Hume’s skepticism about induction rests on three claims.
Today, we considered two attempts to stop the argument at the first step. They challenged the need for a principle. Instead, they maintain, we are justified in drawing conclusions about things we haven’t observed on the basis of observations without the need for the Resemblance Principle.
As several of us noted something like the Resemblance Principle seems to be part and parcel of inductive inferences. Without it, we don’t have any reason for thinking that the things we have observed have any relationship to what will happen in the future. Of course, we rarely bother articulating it. But we take it for granted nonetheless. That’s a good reason for thinking that Hume’s first step makes sense. Why would anyone dispute it?
As Harley pointed out, we have at least one very good reason for trying to block the argument. Hume’s skeptical conclusion is ridiculous. Our observations give us no reason for drawing one conclusion rather than another? That’s crazy!
Of course, Hume didn’t deny that the conclusion is crazy. He tried to show that our inductive inferences are based on something other than reasoning. And as we’ll see when we talk about miracles, he thought that past observations do give us good reasons for drawing conclusions about the future. We’ll have to see if he can show that is consistent with his skepticism.
We considered two attempts to dispute the first step. The first attempt attributed a reason for taking the first step to Hume and claimed that reason was not a good one. The second attempt directly denied that the first step is true.
The first attempt rests on two claims:
In response, I disputed the first claim. I said that we need something like the Resemblance Principle for reasons other than making the move from observations to conclusions certain, like it is in a deductive argument.
This is the point of the Misogynist’s Society example. In that example, the membership list gives me a reason to doubt that the future will resemble the past in some way. Specifically, it gives me a reason to doubt that the next person coming out the door will be male, like the previous 499 people coming out the door were.
So what? This shows that we need a connecting principle for reasons other than establishing conclusions with certainty. Rather, our need for the connecting principle comes from the nature of inductive inference. As many of us said, it involves believing that the future will be like the observed past. When we have reason to doubt that the future will be like the observed past, as we do in this example, we have no reason to draw the inference that our future observations will be similar to the past ones. So we need the principle for reasons that have nothing to do with establishing certainty in our conclusions.
The first attempt denied what it took to be Hume’s reason for asserting that we need the Resemblance Principle to make inductive inferences. The second attempt denies that we need the Resemblance Principle at all. It asserts that what it means to have a good reason to conclude that the next A will be followed by a B is that you have a lot of observations in which As are always followed by Bs.
The problem with this is that it does not distinguish between genuine and spurious correlations. You may have seen me at every moment in my life and never seen me in Arkansas. But the observations alone aren’t enough to justify drawing the conclusion that I will never be in Arkansas. You need some reason for thinking that the observations are relevant to the conclusion you seek to draw. In this case, the observations are not a good guide. This would be obvious if you saw me crossing the border into Arkansas for the first time. The billions of observations of me outside of Arkansas did not establish a causal relationship between being Michael Green and not being in Arkansas.
This is the lesson. In order to reasonably draw a conclusion based on some evidence, I have to believe that the evidence is relevant to the conclusion. This is what the example in the previous paragraph suggests. It is also enough to get Hume’s skepticism about induction under way. Once we admit that we have to have beliefs or principles about what observations are relevant for which predictions, we have given up the second attempt and opened the door for Hume’s argument.
Hume’s argument asks us to explain why any evidence from the observed past is relevant to the future.†† Or any unobserved events, past, present, or future. We can give two possible explanations. It might be relevant because it is inconceivable that the future will be different than the past. Or it might be relevant because past observation shows that the future always has resembled the past. You know how the rest of the story goes.
Hume’s argument poses a question about the relevance of a broad category of evidence (things that you have observed) to a broad category of conclusions (about things you haven’t observed). The second attempt tried to close the argument off by ruling the question out of bounds. It maintained that we did not need to have any beliefs about the relevance of evidence: having made past observations is reason to reach a conclusion all by itself. Once you concede that reaching a reasonable conclusion depends on having beliefs about what evidence is relevant, you will have given up this attempt to rule Hume’s question out of bounds.