Mechanisms of animal learning and memory were traditionally studied without reference to niche-specific functional considerations. More recently, ecological demands have informed such investigations, most notably with respect to foraging in birds. In parallel, behavioural ecologists, primarily concerned with functional optimization, have begun to consider the role of mechanistic factors, including cognition, to explain apparent deviations from optimal predictions. In the present paper we discuss the application of laboratory-based constructs and paradigms of cognition to the real-world challenges faced by avian foragers. We argue that such applications have been handicapped by what we term the 'paradigmatic assumption' - the assumption that a given laboratory paradigm maps well enough onto a congruent cognitive mechanism (or cognitive ability) to justify conflation of the two. We present evidence against the paradigmatic assumption and suggest that to achieve a profitable integration between function and mechanism, with respect to animal cognition, a new conceptualization of cognitive mechanisms - functional cogni-tion - is required. This new conceptualization should define cognitive mechanisms based on the informational properties of the animal's environment and the adaptive challenges faced. Cognitive mechanisms must be examined in settings that mimic the im-portant aspects of the natural environment, using customized tasks designed to probe defined aspects of the mechanisms' opera-tion. We suggest that this approach will facilitate investigations of the functional and evolutionary relevance of cognitive mechanisms, as well as the patterns of divergence, convergence and specialization of cognitive mechanisms within and between species.