Modelling decision-making under uncertainty: A direct comparison study between human and mouse gambling data

Abstract

Decision-making is a conserved evolutionary process enabling us to choose one option among several alternatives, and relies on reward and cognitive control systems. The Iowa Gambling Task allows the assessment of human decision-making under uncertainty by presenting four card decks with various cost-benefit probabilities. Participants seek to maximise their monetary gain by developing long-term optimal-choice strategies. Animal versions have been adapted with nutritional rewards, but interspecies data comparisons are scarce. Our study directly compares the non-pathological decision-making performance between humans and wild-type C57BL/6 mice. Human participants completed an electronic Iowa Gambling Task version, while mice a maze-based adaptation with four arms baited in a probabilistic way. Our data shows closely matching performance between both species with similar patterns of choice behaviours. However, mice showed a faster learning rate than humans. Moreover, both populations were clustered into good, intermediate and poor decision-making categories with similar proportions. Remarkably, mice characterised as good decision-makers behaved the same as humans of the same category, but slight differences among species are evident for the other two subpopulations. Overall, our direct comparative study confirms the good face validity of the rodent gambling task. Extended behavioural characterisation and pathological animal models should help strengthen its construct validity and disentangle the determinants in animals and humans decision-making.