The problem of induction concerns a kind of inference that we make. This inference moves from a set of observations to conclusions about the unobserved. For instance, we do this when we make inferences about the future based on what happened in the past. We also make similar inferences about the unobserved past or present based on what we have observed. These are called “inductive inferences” or induction, for short.
Hume’s contention is that we have no more reason for making inductive inferences than we have for not making them.
He argues that induction depends on a principle concerning the uniformity of nature,** Or that nature is governed by unchanging laws namely, that the future will resemble the past. But that principle does not have a deductive basis nor does it have a probable basis, meaning it cannot be based on what we have observed. Attempts to do the former fail Hume’s conceivability test. Attempts to do the latter involve circular reasoning.
Of course, there might be a third kind of reasoning other than demonstrative and probable reasons. The most obvious is a pragmatic reason: we couldn’t get along without these inferences. We considered this along with several other candidates. But none were fully satisfactory.
The pragmatic reasoning, for instance, is a poor basis for believing the principle that the future will resemble the past. This is because what I want to be true is not sensitive to what is in fact true. I want lots of things to be true for reasons that have nothing to do with whether they actually are true. For instance, I want it to be true that I won the lottery. I want that to be true even though (because, in fact) I haven’t won the lottery.
I actually wrote down most of it. But there’s too much! The main points will have to do.
One exception. I’m still intrigued by Jaron’s suggestion that we should insist on a reason for doubting the principle that the future will resemble the past. I think that Kimbia was right to say that this doesn’t give us a reason to believe it, but it might be a sensible way of resisting Hume’s conclusion that we have as much reason to draw one inference as we do another. It’s still tickling my fancy.