‘basic’ here is controversial among foundationalists. Foundationalists agree, however, in their general understanding of non-inferentially justified, foundational beliefs as beliefs whose justification does not derive from other beliefs, although they leave open whether the causal basis of foundational beliefs includes other beliefs. (Epistemic justification comes in degrees, but for simplicity we can restrict discussion to justification sufficient for satisfaction of the justification condition for knowledge; we can also restrict discussion to what it takes for a belief to have justification, omitting issues of what it takes to show that a belief has it.)
Three prominent accounts of non-inferential justification are available to modest foundationalists: (a) self-justification, (b) justification by non-belief, non-propositional experiences, and (c) justification by a non-belief reliable origin of a belief. Proponents of self-justification (including, at one time, Ducasse and Chisholm) contend that foundational beliefs can justify themselves, with no evidential support elsewhere. Proponents of foundational justification by non-belief experiences shun literal self-justification; they hold, following C. I. Lewis, that foundational perceptual beliefs can be justified by non-belief sensory or perceptual experiences (e.g., seeming to see a dictionary) that make true, are best explained by, or otherwise support, those beliefs (e.g., the belief that there is, or at least appears to be, a dictionary here). Proponents of foundational justification by reliable origins find the basis of non-inferential justification in belief-forming processes (e.g., perception, memory, introspection) that are truth-conducive, i.e., that tend to produce true rather than false beliefs. This view thus appeals to the reliability of a belief’s nonbelief origin, whereas the previous view appeals to the particular sensory or perceptual experiences that correspond to (e.g., make true or are best explained by) a foundational belief.
Despite disagreements over the basis of foundational justification, modest foundationalists typically agree that foundational justification is characterized by defeasibility, i.e., can be defeated, undermined, or overridden by a certain sort of expansion of one’s evidence or justified beliefs. For instance, your belief that there is a blue dictionary before you could lose its justification (e.g., the justification from your current perceptual experiences) if you acquired new evidence that there is a blue light shining on the dictionary before you. Foundational justification, therefore, can vary over time if accompanied by relevant changes in one’s perceptual evidence. It does not follow, however, that foundational justification positively depends, i.e., is based, on grounds for denying that there are defeaters. The relevant dependence can be regarded as negative in that there need only be an absence of genuine defeaters. Critics of foundationalism sometimes neglect that latter distinction regarding epistemic dependence. The second big task for foundationalists is to explain how justification transmits from foundational beliefs to inferentially justified, non-foundational beliefs. Radical foundationalists insist, for such transmission, on entailment relations that guarantee the truth or the certainty of nonfoundational beliefs. Modest foundationalists are more flexible, allowing for merely probabilistic inferential connections that transmit justification. For instance, a modest foundationalist can appeal to explanatory inferential connections, as when a foundational belief (e.g., I seem to feel wet) is best explained for a person by a particular physical-object belief (e.g., the belief that the air conditioner overhead is leaking on me). Various other forms of probabilistic inference are available to modest foundationalists; and nothing in principle requires that they restrict foundational beliefs to what one ‘seems’ to sense or to perceive. The traditional motivation for foundationalism comes largely from an eliminative regress argument, outlined originally (regarding knowledge) in Aristotle’s Posterior Analytics. The argument, in shortest form, is that foundationalism is a correct account of the structure of justification since the alternative accounts all fail. Inferential justification is justification wherein one belief, B1, is justified on the basis of another belief, B2. How, if at all, is B2, the supporting belief, itself justified? Obviously, Aristotle suggests, we cannot have a circle here, where B2 is justified by B1; nor can we allow the chain of support to extend endlessly, with no ultimate basis for justification. We cannot, moreover, allow B2 to remain unjustified, lest it lack what it takes to support B1. If this is right, the structure of justification does not involve circles, endless regresses, or unjustified starter-beliefs. That is, this structure is evidently foundationalist. This is, in skeletal form, the regress argument for foundationalism. Given appropriate flesh, and due attention to skepticism about justification, this argument poses a serious challenge to non-foundationalist accounts of the structure of epistemic justification, such as epistemic coherentism. More significantly, foundationalism will then show forth as one of the most compelling accounts of the structure of knowledge and justification. This explains, at least in part, why foundationalism has been very prominent historically and is still widely held in contemporary epistemology.
See also COHERENTISM , EPISTEMOLOGY, JUSTIFICATIO. P.K.M.