Title:

Circuit and single neuron computations for flexible decision-making

Abstract:

Making flexible decisions in a dynamic environment is a powerful capacity of the mammalian brain. Despite decades of research at the behavioral and cognitive levels, the biological mechanisms implementing computations underlying flexible decision-making remain largely unknown. Here we developed an inferece-based flexible decision-making task in mice and combined in vivo two-photon imaging, large-scale neurophysiological recording and circuit manipulations to investigate the underlying neuronal mechanisms. Using cross-brain region circuit analysis, we show that the orbitofrontal-sensory cortical circuits implement an inference-based algorithm conferring a high degree of flexibility in rule-switching behavior. Using subcelullar two-photon imaging, we found that layer 5 cortical pyramidal neurons compartementalize different types of task information in dendritic and somatic subcellular domains, suggesting a dendritic integration mechanism for rule-based decisions. Using brain-wide electrophysiology, we further show that the representations for task variables of different degree of abstraction are orderly distributed across hierarchical brain regions. Our results provide new insights to the neuronal and circuit mechanisms implementing biological algorithms for flexible decision-making.