"Is Causal Decision Theory Dynamically Inconsistent?"
Abstract: A number of authors -- including Jack Spencer, Caspar Oesterheld and Vincent Conitzer, and Arif Ahmed -- have recently argued that causal decision theory is dynamically inconsistent in the sense that it requires people to make sequences of decisions that are sure to produce suboptimal outcomes. Most of these arguments involve "Death in Damascus"-like decisions in which no act is ratifiable, i.e., no act looks best on the supposition that it is chosen. I will maintain that none of these arguments succeed. Two key components of my response are that (i) rational agents who cannot influence the choices of their future selves should view their future behavior as they would any other state of the world, and (ii) the seemingly plausible notion that agents are always better off with more options breaks down in these kinds of situations. Along the way I will discuss the concept of a deliberational equilibrium, as well as the notion that rational agents might sometimes need to "just pick" among equally desirable options, and that their picking procedures can be objects of uncertainty for them.