In his blog post on the problem of induction, Feser accuses Hume’s argument against induction as being self-contradictory. However, I have heard a recent rebuttal that claims Hume's argument is not self-contradictory, like Feser claims. To see if Feser was correct in his analysis, let us turn to examine Hume and Feser's accusation of contradiction.
Hume argued that induction can never be justified, since there are only two means by which something can be justified, neither of which works for justifying induction. The first way in which induction could be justified is via what Hume terms ‘relation of ideas’, and the second way is what he terms as ‘matters of fact.’
An example of a proposition that is justified as a ‘relation of ideas’ would be ‘all bachelors are unmarried’. What makes this proposition justified in terms of the ‘relation of ideas’ is that it is necessarily true by virtue of the fact that the term ‘bachelor’ logically entails the idea of ‘unmarried man’. Justifying induction in this way would require showing that there is a similar relation between the idea of x happening multiple times and an event y that always follows x. However, as Feser notes, there is no such connection; this because it is at least conceivable that an effect does not follow a cause, bread failing to nourish us for example, in a way that it is not conceivable to imagine a married bachelor.
An example of a proposition known as a matter of fact would be ‘the British drink tea’. For the fact that this is true is not because of the idea that being British logically entails that one drinks tea (although it may seem like it), but rather because it is a contingent empirical fact that the British drink tea. To argue that induction is justified by matters of fact would mean that it is a contingent empirical fact that induction has been reliable and therefore should be justified. But, this is equally, if not more, problematic. This is because to infer from the fact that induction has been reliable in the past to the belief that induction is now reliable is to rely on induction and therefore argue in a circle, as it presupposes the reliability that is trying to be proved. Hume hence concludes that our belief in induction is unjustified.
Feser argues that Hume’s fork is notoriously self-refuting, as the principle itself is neither known as a matter of fact nor is it true in virtue of its constituent ideas and known in terms of the relation of ideas. So, Feser says that Hume’s fork ‘is as metaphysical a principle as any Hume was trying to undermine with it, and its very promulgation presupposes that there is a third epistemic point of view additional to the two Hume was willing to recognize.’ He further argues that any attempt by Hume to salvage the fork from this criticism would also salvage induction.
As stated above, some have asserted that Feser has erred in thinking that Hume’s fork is self-refuting. The proponents of this criticism argue that Hume’s fork is actually justified in terms of the relation of ideas. They argue this due to their reading of Hume’s Fork as stating the truism that all propositions are either known a priori or a posterior, and their belief that this truism is justified in terms of the relation of ideas.
Is this rebuttal any good? I don’t think so. For such a criticism has missed how justification in terms of the relation of ideas is understood by Hume, and as such has let the very point of Feser’s response slip under its nose. We can see this when we expand upon what Hume meant by the relation of ideas. In Hume’s thought, the kind of truths that are justified in terms of the relation of ideas are analytical truths. An analytical proposition is one where it is the definition of the antecedent that makes the consequence true; thus, ‘all bachelors are unmarried men.’ is true because the definition of ‘bachelor’ is ‘unmarried man’. It is for this reason that it can be known a priori. Feser’s point is that Hume’s fork is clearly not known because of any of its definitions that constitute it; and because of this implicitly assumes that there is a third way by which things can be justifiably known, and is hence self-contradictory.
It is through this implicit third way of how we might justifiably believe something that induction could be justified. Other propositions that fall under this third definition could possibly include moral truths, truths about the external world, whether other minds exist or not, etc. etc. To argue, therefore, that all Feser is attacking is the self-evident truth that all propositions are justified either a priori or a posterior is to misunderstand Hume and to concede the very point Feser was trying to make.
