Lawrence Bonjour has dug himself a deep philosophical hole, one about six feet down. He has tried to construct a theory concerning the structure of man’s system of beliefs; but has run into an obstacle of justification. He is forced to salvage his coherentist ideas in order to justify his system of beliefs which is, in and of itself, a justification for accepting new beliefs. As an answer to this dilemma, he proposes the Doxastic Presumption.
The Doxastic Presumption is, for Bonjour, not so much a premise in the logical flow as it is a cognitive practice. He describes it as such realizing that it would, as a premise, require justification. Adjusted thusly, the Presumption states, essentially, that we must assume the veracity of our meta-beliefs; we know that which we believe. This practice, Bonjour claims, is automatic and, because of this, is reliable. He understands that we can question the beliefs in our system, but he feels that the meta-beliefs are not doubtful.
This, unfortunately, is where Bonjour stumbles into that hole which he has dug. This principle requires one to swallow too big a pill. It ask man, who has been staggering through the justification of his cognitive system, to justify that system with an assumption, itself unjustified. Why not simply assume the veracity of one’s initial system and save a few precious memory cells? There is little reason to entertain the Doxastic Presumption when it lacks veracity; when it just “assumes” veracity. Certainly, it is human practice to follow the Presumption. Does, however, humanity always have correct practices? It is conceivable that philosophy’s main barrier to answers is this practice. Could not the most important target of inquery be our meta-beliefs? Or is it impossible that man is mistaken about them? He is, after all, so certain of other “truths.”
Therefore, though the Presumption is a noble effort by Bonjour to escape his unpainted corner, it simply asks too much. It is nothing but an analysis of one aspect of humanity; but, to save himself, Bonjour turns it into an infinite regress arrest. Its failing poses a serious threat to the continued acceptance of externalist coherentism. If one can not even be sure about that which he believes, then a system of these beliefs will not be very trustworthy.