I think the question that Will raises at 30:00 was answered, at least partially, by a paper in the literature of the evolution of preferences by Samuelson & Swinkels
and was described in Slate magazine by Landsburg
The basic idea is that there is a friction between allowing our minds to do the calculation based on our experience and embedding information into our minds via our reward structure. If evolution could simply give us the information directly in our minds, and our minds were capable enough, then we would only care about the results that directly influence our fitness (i.e. successfully raising a child). Since this is not the case evolution gives us crude information by giving us incentives to do intermediate things, eat, have sex, etc.