Implementing Resolute Choice Under Uncertainty

TL;DR

This paper explores implementing resolute choice in sequential decision-making under uncertainty, proposing methods that avoid dominated strategies and Money Pumps, while remaining computationally feasible.

Contribution

It introduces a new interpretation of McClennen's resolute choice based on cooperation among successive Selves, and proposes implementations that prevent negative outcomes.

Findings

Prevented Money Pumps and dominated strategies

Maintained computational tractability

Provided a new interpretation of resolute choice

Abstract

The adaptation to situations of sequential choice under uncertainty of decision criteria which deviate from (subjective) expected utility raises the problem of ensuring the selection of a nondominated strategy. In particular, when following the suggestion of Machina and McClennen of giving up separability (also known as consequentialism), which requires the choice of a substrategy in a subtree to depend only on data relevant to that subtree, one must renounce to the use of dynamic programming, since Bellman's principle is no longer valid. An interpretation of McClennen's resolute choice, based on cooperation between the successive Selves of the decision maker, is proposed. Implementations of resolute choice which prevent Money Pumps negative prices of information or, more generally, choices of dominated strategies, while remaining computationally tractable, are proposed.