What are the choices in an escape room scenario?
Research Unit 1, SCIoI Project 30
A large body of research hypothesizes specific mechanisms for strategy selection and meta-reasoning. A typical model would cast strategy selection or meta-reasoning as rational under some utility. However, a crucial question is what “repertoire of different strategies” is in fact available to humans and animals (citation from the “Ecologically rational strategy selection” proposal). Typical studies as well as formal models of strategy selection often assume a simple decision space (e.g. two discrete choices), in which case problems involving uncertainty can be cast as (non-stationary) bandits or (partially observable) Markov decision process, bounded rationality as sequential information acquisition or limited computational resources, and the computational disciplines readily offer methods to compute rational policies.
However, what are the choices for an animal or human in an escape room scenario? What is the space of decisions and strategies available? We hypothesize that studying rational strategy selection independently from how systems organize (or have evolved) their decision and strategy spaces is fundamentally limited, as the “repertoire of available strategy” is the foundation and major part for enabling ecological rationality.