Flexible sensorimotor linkage during decision-making in changing environments

Abstract

In the present study we probed into the dynamics of hierarchical decision-making, necessary for adaptive behavior in a setting in which the state of an uncertain and unstable environment determines the best behavioral strategy for more basic sensorimotor decisions. We combined a novel decision-making task with fMRI and pupillometry. Through sophisticated Bayesian computational modeling of behavioral data we showed that the participants' behavior was well captured by a dynamic belief updating process regarding the state of the uncertain environment and we gained access to continuous estimates of computational quantities of this updating process. By relating these computational readouts to fMRI signal through general linear modeling and model-free analyses, we identified i) brain regions in which activity reflected important components of the belief updating process and the basic sensorimotor decisions as well as ii) cortical brain areas that were significantly more active in trials that required a coupling of the basic decisions to the belief updating process. Critically, brainstem imaging revealed that activity in certain neuromodulatory brainstem centers was also enhanced during such trials that required a coupling. Finally, pupillometry allowed us to relate pupil diameter - an established proxy for brainstem arousal systems - to change-point probability, a measure for the probability that the underlying environment has undergone a state change.