Information Processing and the Evolutionary Ecology of Cognitive Architecture

Abstract

The individual's behavioral response to the environment is a consequence of three processes. Information from the environment must be encoded into a form that can be manipulated and stored in the nervous system. Encoded information is then subject to computational operations that lead to a representation of the environment that forms the basis for decisions and actions. The design of information-processing schemes acts as both a constraint on behavioral response and an aspect of the organism's phenotype that may be subject to natural selection. My experiments with bumblebees suggest that individual workers process information about floral rewards according to expected utility theory. Utility is here defined by a biomechanical function relating floral reward sizes to rates of net energetic gain. This model also suggests that bees process information from successive visits so as to maximize expected short-term (rather than long-term) energetic gain. Short-term computational algorithms may prove advantageous when environments are spatially or temporally autocorrelated or when individuals are subject to short-term memory constraints. Short-term computational algorithms may lead to perceptual biases, for example, underestimation of low-probability events. Observations on the computational rules employed by individuals may reveal aspects of nervous system organization and operation. At the same time, the floral preferences that result from individual decision making may have profound consequences for the evolution of pollinator-plant interactions and ecological community organization.