M. The Problem of induction
Whenever we say "All A's are B's" we are saying not only that all A's have been B's in the past, but also that all future A's will be B's. Thus we give evidence from the past as reasons for making claims about the future. But will the future resemble the past? How can we know this? This is the problem of induction.
Inductive reasoning presupposes the principle of uniformity, i.e., the belief that scientific laws (like the universal law of gravity) will hold in the future as they have in the past. But the principle of uniformity cannot itself be demonstrated without assuming that it is itself true.
To show that any future even will be like a past even, we have to assume uniformity, i.e., that it will be like a past event, as, e.g., A causing B. Induction cannot be demonstrated; i.e., we cannot KNOW that the future will resemble the past (Hume, Russell and skepticism generally).
Opposing views (from Will, Strawson and others) argue that skepticism disqualifies evidence by allowing nothing to count for induction, i.e., by disqualifying evidence as soon as it applies to the case at hand. When future futures become present and past futures, they are disqualified as evidence for reasoning about other future futures.
The use of the principles of uniformity and induction is justified pragmatically, as is the use of logic.
Without these principles, knowledge would simply be impossible. With them, knowledge is, or at least may, be possible.