BonJour on Foundationalism

Both the ‘coherentist’ and the ‘foundationalist’ theories of the justification of our claims to empirical knowledge are subject to considerable difficulty, as Laurence BonJour admirably brings out in his book The Structure of Empirical Knowledge. (Coherentism is the view that justification of a proposition is always and exclusively a matter of its coherence with a whole system of propositions; foundationalism, that it is at least sometimes also a matter of its having foundations in things which are not propositions, typically a range of sense-impressions or observations of physical objects.) BonJour argues, however, that while the difficulties encountered by the foundationalist view are insuperable, a version of the coherentist position can be salvaged by considerations which render it proof against the usual objections. I myself maintain exactly the opposite view, which I shall try to defend in what follows. If every empirical belief requires justification by another empirical belief, we seem to be faced with either a vicious circle or an infinite regress; and the sceptical conclusion appears to follow, that there is no reason to think that any empirical belief is true. The foundationalist tries to stop the regress by maintaining that some empirical beliefs are justified in a way that does not depend on inference from other such beliefs. One influential way of doing this (and the one which I shall try to defend) is to maintain that while indeed all empirical beliefs have to be justified, in the case of some of these beliefs, the justification consists of reference to mental states other than beliefs which do not themselves have to be justified; these are described in various ways by various authors, but are usually said to be ‘given’, ‘represented’, or something equivalent.