epistemic rationality See IRRATIONALITY. epistemic regress argument, an argument, originating in Aristotle’s Posterior Analytics, aiming to show that knowledge and epistemic justification have a two-tier structure as described by epistemic foundationalism. It lends itself to the following outline regarding justification. If you have any justified belief, this belief occurs in an evidential chain including at least two links: the supporting link (i.e., the evidence) and the supported link (i.e., the justified belief). This does not mean, however, that all evidence consists of beliefs. Evidential chains might come in any of four kinds: circular chains, endless chains, chains ending in unjustified beliefs, and chains anchored in foundational beliefs that do not derive their justification from other beliefs. Only the fourth, foundationalist kind is defensible as grounding knowledge and epistemic justification.
Could all justification be inferential? A belief, B1, is inferentially justified when it owes its justification, at least in part, to some other belief, B2. Whence the justification for B2? If B2 owes its justification to B1, we have a troublesome circle. How can B2 yield justification (or evidence) for B1, if B2 owes its evidential status to B1? On the other hand, if B2 owes its justification to another belief, B3, and B3 owes its justification to yet another belief, B4, and so on ad infinitum, we have a troublesome endless regress of justification. Such a regress seems to deliver not actual justification, but at best merely potential justification, for the belief at its head. Actual finite humans, furthermore, seem not to be able to comprehend, or to possess, all the steps of an infinite regress of justification. Finally, if B2 is itself unjustified, it evidently will be unable to provide justification for B1. It seems, then, that the structure of inferential justification does not consist of either circular justification, endless regresses of justification, or unjustified starter-beliefs.
We have foundationalism, then, as the most viable account of evidential chains, so long as we understand it as the structural view that some beliefs are justified non-inferentially (i.e., without deriving justification from other beliefs), but can nonetheless provide justification for other beliefs. More precisely, if we have any justified beliefs, we have some foundational, non-inferentially justified beliefs. This regress argument needs some refinement before its full force can be appreciated. With suitable refinement, however, it can seriously challenge such alternatives to foundationalism as coherentism and contextualism. The regress argument has been a key motivation for foundationalism in the history of epistemology.
See also COHERENTISM , EPISTEMOLOGY, FOUNDATIONALIS. P.K.M.