Tapping artificially into natural talents

Abstract

The recent Viewpoint article by Gerlai and Clayton1 declares that ‘only naturalistic studies can inform us about the cognitive processes that are sensitive to natural selection’. In this context they also postulate that ‘artificial tasks are less sensitive to differences between mutant and wild-type animals’ and that ‘a learning paradigm that ignores the behavioural ecology of the species might be swamped by environmental error variation.’ None of these assertions is axiomatic. Not only do the authors never define the meaning of a ‘naturalistic study’ versus an ‘artificial task’, but the tenets of their paper remain ill conceived. In fact, even a cursory glance at the past ∼50 years of research in comparative cognition2 has generated substantial insights about the ‘cognitive processes that are sensitive to natural selection’ by devising largely artificial paradigms for probing defined cognitive functions. Although natural tasks might be useful in identifying phenomena of interest, they often preclude the rigorous control and systematic manipulation of variables that are essential for meaningful investigations of behaviour, as for any other scientific subject matter. Consequently, naturalistic studies have only limited potential in the identification of the specific cognitive or behavioural function altered by, for example, a genetic manipulation. Moreover, data from studies that employ presumably more naturalistic settings that exploit species-specific capabilities (for example, the use of olfactory stimuli in rodents) do not generalize readily to other species. In fact, it is doubtful that we would know much about the often unexpected cognitive capacities of various species, including humans, if research had followed the authors' major claim that only naturalistic studies would reveal those abilities. Likewise, research aimed at identifying the neuronal circuits that mediate high-level cognitive functions would remain severely restricted if researchers did not devise highly artificial, yet valid, tasks for probing such functions in laboratory animals and relating the data from those animals to other species, including humans (for more discussion and examples see Refs 3–5). Thus, the authors' declaration that the discovery of behavioural commonalities between animals and humans requires a focus on naturalistic paradigms is precisely the opposite of an optimal strategy.