The chicken fed every morning has no idea what tomorrow brings.
The sun has risen every day of your life, so you're sure it will rise tomorrow. But why? David Hume argued in 1739 that there's no non-circular way to justify learning from the past. Bertrand Russell's grim illustration: a chicken fed daily by the farmer, until the day he wrings its neck.
Induction is reasoning from what we have observed to what we have not: a woman has walked her dog past the market every Monday, so she will next Monday. We rely on it constantly, and science depends on it. The problem of induction asks what, if anything, makes this reasoning rational.
David Hume, the Scottish Enlightenment philosopher, set out the problem in A Treatise of Human Nature in 1739 and again in his Enquiry concerning Human Understanding. He distinguished truths of pure reason, such as those of geometry, from matters of fact, which we learn by experience. Our knowledge of causes and effects comes from seeing things constantly occur together. But no logical argument proves that unobserved cases will resemble observed ones, and any argument that induction has worked before, so will work again, is itself inductive, and therefore circular. Hume went further and denied we could even call such predictions more probable without assuming that the future resembles the past.
Hume was not the first to notice. The Greek sceptic Sextus Empiricus observed that a general rule built from particular cases is insecure unless you check every case, which is impossible. In India, the materialist Cārvāka school and the ninth-century sceptic Jayarasi Bhatta attacked inference along similar lines. Medieval thinkers such as al-Ghazali and William of Ockham asked how we could be sure the world would keep behaving as expected when God could change it at any moment.
Hume's own answer was practical rather than logical: we reason inductively out of custom or habit, and we could not live otherwise. The philosopher C. D. Broad called induction 'the glory of science and the scandal of philosophy'. Karl Popper responded that science does not really use induction at all but proposes hypotheses and tries to falsify them. In 1955 Nelson Goodman added a 'new riddle' with the made-up word 'grue', showing that even deciding which patterns to project is not straightforward.
Source: Wikipedia — Problem of induction · Text summarised from Wikipedia (CC BY-SA 4.0)
