Transformation and recombination of neural information in a brain network

Mammalian brain function relies on integrated interactions among interconnected neural structures. The principles by which regional activity is transformed into projection-specific signals and then recombined at targets elsewhere in the brain are fundamental to brain processing, but remain poorly understood. Here we study these phenomena using a genetically encoded probe that provides neurophysiological readouts from virally labeled projections on a brain-wide scale via functional magnetic resonance imaging (fMRI). By analyzing outputs from thalamic and cortical somatosensory processing regions in rats, we find that projection-specific neural population activity undergoes shifts in tuning and temporal characteristics as it emanates from source regions. Patterns of neural information flow to targeted brain structures reconfigure under different conditions of stimulation and rest, contrasting with intrinsic fMRI functional connectivity profiles, which remain constant. Excitatory and inhibitory projections are coactivated during stimulation, but their relative response amplitudes change dynamically between stimulus conditions and across repeated stimuli, suggesting mechanistic roles for network-wide shifts in excitation/inhibition balance. Our results thus reveal how information flow throughout a neural system reshapes to promote stimulus selectivity and provide underpinnings of large-scale brain phenomena.