Midgley, Mary. "Reason Eats Itself." NEW SCIENTIST July 23, 2008.

Like the 17th-century philosopher René Descartes and the ancient Greek philosophers, Hume thought that real knowledge had to be infallible. In order to achieve this they had proposed that it is guaranteed, as mathematics is, by logical necessity. But this deductive model is actually only one in great spectrum of patterns that we use in making sense of the world around us. To go back to Franklin's example of death, the mere formal structure of the familiar argument "All men are mortal – Socrates is a man – so Socrates is mortal", does nothing to protect its premises. The bold claim about "all men" has no logical backing; and the doomed attempt to invent one for it gave rise to endless inconclusive worrying about the "problem of induction". Seeing this difficulty, Hume concluded that real knowledge simply didn't exist. If, however, we ask how we actually form such generalizations the answer surely is that we do it not by logic but by pattern-recognition, using a great complex background of comparisons and analogies which we are naturally able to spot and, when necessary, to criticise. . . . Read the rest here: http://www.newscientist.com/article/dn14387-reason-special-reason-eats-itself.html.