Different computations over the same inputs produce selective behavior in algorithmic brain networks

A key challenge in systems neuroscience remains to understand where, when and now particularly how brain networks compute over sensory inputs to achieve behavior. We used XOR, OR and AND functions as behavioral tasks, because each requires a different computation over the same inputs to produce correct outputs. In each task, source-localized magnetoencephalographic activity progresses through four systems-level computations identified within individual participants (N = 10/task): (1) linear discrimination of each visual input, first contra-laterally in occipital cortex then (2) jointly in midline occipital cortex and right fusiform gyrus, followed by (3) nonlinear task-dependent input integration in temporal-parietal cortex and finally (4) behavioral response representation in post-central gyrus. Our results show how network algorithms differently compute over the same inputs to produce different behaviors. One sentence summaryFour stages of task-specific computations over the same visual inputs achieve different behaviors in dynamic brain networks