SEX-BIASED COMPUTATIONS UNDERLYING DIFFERENTIAL SET SHIFT PERFORMANCE IN MICE

Cognitive flexibility can be defined as the ability to adaptively shift between choices or strategies based on environmental feedback. Maladaptive cognitive flexibility has been implicated in numerous neuropsychiatric conditions. Understanding drivers of altered cognitive flexibility between individuals can thus be poised to reveal mechanisms of neuropsychiatric risk and resilience. We demonstrate a newly-developed operant touchscreen Set Shift task that permits robust and continuous testing in mice, and use this task to interrogate sex differences in the computations supporting cognitive flexibility. This task requires animals to both commit to a rule-specific choice during periods of relative certainty (stable rule responding) and to flexibly explore alternative options during periods of uncertainty (rule shifts). Using this task, we discovered that female mice completed significantly more rule shifts with fewer errors than males. We next employed a suite of computational models that revealed sex-biased individual differences in the computations underlying cognitive flexibility, including an input-output hidden Markov model to label latent cognitive states. Animals occupy different states as they navigate unsignaled changes in this environment–exploiting the current rule state and exploring to learn the new rule state following a rule shift. The most optimal strategy is to transition between these states as rapidly as possible following feedback. We hypothesized that female mice were faster at at least one of these processes. Our results suggest that following rule shifts, female mice learn the new rule faster and commit to exploiting rule choices sooner compared to males–sometimes because they commit to multiple rules simultaneously.