Brainstem arousal systems adaptively shape large-scale cortical interactions for flexible decision-making

Abstract

AbstractMany cognitive tasks require a flexible mapping from specific features of sensory input to motor output. Such flexible input-output mapping is reflected in intrinsic correlated variability of activity within the cortical network that implements the decision process, and might rely on rapid plasticity mechanisms that are under neuromodulatory control. Here, we test for a role of neuromodulators in flexible decision-making by combining brainstem fMRI and pupillometry with time-resolved tracking of feature-specific intrinsic correlations within the human sensory-motor network. Human participants reported visual orientation judgments where the correct responses were contingent upon an active rule that could switch unpredictably. Rule switches evoked brainstem and pupil responses and changes in latent variables of behavior that were quantified with a computational model. Behavioral variables in turn were encoded in pupil dynamics. Brainstem activity and pupil dilation preceded fluctuations of stimulus-action coupling strength within the cortical network that implemented the decision. Brainstem arousal systems may thus instigate a context-dependent reorganization of selective cortical pathways for flexible decision-making.