Abstract

Awareness growth is a persistent challenge to Bayesian theories of rational belief. Cases of awareness growth include coming to consider a completely new possibility (called expansion) and coming to consider finer distinctions through the introduction of a new partition (called refinement). Recent work has centred on “Reverse Bayesianism”, a proposal for rational awareness growth developed in the economics literature.

This essay develops a Reserve Bayesian (RB) position in light of two challenges. Anna Mahtani argues that RB yields the wrong result in cases where the growth of awareness constitutes an expansion relative to one partition, but a refinement relative to a different partition.

Steele and Stefánsson argue that RB cannot deal with new propositions which are evidentially relevant to old propositions. I show that each objection rests on a lack of clarity about the process being modelled. Modelling awareness change requires specifying how propositions in the agent's old algebra are identified with propositions in the new algebra, which has hitherto not been done.

I introduce a lattice-theoretic model for this purpose, which resolves Mahtani's problem cases. Steele and Stefánsson’s remaining cases confuse questions of awareness change with those of belief revision.  Applying my RB model, followed by a generalised belief revision procedure, resolves Steele and Stefánsson’s cases.