ABSTRACT
In the previous two chapters we have looked at attempts to solve or avoid Hume’s sceptical problem of induction. The time has come to show where I think our best hope of a solution lies. First, we need to be clear about what the problem is. Sometimes the problem is stated as that of “justifying induction”. I think this is potentially misleading. There is no one thing “induction” which stands in need of justification. Strawson was right to point out that there are many inductive arguments, some of which are reasonable and others which are not. We also saw that any attempt to construct a super-inductive argument, for instance one using the uniformity premise, is doomed. Any such superinductive argument is a bad inductive argument, as it clearly leads to false conclusions. Secondly, if we mean by “justification” the task of showing that an inductive argument will lead to or has a propensity to lead to true conclusions, then that task is not a philosophical one. Even if it is the case that we can have a priori grounds for thinking that an inductive argument is reasonable (as Strawson claims), we cannot have such grounds for thinking that someone who uses that argument will thereby achieve a high proportion of true conclusions. For that is a contingent issue. The world might be irregular. There may be no laws of nature. Hence any inductive argument might be utterly unreliable. Therefore, no philosophical argument can show that it is reliable. If the proposition that p is a contingent proposition of science, then the philosophical problem is not: Do we know that p? Rather it is: Could we know that p? That is to say, this issue is one of showing that circumstances could exist in which we do have knowledge as a result of using an inductive argument. In addition, of course, we would like to know whether such a possibility is consistent with what we do know of the circumstances we actually find ourselves in.
