A Rational-Ecological Approach to the Exploration/Exploitation Trade-Offs Bounded Rationality and Suboptimal Performance

This chapter describes a rational-ecological approach to derive the processes underlying the balance between exploration and exploitation of actions as an organism adapts to a new environment. The approach uses a two-step procedure: the general environment is first analyzed to identify its invariant properties; then a set of adaptive mechanisms are proposed that exploit these invariant properties. The underlying assumption of the approach is that cognitive algorithms are adapted to the invariant properties of the general environment. The current proposal is that suboptimal performance can be often explained by the interaction of the cognitive algorithms, information samples, and the specific properties of the new environment so that the obtained samples of the environment may provide a biased representational input to the cognitive algorithms. The current approach is applied to analyze behavior in two information-seeking tasks. It is shown that suboptimal performance is often an emergent property of the dynamic interactions between cognition, information samples, and the characteristics of the environment.