Abstract:
The problem of induction seriously threatens any attempt to show the reasonableness of the foundations of science. It is usually argued that if the problem is conceded to the sceptic, then, as Russell famously objected, the only difference between the man of science and the man who believes he is a poached egg is that the latter is in a minority. Recently, some philosophers, most famously John Norton, have rejected the argument that science relies on any universal inductive schema. This scepticism of induction qua universal inference schema is a scepticism shared and anticipated by Popper, who also argues that the problem of induction, as traditionally stated, is insoluble, but unproblematic given that science does not use any universal inductive schema. Given the momentum of Norton’s anti-universalist programme, new ammunition is lent to the Popperian programme. This thesis attempts to resuscitate and amend the Popperian programme, while also attempting a reconciliation with some of the Bayesian ideas from which it has become unfortunately, and dogmatically, estranged. Popperian anti-inductivism is presented in its strongest possible light. Popper’s corroboration formula is mended and relativized, such that theoretical appraisal can be carried out only with respect to competing theories. Lastly, by adopting some popular Bayesian methods, I show that corroboration, a non-inductive method, is able to guide rational action and practical prediction-making in an uncertain future.