Investigating Sensorimotor Decision-Making in Depressed Individuals using a Bayesian Approach

Depression is one of the most common mental disabilities worldwide. People with depression may experience impaired cognitive functions, unusual reward and punishment processing, and other executive dysfunctions that affect daily life. Depression has also been associated with alterations in sensory perception, sensitivity, and processing speed. However, the relationship between depression and sensorimotor decision-making under uncertainty remains understudied. Our study aims to understand how individuals with depression process and weigh prior and current sensory information during sensorimotor decision-making. Here, we employed a visuomotor coin-catching task and the Bayesian Decision Theory framework to examine the relative reliance on current information in depressed and non-depressed individuals; specifically, we investigate if and how the reliance changed depending on the uncertainty of prior and current information. Behavioral analyses and computational models revealed that both groups combined prior knowledge and sensory input in a qualitatively Bayesian manner, giving more weight to current information when it was less uncertain and/or when the prior information was more uncertain. In addition, the groups did not significantly differ in their overall relative reliance on current information. Despite these overall similarities, several group differences emerged. Depressed participants responded faster when prior uncertainty was high, whereas non-depressed participants did not show this pattern. Moreover, while both groups adjusted their sensory weights as prior uncertainty changed, non-depressed participants exhibited more than twice the shift in sensory weighting compared to depressed individuals. Model-based analyses further showed that although Bayesian models fit both groups better than heuristic alternatives, model fit was worse in the depressed group, as indicated by higher mean squared errors (MSEs). In summary, these findings suggest that depression does not abolish Bayesian-like integration, but may attenuate sensitivity to changes in prior reliability and lead to faster, but potentially less adaptive, responses under prior uncertainty. Future work should examine how individual variability in clinical symptoms, task demands, and cognitive flexibility may moderate these effects.