How not to be a coherentist

Abstract

Recall the gist of my dilemma. Internalist coherentism states that one's overall belief-system is one's only source of justifying reasons, and that one must have access to one's justifying reasons. The dilemma's first horn is: If the required access is cognitive in BonJour's sense (i.e., essentially a belief-like judgement) and requires justification, then we get a vicious endless regress of required justified beliefs. The second horn of my dilemma is: If the required access is cognitive but does not itself require justification (or even if it is noncognitive), then BonJour's coherentism loses its main motivation as an alternative to foundationalism. BonJour's key anti-foundationalist stricture (SEK, p. 69) is that a belief-like judgement must itself be justified to play an essential role in epistemic justification. BonJour also holds that the access in question is indeed cognitive (pp. 32, 43, 80, 123). BonJour believes he can slip through the horns of my dilemma with help from his Doxastic Presumption: the presumption that one's representation of one's belief-system is at least approximately correct. On BonJour's view the Doxastic Presumption is unjustified and unjustifiable (pp. 106, 147). Can the Doxastic Presumption take us through the dilemma's horns? Surely it cannot. The second horn of my dilemma stems directly from the Doxastic Presumption. The penultimate paragraph of 'Internalism and Coherentism' (ANALYSIS above, p. 163) anticipates an appeal to the Doxastic Presumption (see the pages cited), and it specifies why such an appeal is ad hoc and troublesome. In fact, that very paragraph erects the second horn of my dilemma. So the Doxastic Presumption impales the coherentist on the dilemma's second horn. I have suggested that BonJour's use of the Doxastic Presumption fails to preserve the essential connection between justified belief and adequate likelihood of truth. BonJour's reply (ANALYSIS above, p. 165): 'this result follows only if adequacy requires that likelihood of truth be established in a way which requires no background assumptions of any sort ... [but] any imaginable epistemological view is inadequate when judged by such a standard'. I doubt both parts of this reply. We plausibly can acknowledge an epistemic role for background assumptions so long as they are justified but do not depend for their justification on other beliefs. Also I happen to have formulated a theory of evidential probability (I trust it is at