ABSTRACT
Advocates of the 'causal theory of knowledge' have sometimes shied away from the task of identifying a causal condition of a priori knowledge just because they have seen the causal condition of factual knowledge as a matter of the right causal route from the fact known to the belief, a formulation inapplicable to a priori knowledge. A priori facts have no natural effects. Another consequence of their approach has been the entangling of the causal condition of knowledge with that other objective condition of knowledge, the truth of what is believed. For if the causal condition consists in a relation between fact and belief, it would seem to entail or require that the belief be true. If, however, we keep separate the causal factors which the causal theory tends to lump together (the operations of cognitive faculties, the relations between effects inferred from causes or causes inferred from effects, the passage of information from one person to another and so forth), it becomes possible to find a causal condition of a priori knowledge: roughly, that it should be the product of intelligence. It also becomes possible to see that the causal condition of knowledge may sometimes be fulfilled when the condition of truth is unfulfilled. For example, since the standard sensory illusions arise when the senses are functioning normally and well (even if they fall short by comparison with some ideal mechanism) it can be said that, when they occur, the causal condition of knowledge is satisfied even though the condition of truth is not. Analogously, the reason why belief in a true a priori conclusion arrived at by a fallacious (but ingenious) argument or from a false (but intellectually plausible) premise does not constitute knowledge is not because the causality of the belief is not in general of the right kind, but simply because of the falsity of the premise or the fallacy in the argument. There need be nothing wrong with the intelligence of someone who accepts a sophistical a priori argument; otherwise most philosophers would suffer from defective intelligence. Indeed, only by separating out truth and validity on the one hand from appropriate causality on the other can we avoid defining rational mistakes out of existence and so plunging back into the Cartesian psychology, shared in this respect by Locke, which attributes error solely to such irrational factors as passion and prejudice. That is to revert to the sufficiently discredited notion that reason is in itself infallible, a natural faculty defined by a logical criterion. The more comfortable view is that intelligent mistakes
are possible, and that the deliverances of intelligence, like the deliverances of sense, are defeasible.
