How does cognition determine an individual's fitness? A systematic review of the links between cognition, behaviour and fitness in non-human animals

Abstract

Recent efforts to understand the evolution of cognition have adopted a within-species approach where cognitive performance, and its effects on behaviour, is correlated with fitness benefits and, if related, is assumed to be selected for. We reviewed 45 studies taking this approach, involving 26 species and describing 211 relationships between behavioural measures of cognition and fitness, to explore broader patterns underlying the evolution of cognition. First, we explored patterns in the strength and direction of selection. We found generally weak support for a relationship between cognition and fitness, with >70% of raw published relationships being statistically non-significant, with an even smaller likelihood once co-variates were accounted for. Where significant relationships were found, they were predominantly, but not exclusively, positive, with individuals exhibiting faster learning or more accurate memory also displaying greater (proxy) fitness. Second, we tested how selection might act under different circumstances. A relationship with fitness was more likely when general, rather than specific cognitive entities were considered, and when the fitness measure corresponded to survival rather than reproductive output. Consequently, the study of within-species cognitive evolution remains in its early stages, with evidence that is both incomplete and inconclusive. However, it potentially offers a powerful opportunity to explore the structure of cognition, trade-offs, constraints and the way that it links to behaviours.This article is part of the Theo Murphy meeting issue 'Selection shapes diverse animal minds'.

Keywords: cognition; cognitive ecology; cognitive performance; fitness; reproductive output; survival.

Figure 1.

Summary of literature search and selection of studies for use in this review.

Figure 2.

Probability of a test finding a significant effect derived from the fitted Any Effects model. (a–c) The predicted probability of a test reporting a significant effect given (a) the cognitive level tested, (b) the sample size, and (c) the fitness proxy used. (d) The difference in predicted probability of reporting an effect at each cognitive level tested for tests using survival versus reproduction fitness proxies. Predicted differences above 0 indicate that the probability of a study reporting a significant effect is higher when survival rather than reproduction is used as the fitness proxy. In each panel, points/lines show the posterior mean, thicker error bars/darker ribbons show the 50% credible interval, and thinner lines/lighter ribbons show the 95% credible interval.

Figure 3.

Predicted probability of a test reporting a significant positive or negative effect, derived from the posterior of the fitted Effect Direction model. Points show the posterior means, thicker error bars show the 50% credible intervals and the thinner error bars the 95% credible intervals.

Figure 4.

Depiction of how the terms used in this paper referring to cognition relate to one another. Terms in bold are defined in the Glossary at the end of the article.