Evolutionary studies of animal behaviour have benefited enormously from the integration of mathematical models and field data. I suggest that the same harmonious combination could transform our theorizing about science. To support this, I try to show how simple models can illuminate various aspects of scientific practice, and also how such models need to rest on empirical psychological and sociological studies. As an illustration of the possibilities and the problems, I consider the issue of public involvement in scientific decision making.

An analogy

Between the 1950s and the 1970s, the integration of mathematical models and field research transformed earlier ventures in natural history into a sophisticated branch of ecology. Recognizing that natural selection affects the morphology, distribution and behaviour of organisms, mathematically inclined biologists formulated hypotheses about the expected characteristics of plants and animals by devising rigorous models about the constraints on, and optimal forms of, organismal phenotypes. Their endeavours drew on the observations of generations of field naturalists and, in turn, their results supplied concepts and hypotheses that a new cohort of observers could take into the field. At its best, the joint activity of mathematical theorizing and painstaking observation proved extraordinarily fruitful. So, for example, in Geoffrey Parker's extended study of the behaviour of dung-flies, precise expectations about the times spent foraging, searching for mates and in copula, have been formulated, tested and refined in an ever-more penetrating and detailed understanding of the natural phenomena (see, for example, Parker, 1978, and the discussion of this work in Kitcher, 1985).