## Speaker:

Willem Labuschagne, Department of Computer Science

## Title:

Inductive reasoning

## Location:

Archway 2 - 1:00 pm, Friday 4 May

## Abstract:

Inductive reasoning has endured a history of controversy. After briefly reviewing Hume's argument that inductive reasoning cannot be justified, I consider induction from two contrasting perspectives.

On the one hand there is modern nonmonotonic logic, in which the notion of justification is looser and induction finds a natural framework.

On the other hand there is the profound difference between continuous and discrete models. The recent mathematical result by Hardin and Taylor on the possibility of predicting the values of a real-valued function implies that there are strategies for prediction that do not rely on the uniformity of nature and that are successful more often than not.