Optimal decision-making in inhibitory control
Angela Yu
UCSD
Optimal decision-making in inhibitory control
Inhibitory control, the dynamic ability to modify or cancel inappropriate actions in response to changing task demands, is an important aspect of cognitive function. I will present a normative model of inhibitory control in the stop signal task, in which a prepotent "go" response is sometimes interrupted by a "stop" signal. I will demonstrate that optimal integration of prior expectation with accumulating sensory evidence (Bayesian inference) and continual weighing of the relative value of "go" and "stop" (Bayes risk
minimization) naturally give rise to the array of behavioral phenomena observed in the task, including contextual effects such as statistical frequency and motivational factors. I will also show where some of the necessary computations are carried out in the brain, using model-based analysis of fMRI data. Finally, I will discuss the role of the neuromodulator norepinephrine as a signal for unexpected uncertainty, by showing how the normative model can account for behavioral alteration due to atomoxetine, which is an approved ADHD drug and norepinephrine reuptake inhibitor.