In our latest episode, Peter Kail addressed a popular misreading of David Hume’s views about induction—the process of inferring things about the future on the basis of facts about the past. According to this reading, Hume is a skeptic about induction. Let’s distinguish skeptical from non-skeptical views about induction like this:
Skepticism about induction: we are never justified in believing things about the future on the basis of facts about the past.
Non-skepticism about induction: we are at least sometimes justified in believing things about the future on the basis of facts about the past.
Skepticism about induction, then, is the view that it’s never OK to believe that some event will happen in the future just because similar things have happened in the past under similar circumstances. To borrow the example Matt gives in our interview, a skeptic would say that the fact that the sun has risen every day of my life gives me no reason at all to believe it will rise tomorrow. Non-skepticism about induction is just the view that this is false, i.e. that the way things have gone in the past can sometimes serve as evidence of how they will go in the future.
Hume famously says that what causes us to make inductive inferences is not our grasp of a sound principle that we know through observation or through a priori reasoning. Instead, we make inductive inferences because we are habituated to do so because of the way our minds and sensory faculties are wired. For example, I have been exposed over and over again to the “constant conjunction” of two events: my releasing a cup in my hand and it falling to the ground. Every time I let go of a cup that’s in my hand, I see it fall to the ground. All these experiences have conditioned me to expect that, in the future, a cup in my hand will fall to the ground when I let go of it. So what causes us to reason inductively is habit or what Hume calls “custom”.
The popular misreading that Prof. Kail addressed takes this to mean that Hume endorses skepticism about induction. But this conclusion isn’t supported by the text. As Prof. Kail points out, Hume is only trying to make a more limited point about what causes us to reason inductively. In the passages in question, Hume is not saying anything directly about whether or not we are justified in doing so.
That leaves an important question up for grabs: is Hume a skeptic about induction? According to Prof. Kail, the consensus view among Hume scholars is that he is not. In fact, the consensus view is that Hume is a non-skeptic about induction, and that he thinks we are justified in making inductive inferences because the wiring of our brains and our senses allows those inferences to track regularities in nature. The nice thing about the inductive inferences we make is that they’re pretty reliable. I am (almost) always right to expect a cup to fall to the ground when I drop it, and my predictions that the sun will rise the following day have all been right so far. Hume thinks we are justified in making inductive inferences because they are in fact largely successful.
How do we know that this is really Hume’s view about induction? Prof. Kail offered a few pieces of evidence for this reading in our interview. Here is another, longer piece of textual evidence. In An Enquiry Concerning Human Understanding, Hume says:
“Custom, then, is the great guide of human life. It is that principle alone which renders our experience useful to us, and makes us expect, for the future, a similar train of events with those which have appeared in the past. Without the evidence of custom, we should be entirely ignorant of every matter of fact beyond what is immediately present to the senses. We should never know how to adjust means to ends, or to employ our natural powers in the production of any effect.” (Chapter 5, Section 6, my italics)
There are a couple of things about this passage that are telling. The first is Hume’s claim that induction is extremely useful to us. It’s probably fair to assume that induction is useful to us precisely because our predictions about the future tend to be true. The second is Hume’s reference to “the evidence of custom,” which would be a weird turn of phrase for a skeptic about induction. The third is the claim that if it weren’t for custom, we wouldn’t be able to know or make reliable predictions about the future. We would, Hume says, be entirely ignorant about everything that isn’t immediately present to our senses. The implication here is that induction does allow us to know things about the future. Think about it this way: It would be strange to say “if it weren’t for induction, we wouldn’t know anything about the future” if you thought that induction doesn’t give us any reason at all to believe anything about the future! So this passage, at least, seems to support a non-skeptical reading of Hume.
If this is the right reading of Hume, then it sounds like Hume holds a strange set of views about induction. On the one hand, he thinks that we can be justified in making inductive inferences, because those inferences are caused by mechanisms in our minds and senses that track regularities in nature. On the other hand, Hume thinks that we have no basis for thinking that the future will resemble the past! That sounds like a strange combination. How can we hold these two claims together?
It seems like Hume’s views about induction presuppose at least two claims about the justification of belief. The first is the falsity of the position known as internalism about justification. The basic idea of internalism is that whether I’m justified in believing something depends on whether I’m aware of good reasons for believing it, or on whether I could easily become aware of good reasons for believing it through reflection. Suppose I believe that the streets outside are wet. It might be true that the streets outside are wet, but that doesn’t necessarily mean that I’m justified in believing it. If you asked me why I believe the streets are wet and I said, “I believe the streets outside are wet because the little elf who lives in my closet told me so,” you might be inclined to think that I don’t have a justified belief about the streets being wet. On the other hand, if I said “well, if it’s raining, then the streets outside are wet. It’s raining. Therefore, the streets outside are wet,” you would probably be inclined to say that my belief is justified.
If what Hume says about induction is right, this version of internalism has to be false. It has to be the case that we can sometimes be justified in believing something without being aware of the good reasons for believing it. The reason is simple: Hume thinks we can’t have access to the reasons that justify inferring things about the future from facts about the past, because we can’t know that the future will resemble the past.
The second claim that Hume’s account presupposes is the truth of what is known in epistemology as reliabilism, according to which a belief is justified if it was formed by a psychological process that tends to track the truth. The basic idea here is that whether a belief is justified depends on whether it is the product of a reliable process of belief formation, where a reliable process is one that produces a high percentage of true beliefs.
Hume’s account certainly sounds like a kind of reliabilism about induction. On his view, we make inductive inferences as a result of various habits. These habits are formed by mechanisms that reliably track regularities in nature, and the beliefs they cause us to have are justified for that very reason.
Hopefully by now we’ve gotten a clearer idea of what Hume’s position is. If so, we can move on to various challenges that Hume’s account of induction might face. There are plenty of forceful arguments for internalism about justification, and plenty of powerful objections to reliabilism. We can’t canvass those objections here, but take a look at the links below if you’re interested in learning more.
One challenge worth considering here is whether reliabilism already assumes the soundness of inductive reasoning. If so, we cannot justify our inductive inferences by pointing out how reliable they are. That would be assuming that induction is justified in order to justify induction, and that would be begging the question. Why should we think that reliabilism assumes the soundness of induction? Well, again, the basic idea of reliabilism is that we are justified in holding beliefs that are formed by processes that tend to track the truth. But why is that? Why is it that I’m justified in believing something just because the process that led me to believe it is a reliable one? Presumably, it’s because we can expect those processes to continue to be reliable in the future (or in the present case). But how can we be justified in expecting that, unless we are entitled to assume that the future will resemble the past? Since Hume thinks we are not entitled to assume that, it’s not clear how the past reliability of our inductive inferences can justify any new inductive inferences. If Hume is a reliabilist about induction, this seems like a pretty serious challenge to his account.
Links and Other Readings:
Alvin Goldman, Reliabilism.
George Pappas, Internalist vs. Externalist Conceptions of Epistemic Justification.
David Hume, An Enquiry Concerning Human Understanding.