Like structural or physiological traits, the components of cognition can be thought of as solutions or adaptations to problems posed by natural environments. Therefore, a complete understanding of a cognitive function requires an understanding of the problem it was designed to solve. Of these problems, one of the most vexing is the variability found in nature. Such unpredictability would seem to diminish or defeat the value of learning and memory. Of what use is a memory for a good foraging site if that site is as likely to be poor as to be good on later visits? Drs. Lynn Devenport and Jill Devenport have begun their research by establishing the optimal or ideal solution to the variability problem and expressing it in a quantitative model. Put in words, the model says that as time passes and information ages, decisions should be based less and less on recent outcomes (say, a good harvest at a foraging site) and more and more on long-term regularities (the average harvest at that site). The researchers will conduct field experiments to find how closely animals approach this ideal. To do this, they will provision wild populations of ground-dwelling squirrels (least chipmunks) at two or more artificial "patches" of food (unhulled sunflower seeds) whose variability and abundance the researchers control. Some patches are occasionally very productive, but in the long run poor, others are sometimes poor but in the long run quite good, and yet others fluctuate both in the short and long term. Having provided this experience, the researchers impose delays of minutes or days before the animals can resume foraging at the patches. The model specifies how animals should choose among patches if they were to make the best use of their information. Beyond the aim of determining how closely (if at all) animals approach this ideal, the researchers will make the problem complex enough to ensure a plausible correspondence to natural foraging conditions. The project should help identify cognitive mechanisms that minimize uncertainty, and point to their evolutionary origins.
