Comparative psychometrics: establishing what differs is central to understanding what evolves

Abstract

Cognitive abilities cannot be measured directly. What we can measure is individual variation in task performance. In this paper, we first make the case for why we should be interested in mapping individual differences in task performance onto particular cognitive abilities: we suggest that it is crucial for examining the causes and consequences of variation both within and between species. As a case study, we examine whether multiple measures of inhibitory control for non-human animals do indeed produce correlated task performance; however, no clear pattern emerges that would support the notion of a common cognitive ability underpinning individual differences in performance. We advocate a psychometric approach involving a three-step programme to make theoretical and empirical progress: first, we need tasks that reveal signature limits in performance. Second, we need to assess the reliability of individual differences in task performance. Third, multi-trait multi-method test batteries will be instrumental in validating cognitive abilities. Together, these steps will help us to establish what varies between individuals that could impact their fitness and ultimately shape the course of the evolution of animal minds. Finally, we propose executive functions, including working memory, inhibitory control and attentional shifting, as a sensible starting point for this endeavour.This article is part of the theme issue 'Causes and consequences of individual differences in cognitive abilities'.

Keywords: comparative cognition; construct validity; executive functions; individual differences; inhibitory control; multi-trait multi-method test batteries.

Figure 1.

Mean performance (±s.e.) in the cylinder task (per cent correct out of 10 trials) of the MacLean et al. [24] dataset as a function of species (ranked by overall performance in the cylinder task) and performance in the A-not-B error task (correct/incorrect, 1 trial). Only the performance of individuals that completed both tasks is shown.