On Behalf of Feldman contra BonJour on Having Evidence

In the January issue of Mind Larry BonJour reviews Conee and Feldman's _Evidentialism_. It is very complementary so I have few bones to pick but toward the end he says he has an objection to a view of Rich's on what constitutes having evidence. I'm not sure its relevantly different from what Rich does consider but I want to defend the view from the objection.

In "Having Evidence" Rich defends the following proposition (if I recall correctly he's said in conversation that he's not too committed to this view, but I like it).

(HE) S has p available as evidence at t iff S is currently thinking of p.

First let me see that given that this is an account of available-evidence-at-t it seems almost trivially true. How could anything but what's occurant at t (even if subconsiously) contrubute to justification *at t*.

At any rate here's BonJour's objection. He says that (7) runs into trouble with "inferential conclusions for which complicated sets of premisses [sic] are required" as well as "many claims that rest on complicated patterns of sensory or memorial evidence." For the first version of the objection I think there is a plausible fairly quick reply, the latter case is more complicated simply because such cases of belief are more complicated already.

So consider a complex inference consisting of, say, a dozen steps. I doubt I can hold any 12 propositions before my mind at any one time, so it seems that there's no way I could ever reach a point of justification for the conclusion (assuming that all the steps in the inference were necessary to support the conclusion). However I think this is merely an illusion. I'll draw an analogy with Lemmon-style proof annotations. In Lemmon's _Beginning Logic_ (the same system is used in Allen and Hand's Primer) you don't just annotate lines on the right, you keep track on the left of all the lines above which the current line depends on. So each line is a proper terminus of a proof. In a really long proof there's no need to remember all the steps precisely, it's enough to know that the current stage of the proof depends only on the lines annotated at the left.

Notice that you don't even need to remember what those lines said. If you reasonably believe that you did the steps correctly then you an reasonably believe the current line (if you also reasonably believe the previous lines were sufficiently well supported). So toward the end of a fairly long proof your actual evidence for the current lemma will not be your reasonably believing the conjunction of all the prior steps in the inference, rather it will consist in apparent memory states to the effect that you've done the proof right up to present. Likewise, at the end you can reasonably believe the conclusion on the basis of your apparent memories of having gone through the proof correctly (and that the premises were well supported). Since your evidence is what supports your beliefs your evidence just isn't the conjunction of the premises, it's the (apparent) memory of having done the proof (the memorial seemings).

I think a similar reply--mutatis mutandis--can be made for the other kind of case.