Motto:

"There are none so blind as those who will not see." --

Google+ Badge

Wednesday, September 11, 2013

A Problem for Hume's Problem of Induction

Could Hume consistently believe that his argument to the effect that inductive inferences are not justified is successful? In this post I put forward reasons to think the answer is "no."

Hume, very basically, argued that inductive inferences are not justified because there are only two ways that that they could be supported: Either through a priori reasoning, or through further inductive inferences. A priori reasoning cannot support inductive inferences, because there is no contradiction in the supposition that the course of nature may change, and hence it is possible that it could. Nor could inductive inferences rest on further inductive inferences for their support, for they all rest on the supposition that the course of nature will not change, and cannot support that supposition without begging the question. Hence, inductive inferences are not justified.

Let's see what happens when we apply Hume's Fork to the conclusion of Hume's argument. We get the following result:

           "Inductive inferences are not justified" states either a relation of ideas, or a matter of fact.

If we accept the first alternative, then, if it's true, it only states a relation between certain of our ideas--namely, our ideas of justification and inductive inference--and says nothing significant about matters of fact. It is no more informative, in any real sense, than "a = a". If we accept the second alternative, it states a matter of fact, which according to Hume's own principles could only be confirmed or refuted through induction. If inductive inferences are justified, or justified for the most part, it cannot establish that they are unjustified. And if they are unjustified, they cannot justifiably establish their own lack of justification. Thus it seems that even if inductive inferences are not justified, Hume cannot justifiably believe that they are not. If one is inclined to accept Hume's argument, one should reject Hume's Fork as a false dichotomy; conversely, if one is inclined to accept Hume's Fork, one should either reject Hume's argument as unsound, or conclude that it may be sound, but if so, we cannot know that its premises are true.

2 comments:

Tristan Haze said...

Interesting! A couple of criticisms about perceived leaks in the argument:

'If we accept the first alternative, then, if it's true, it only states a relation between certain of our ideas--namely, our ideas of justification and inductive inference--and says nothing significant about matters of fact. It is no more informative, in any real sense, than "a = a".' - Is Hume really committed to the thesis that all judgements on the 'relations between ideas' prong of his fork are as uninformative as 'a = a'? That seems contrary to his allowance that a book which contains reasoning about quantities may be spared commission to 'the flames'.

'If we accept the second alternative, it states a matter of fact, which according to Hume's own principles could only be confirmed or refuted through induction.' - Does Hume really hold that all judgements on the factual prong of his fork can only be confirmed through induction? What about the judgement that I am currently having an impression?

I think these and similar worried have to be expicitly and convincingly dealt with for your problem to acquire force.

Jason Zarri said...

Hi Tristan,

Thanks for your comment.

Re. your first point:

I think he is committed to it, not in the sense that he would necessarily acknowledge it, but in the sense that his position entails it. For Hume all distinct ideas are separable, so relations of ideas must reduce to identities or partial identities. In the present case, Hume would presumably say that the idea non-justification is included in the idea of inductive inference. Is it really much more informative to be told that something which is included in an idea is in fact included in it, than that "a = a"? Perhaps he could say it is something like a Kantian "judgment of clarification" in that it clarifies what we confusedly thought in the concept already. But my real concern is not whether it could be informative in that sense, but whether it tells us about matters of fact; which, if Hume were to accept this prong of the fork, it certainly does not. If inductive inferences are unjustified solely (roughly speaking) as a matter of definition, why not just re-define what 'unjustified' means?

Re. your second point:

Certainly judgments about immediate experience don't need to be confirmed through induction on Hume's view, but he seems pretty clear that knowledge of empirical matters outside of immediate experience depends on our knowledge of causal relations, which he seems to think can only come through induction. Clearly knowledge that induction is unjustified, if empirical, is not a matter of immediate experience. How the could Hume claim to know about it if not through causal relations / induction?