We often would like to determine whether we are justified in believing some proposition P. Take this P to be the statement, “It is sunny outside today.” If we find ourselves inside of a room, how would one of us go about justifying this claim? You might declare the following proposition P’, “Because I looked through the window, I am justified in believing that it is sunny outside today,” from which the first proposition P should follow. But we would also like to know how we are justified in believing P’, or in this case why having looked through the window constitutes a proper justification for P. This seems to lead to a problem, as while we might begin to discuss how a window offers a view to the outside world, and how our eyesight can be taken as generally reliable under normal circumstances, we can also demand justification for these further P’’ claims, which should lead us to question whether this chain of justification will ever end.

For Descartes, all statements are ultimately grounded in self-evident truths, most namely the cogito ergo sum, out of which all other possible true beliefs rationally follow. Some modern theories follow a more modest approach toward determining their foundational truths by basing them on our potentially fallible empirical observations, which they claim can be taken as self-justifying in so far as they are to be generally accepted as necessarily passively pressed on to the mind, at least until we obtain stronger evidence in any specific case leading us to believe otherwise. This could involve situations where we are determined to be in a state of hallucination and so the perceptual system then necessarily cannot be taken as trustworthy as it normally would. In this way we wind up with foundational beliefs concerning not only one’s own subjective mind, but also the external world, which also manage to avoid the sort of arguments made against infallible truths found in strong Foundationalism such as Descartes’. Contrarily, Coherentists argue that there are no such basic, non-inferential truths at all, and that every belief is justified through the degree of coherence it shares with the rest of one’s entire belief system. In this way, even one’s most basic perceptual beliefs would still be justified through their relation with the general reliability of all our past experiences, among the other relations it shares with the rest of one’s beliefs.

One response to the justification regress problem is to deny its significance entirely. Coherentism holds that a belief is justified as long as it belongs to the coherent set of beliefs. No member of this set is taken to be justified prior to any other as in the case of Foundationalism, instead they all equitably offer justification for one another in so far as they together form the most coherent set of beliefs. The problem for the Foundationalist is that they have difficulty providing a reason for determining which beliefs they choose to hold as being foundational, as this would seem to entail performing some sort of justificatory process. It always seems appropriate to ask for the justification behind a claim, even those of which Foundationalism would hope to claim are self-justifying.

We only hold the perceptual system to be generally reliable because its past reliability offers justification for this belief, and all of our other beliefs seem to be causally related to it in certain ways. It does not seem like we could even hold the perceptual system as being generally reliable if it weren’t for its relation with certain other rational structures present before the mind, as on its own we realize we are subject to entirely passive impressions. It is only through the patterns and regularities which we happen to observe in nature that allow us to derive any significance out of our perceptions, and were it subject to random static noise it seems unlikely we could form any foundational perceptual beliefs out of it at all. Coherentism holds justification for such perceptual beliefs in exact relation with justification for any other sort of belief in that they are all subject to a nonlinear process of justification in relation to the rest of the entire set of their coherent beliefs. In this way, the regress argument dissolves altogether, as justification for any individual belief is always determined through a single measure of its relation with the rest of the set as a whole.

From a perspective outside of a Coherentist theory of justification it seems that we could form coherent sets of belief which entail interrelations that work very well to justify one another, but which share no relation with the world whatsoever. We might have a proposition P which supports a proposition Q and vice versa, but neither of which are grounded in reality in any way. If this is the case, then what worth could such a theory of justification behind potentially absurd sets of arbitrarily self-consistent beliefs hold? Foundationalism argues that Coherentism always relies on just this sort of fallacious circular reasoning in order to justify its beliefs. Coherentism responds that the most coherent set of beliefs will always cohere best with the world, as this is entailed by one’s conception of the world being made up specifically of the most coherent set of beliefs in the first place. Justification under Coherentism does not revolve in a circular pattern, but is rather a holistic process which takes into account the entire set of beliefs as a whole in determining an individual belief’s degree of justification immediately.