The distinction in biology between homology and analogy is examined for possible application to studies in behavioural pharmacology. It is argued that the concept of homology is central to understanding the 'construct validity' of animal models of human cognition. It is suggested that we capitalize on known correspondences across species in brain structure and development which may mediate homologous behavioural functions. Manipulation of specific receptors in defined areas may be achieved by local and systemic administration of drugs with relatively specific actions; this will complement an alternative criterion of construct validity, based on clinical treatments. This argument will be illustrated by a critique of successful extrapolations across species that are sometimes initiated by work in animals and sometimes by work in humans, especially in the clinical setting. The examples used will include analyses of spatial working memory, spatial attention and attentional set-shifting, studied in rats, monkeys and humans.