Epistemic Blowback in BonJour

Test-balloon that's open for popping by discerning readers:

In "Can Empirical Knowledge Have a Foundation?" Laurence BonJour made the claim that for any basic belief B to be justified for S, it must have some feature Φ, which makes B "highly likely to be true." [1] He then argues that either a foundationalist either must hold that this principle must be known by the B by "intuitions or immediate apprehensions" or it is not known. Since the kind of basic beliefs we have in mind are empirical, it just wouldn't do to have another empirical belief do that justification, nor would it do to have it justified by inference (on pain of regress). Then BonJour proceeds to remove the bunk from each lemma, leaving the foundationlist in a world of hurt. More could be said about his argument to this end, but to get onto juicier material, I'll skip it.

I've got in mind a tu quoque against coherentists who use this argument, such that, inasmuch as this is a problem for the foundationalist, it's a problem for the coherentist.

Suppose the following picture of coherentism: S's belief B is justified, it is justified by its coherence with his set of beliefs, M. Presumably this is some relation R that B bears to M. Something quite like BonJour's metaepistemological principle for basic beliefs seems to apply to relations like R. There are all sorts of relations beliefs bear to sets of beliefs, and some of them seem to confer justification for the coherentist and others don't (at very least the incoherence relation). Thus, let us say that there is some property γ such that if any R has γ, then a belief bearing R to his set of beliefs is highly likely to be true call this metaepistemological principle 2. But then there are two possibilities for any coherence theory. Either 1) S knows/believes/justifiedly-believes R has γ or 2) S does not.

1) is the interesting disjunct in this post. 2) allows some form of broadly externalist coherentism, which BonJour would be likely to reject for the same reasons he rejected broadly externalist foundationalism. Suppose the coherentist picks 1): then S has a new belief B* (R has γ) that must itself be justified by coherence with M. B* is a new belief that bears a unique relation R* to M. But R* must have γ in order to justify B*. And, consistent with 1) above, S must believe that R* has γ, B**. I think it is clear that this is clear that the coherentist who picks 1) is on an infinite regress. What's more, because each belief bears it's own unique relation to M, B will never "reappear" in this justificationatory regress.

Suppose that the following is true: BonJour's arguments against foundationalism are devastating, and that 2), being externalist, shares in the calamity. If I'm right, then 2) will trap the coherentist in an infinite regress and (providing there are no decisive objections to infinitism) infinitism is the right response to the regress.

Now, that's a lot to suppose - since I'm not sure BonJour's argument against foundationalism is good. But I'm inclined to think, from this reasoning, that if BonJour's right, he has proved more than he intended to (assuming he was trying to argue for coherentism - which I admit, he didn't say he was doing.)

-Bill

[1] "Can Empirical Knowledge Have a Foundation?" in Epistemology: An Anthology ed. Sosa, Kim, McGrath, Blackwell: Malden, MA, 2000., 265.