2026/07/01 by Harry F. Suter, Andrew N. Radford, Alasdair I. Houston · 1 voice
Psychology · Computer Science · Decision Sciences · #Behavioral Health and Interventions #Reinforcement Learning in Robotics #Game Theory and Applications
paper · doi:10.1098/rspb.2026.0488
To measure self-control in non-human animals, subjects are typically tested on how long they will wait for a larger-later food reward instead of taking a smaller-sooner food reward. Animals that wait longer are deemed to have exhibited better self-control. We develop an optimal foraging model to show that waiting for longer for a larger-later food reward does not always increase fitness whereas minimizing the difference between predicted and observed wait times does-a measure we propose as a novel performance metric. We use our model to generate qualitative predictions that capture trends in wait times observed in empirical studies across species. Moreover, we show that, all else being equal, waiting time increases with basal metabolic rate. Together, our results suggest that variation in wait times may be better understood in the context of animals attempting to meet their individual metabolic demands rather than an inability to wait. Our work raises questions about a large body of empirical support for key hypotheses in cognitive evolution, while our novel performance metric provides a framework for future empirical work to compare self-control appropriately between species.