Learning and Adaptation

MAKING. Edited by ReuvenDukas. Chicago (Illinois): University of Chlicago Press. $95.00 (hardcover); $30.00 (paper). ix + 420 p; ill.; index. ISBN: 0-226-16932-4 (hc); 0-226-16933-2 (pb). 1998. E L Thorndike (1874-1949) was remarkable in two respects: first, he was one of the founders of modern experimental psychology and the study of animal learning. Second, he was the first behavioral scientist since the publication of The Origin of Species to completely ignore Darwin and his ideas. Ever since Thorndike, psychology and evolutionary ecology have taken separate paths. In the introductory chapter of his new book, Dukas hits the nail on the head by saying that those paths ought to intersect. We cannot understand how and why brains function without knowing how they evolved; we cannot understand how animals interact with their environments without understanding their mechanisms of information processing. Dukas brought together a group of excellent scientists whose recent work has been poised at the boundary between these two divergent disciplines, and who are likely to provide important stimuli in the newly budding field of cognitive ecology. Enquist and Arak criticize the classic ethological view that the evolution of communication systems will lead to mutually and optimally tuned signals and receivers. Using artificial neural networks, they point out that biases may be a general property of perceptual systems, and these may represent a strong force in driving the evolution of the many striking biological signals we see. Although the models are convincing, an emphasis on how sensory mechanisms really function would be helpful. There is a vast literature on the little tricks and simple solutions sensory-perceptual systems use to cope with complex problems, and they rarely provide a perfect picture of the real world: almost invariably they involve biases and distortions. The chapter on constraints on information processing (by Dukas) reviews theoretical considerations and evidence from human subjects to show that there might be limitations in attending to several tasks simultaneously and efficiently. The psychology literature, however, also shows that for many pairs of tasks, there are no such tradeoffs, and thus we must demonstrate them empirically. But Dukas is asking extremely important questions such as: are there costs to maintaining long-term memory? Are working memory limitations relevant for foraging? Are there difficulties in remaining vigilant for predators for extended periods of time? Will cognitive constraints favor specialization when prey is cryptic? In his chapter on the evolutionary ecology of learning, Dukas reveals how much we still have to learn. We are now certain that learning, at least in simple forms, is a phenomenon shared by all animals, and that they learn about surprisingly diverse aspects of their lives. It is intuitively appealing to assume that this learning is related to their fitness, but we do not yet understand how learning mechanisms evolve, let alone how they adapt to the particular environmental needs of each species. This is because we need more phylogenetic studies, more data on interindividual variance within species, biogeographical studies, and fitness tests related to learning. Without such studies, all considerations of learning adaptations will remain "just-so stories." Beecher et al. describe the fascinating ways in which young sparrows acquire songs, how they decide which songs to acquire, how they combine elements copied from songs of others to form their own individual repertoire, and how these songs are used to communicate with territorial neighbors. They present convincing arguments that the study of song learning must take into account an animal's socioecological context, and therefore field studies are essential. In the chapter, Cognitive Ecology of Navigation, Dyer first maps out the diversity of problems faced by animals in search of familiar goals. Animal orientation is a mature field, but the neuroethologists and physiologists who studied it were largely inter-