How and why are some species so smart? A comment on Rowe and Healy

Examining how individual cognitive variation relates to fitness holds the promise of providing a richer understanding of how cognitive traits evolve (Thornton and Lukas 2012). However, as Rowe and Healy (2014) point out (see also Thornton et al. 2014 in this issue), this endeavor is rife with pitfalls and requires more psychological rigor than has been employed in most studies to date. I agree wholeheartedly with them that arbitrary “problem-solving” tasks should be abandoned in favor of psychologically grounded tests targeting specific cognitive mechanisms (Thornton et al. 2014) and hope that future work will heed their advice in designing tests and accounting for noncognitive influences. Here, I want to focus on 2 additional, interrelated issues: which species should we focus on and what cognitive mechanisms should we examine? Choices of study species will largely be determined by practical considerations such as the need for an adequate sample size. Abundant, tractable, yet seemingly cognitively unsophisticated model systems may prove useful in developing methods and testing general principles. Such animals have long been a cornerstone of experimental psychology and could be tested in the field to illuminate general principles including trade-offs in cognitive processing (e.g., speed vs. accuracy; Sih and Del Giudice 2012) or between cognition and other traits (e.g., Mery and Kawecki 2003) in the face of natural selective pressures. I would argue that yet greater insights may be gained by examining the fitness consequences of cognitive variation in animals whose cognitive abilities appear in some way remarkable. Given the costs of investment in neural tissue, what are the fitness benefits that generate selection for these abilities? Food-storing birds provide the paradigmatic example, as there is strong evidence linking the challenges of retrieving stored food with elevated hippocampal volume and long-term memory retention (see Rowe and Healy 2014). Spatial memory tests may offer similar opportunities in other animals, such as female brood parasites, which must remember the locations of potential hosts’ nests (Reboreda et al. 1996), though the relevant facets of spatial memory are yet to be determined. Here, individual-based studies can help pinpoint which, if any, aspects of spatial memory confer fitness advantages and specify what those advantages may be. Arguably, an even greater mystery is the striking performance of some species not only in a single, ecologically predictable cognitive domain but across domains. For instance, meta-analyses show that some primate species consistently outperform others across different laboratory tasks (Deaner et al. 2006). Among birds, corvids may also exhibit similar cross-context cognitive prowess, with evidence for remarkable abilities in memory, planning, rule learning, inferential reasoning, and aspects of physical and social cognition (Seed et al. 2009). One fruitful avenue of research may be to test the performance of wild individuals across batteries of tasks targeting different cognitive processes (cf. Isden et al. 2013). Although it is difficult to generate straightforward predictions, such work may help elucidate the mechanistic structure of cognition (e.g., the extent to which general mechanisms explain individual performance across different tasks) and understand whether selection acts on cognitive mechanisms as discrete traits or part of an interrelated complex. As Rowe and Healy (2014) remind us, it would be naive to suppose that individual cognitive performance will necessarily correlate positively with fitness. However, exploring the relationships between suites of cognitive measures and multiple fitness components may help build a picture of the costs and benefits of cognitive traits. We have argued that there is much to be learned from psychometrics, a branch of psychology specifically concerned with quantifying individual cognitive differences (Thornton et al. 2014). In particular, psychometric tests aim to reduce the influence of noncognitive factors and generate continuous measures of cognitive performance. These tests typically specify clear criteria as to what constitutes a “correct” outcome, but apparent mistakes will also be informative, for example, in revealing individual differences in sampling strategies (Seed et al. 2012). By uniting psychometric approaches with conceptual and methodological tools from behavioral ecology, we can start to hone in on trade-offs and understand whether and how selection acts on cognitive variation. A.T. was supported by a Biotechnology and Biological Sciences Research Council David Phillips Fellowship (BB/H021817/1). I am grateful to J. Madden for helpful comments on previous drafts of the article.