No evidence for general intelligence in a fish

Abstract

AbstractThe average mammal or bird has a roughly ten times larger brain relative to body size than the average ectotherm vertebrate. It has been surprisingly challenging to determine how this translates into increased cognitive performance. In particular, it is unclear whether the brain size differences translate into qualitative differences in specific cognitive abilities. Here, we provide a first exploratory study to examine the possibility that the larger brains of endotherms support a different organisation of information processing, rather than specific differences in cognitive processes. In mammals, individual performance across domain‐general cognitive tasks is positively correlated, resulting in the psychometric factor g. The value of g is positively correlated with brain size. We tested wild‐caught female cleaner fish Labroides dimidiatus, known for its highly sophisticated social behaviour, in four ecologically nonrelevant cognitive tasks that have been used to varying degrees to assess g in mammals. Cleaner fish solved three of these four tasks, flexibility (reversal learning), self‐control (detour around an obstacle) and numerical competence (simultaneous two‐choice task), while also providing enough interindividual variation to test for g. They did not perform above chance levels in the fourth task, which tested for object permanence. For the three retained tasks, individual performance did not load positively on one principal component. Furthermore, all pairwise correlation coefficients were close to zero. These negative results contradict a frequent criticism of g studies, which proposes that g is a default result of how brains are designed. Rather, the results provide a first indication that endotherm and ectotherm vertebrates may process cognitive tasks in fundamentally different ways due to differences in brain organisation. Our relatively low number of experiments compared to mammalian studies enhances this hypothesis, as the probability of finding a g factor by chance would have been higher.