The use of models and metaphors in developmental biology.

That models in science are of a variety of kinds and fulfil a variety of purposes has been particularly evident over the past 30 years in the field of developmental biology. At the beginning of this period, molecular and cellular mechanisms for developmental events were largely lacking, and theoretical models were developed faute de mieux. This gave way to a time in which a number of competing paradigms were present. Now, there is an increasingly detailed knowledge of underlying mechanisms, at a range of levels from genes through cells to organs and organisms. This is therefore a good time to review the usefulness of model making in general, with reference to this discipline. It is suggested that models are of two main kinds: models of structure and models of function. They fulfil three main roles, each of which operates at different stages of maturity of the field of interest. When the causes of some phenomenon are entirely mysterious, then a model may help generate hypotheses for subsequent testing. Second, when competing biological explanations are available, a model may help discriminate between hypotheses. Third, when a well-established hypothesis is available, models may facilitate use of the hypothesis. However, unlike physics, biology is stochastic and contingent, and cannot be entirely deduced from first principles. Mathematical modellers and biologists must remain in constant communication.