Induction About Induction

Induction about Induction

When we reason about the future we use induction. The sun has always been observed to rise, so we infer inductively that it will rise in the future. The skeptic questions the validity of such inferences. But can’t we apply inductive skepticism to induction itself? Is there any justification for believing that we will carry on inferring inductively in the future? True, we have used inductive inference in the past, but does this give us any ground for supposing that we will keep on using it in the future? Just as the sun may not rise tomorrow, so we may not inferthat it will rise. We may cease to use induction as a rule of inference. From the fact that the human mind has used induction in the past it does not follow that it will use induction in the future. Maybe we will start to use counter-induction: instead of inferring that the sun will rise tomorrow, based on its behavior in the past, we will use that same behavior to infer that it will not rise tomorrow (yet it still stubbornly rises). This may not be a good inference, but it doesn’t follow that we couldn’t adopt it. Who is to say that we might not become completely irrational tomorrow? The future of the human mind is no more predictable from the past than the future of the physical world is. Hume says that induction is a human instinct, but by his own arguments it might change into another instinct tomorrow.

            We can therefore apply inductive skepticism to inductive inference itself: we have only induction to justify our belief that we will carry on reasoning inductively—the mind might abruptly change its ways. It might be objected that this is wrong because the mind necessarily reasons inductively. The thought here would be that logic is not optional for anything deserving to be called a mind: to be a mind a thing has to obey the rules of logic. Might we stop using deductive logic tomorrow? Some say that a principle of charity implies that all minds must obey the same logic, i.e. accept the same rules of inference—and that includes our future minds as well the minds of our contemporaries. This would have the result that we can know a priori that our minds will always use the same logical principles. And that would show that in one limited area at least inductive skepticism breaks down—we canknow by reason alone (not by fallible induction) that our future minds will resemble our past minds. This would be analogous to showing that we can know that future material objects will always be extended, since it is in the very nature of material objects to be extended (the same is not true of the sun rising). If it is in the very nature of the human mind to use induction (a necessary truth about the human mind), then we can know that future human minds will always use induction.

            But that argument lacks cogency: is it really plausible that human minds necessarily follow standard inductive logic? People can become logical intuitionists, so why can’t they change their commitment to inductive logic, becoming counter-inductivists? All minds may have to use some logic perhaps, but it doesn’t follow that there is some logic they all have to use. Couldn’t we come across people who simply don’t reason inductively (a tribe of fanatical Popperians)? Maybe they would not last long if they disregarded the past as a guide to the future, but that is not to say that they are inconceivable. So our minds might change so as to cease to reason inductively; we can’t rule that possibility out on conceptual grounds. Thus we only have induction to justify our belief that we will carry on using induction. Inductive skepticism applies to induction itself: just as I can’t justify my belief that the sun will rise tomorrow, so I can’t justify my belief that I will think it will rise tomorrow. The sun might not rise and I might not believe that it will rise. So I don’t have a skepticism-proof reason to believe that I will make inductive mistakes in the situation in which the future doesn’t resemble the past. For by then I might have stopped supposing that the past is any guide to the future. I might have abandoned induction. For all I know, I might stop reasoning inductively two minutes from now, becoming a complete agnostic about the future. Induction certainly can’t persuade me otherwise.  [1]

  [1] We can construct a grue-type case for induction itself. Let “inductive*” mean “inductive up to time t and agnostic after t”: then what evidence do we have that we are not inductive*, as opposed to inductive? At some future time t we might all become counter-inductive. Human behavior is subject to inductive skepticism, including inductive behavior.