Motivation, Learning, and Motivated Learning

Abstract

As many of the papers in this volume illustrate, the behavior of an organism can be influenced by its own physiological state, by the state of the environment, and by the information that the organism has about the state of the environment. In this chapter, I develop a functional (i.e., evolutionary) approach that can be used to both separate and integrate physiology and environmental information, since each is connected with changes of behavior as a result of experience. The approach is based on dynamic, state-variable modeling (Mangel and Clark, 1988) which explicitly couples physiology and ecology within the framework of a Darwinian measure of fitness and thus responds to Kamil’s (1983) call to integrate the “optimization approach” to behavior with other methods of ethology and psychology. Functional interpretations of learning require an assessment of the fitness, measured in terms of expected reproduction, of suites of behaviors. The technique used to determine fitness is called stochastic dynamic programming. Ward (1987) gives a simple example of stochastic dynamic programming for habitat acceptance; this example is in fact a special case of the methods developed by Mangel and Clark (1986).