It's about time for a post with actual postrational content. I'm reading George Ainslie's "Breakdown of Will", which is one of the most postrational treatments of the subject of willpower and decisionmaking that I've seen yet. It occurred to me that a his model of intertemporal bargaining ties decision and prediction in a way that can be modeled by a Bayesian as follows: each individual decision is distributed where is a hidden parameter. If we place a Dirichlet prior on and integrate it out we get the classic "rich get richer" dynamics, with repeated observations being more likely to repeat again in the future. This formally couples the individual decisions. Would this model be easier for a postrationalist to accept than the more intuitive belief that your current choice influences future choices? I'm not sure, since akrasia is not about rational modeling anyway. It does seem to me a fairly intuitive way of formulating the statement "I am the kind of person who acts X in condition Y" which is what Ainslie tries to use to replace "I now choose to do X independently of future choices in condition Y". A potential downside of this formulation is that it hides the role of free will even more than the intuitive statement.

# Intertemporal bargaining and Dirichlet Processes

Leave a reply

Advertisements