Inverse Optimal Control Model of Driving Behavior in Depressed Individuals

Poor performance in goal-oriented sensory motor tasks is a common symptom among depressed individuals. However, it is unclear what the underlying causes of these deficits are. Elucidating the underlying mechanisms is an important first step to develop more targeted behavioral interventions. Here, using simple motor-control tasks, we propose an inverse optimal control approach to analyze and factorize performance deficits into two components of subjects’ behaviors: 1) sensory motor speed, 2) reward-processing. In Task 1, subjects with Beck Depression Inventory score ranging from 0 to 36 were instructed to push a joystick as quickly as possible once they observe motion onset of a virtual car. In Task 2, they were instructed to drive a virtual car as quickly as possible and stop it as close as possible to a stop sign. Based on the continuous joystick actions for each individual subject, we estimated perceptual motor efficiency parameters and recovered the underlying reward function that best explained the subject’s behavior. Initial results suggest, that relative to healthy controls, depressed individuals: 1) have deficits in sensory-motor processing speed, 2) have different goals but not significantly different accuracy/effort ratio. The results suggest that inverse optimal control may be a viable computational approach to quantify and factorize the underlying causes of sensory motor deficits in individuals with depression.