Item

Abstract

When suitably freed, subjects generally elect to divide their time between work and leisure, rather than consumingjust one. Existing accounts of this division have largely characterised behaviour at a macroscopic level of detail,ignoring the wealth of regularities present in the microscopically fine temporal patterning of responding. Recently,as a substantial extension to Niv et al (2007)’s work on the vigour of responding, we posited a general optimalcontrol framework in which subjects make stochastic, approximately-optimizing microscopic choices: whetherand when to work or engage in leisure, and how long to do so for (Niyogi et al, CoSyNe 2012). They can gain benefits from both work and leisure, but pay an automatic op- portunity cost for the time they allocate. Thisframework spans computational questions, such as the optimality of behaviour; algorithmic issues such as thenature of the response policy; and provides a detailed tie to the neural mechanisms of free operant choice. Here,(a) we show that the critical construct for fitting behaviour is the function quantifying the benefit of leisure as afunction of leisure duration; (b) we reverse engineer the empirical benefit of leisure functions in a number of rats,demonstrating that they are non-linear; and (c) we explore the formal relationships between this account and thetraditional macroscopic accounts.