Duane asked:

What is induction and why can it be said that this form of reasoning is ultimately circular?

Answer by Helier Robinson

Induction is a fancy name for generalisation. Generalisation might be called animal learning. If your cat is fed cat food out of cans (tins, in Britain) and you do not eat canned food, then the cat will quickly learn that the sound of a can-opener means food, and will come running. The unreliability of this kind of learning would be shown if one day you decide to eat some canned peaches, and so disappoint the cat. Humans generalise in this way, and often very badly, as shown by stereotyping (‘All Canadian men are either lumberjacks or Mounties’) and by superstitions (touching wood or crossing the fingers to avert bad luck).

The word induction is meant to refer to generalisation that is well done, as in formulating scientific laws, but a major problem in philosophy of science is the problem of induction: how to justify induction while condemning superstition and stereotyping. One attempted solution to this problem is to postulate the principle of uniformity of nature: if nature has uniformities then descriptions of those uniformities can safely by generalised; another attempted solution is to say that scientific induction clearly works well and so is justified. Both of these fail because the only knowledge we can have of the principle of uniformity of nature and of the success of scientific induction is inductive, thereby seemingly making induction ultimately circular. However it is not in principle ultimately circular. If, for example, you could prove that Leibniz’ claim that this is the best of all possible worlds (that is, the world of all the underlying causes of phenomena is the best of all possibles) then scientific induction would have to work, since if it did not then the world would be inferior to the best.