Differential participation in cognitive tests is driven by personality, sex, body condition and experience

Failure to participate in a cognitive test may result in sampling biases when measuring inter-individual variation in cognitive performances in both captive and wild populations. This would be problematic if particular classes of individuals consistently fail to participate, skewing data and making generalisations or comparisons difficult. We presented 144 pheasant chicks, raised under standardised conditions, with a battery of cognitive tests to investigate whether sex, body condition or personality traits, measured by differences in latencies to explore a novel object, novel environment or unknown conspecific, predicted individual variation in voluntary participation across 37 test sessions. In general, participation increased across testing sessions, yet patterns of participation differed with sex and body condition. Males with a high body condition were more likely to participate in early test sessions compared to males with a low body condition or females. While participation among males in high body condition was consistent across sessions, males with a low body condition and females, regardless of body condition, were more likely to participate in later, rather than earlier sessions. Individuals also showed repeatable behaviours across time and different contexts, revealing not only that the exploration of novelty, but also that the order that subjects entered the testing arena and their latencies to acquire a freely available meal-worm reward may be considered valid proxies for different personality traits. During each test session, those individuals that were among the first to voluntarily enter the testing arena were more likely to participate in subsequent trials. Moreover, when isolated in the testing arena, individuals that rapidly acquired a freely available meal-worm, positioned on the testing apparatus, were also more likely to participate in a cognitive test. Our findings therefore reveal that sex, body condition and personality traits, along with habituation to the testing paradigms, all play important roles in determining whether or not particular individuals participate in cognitive tests. Sampling biases may therefore misrepresent our understanding of variation in cognitive performance in wild and captive populations, making individual differences in cognition difficult to interpret.