Recurrent cortical networks encode natural sensory statistics via sequence filtering

Recurrent neural networks can generate dynamics, but in the sensory cortex, it has been unclear if any dynamic processing is supported by the dense recurrent excitatory-excitatory network. Here, we show a role for recurrent connections in the mouse visual cortex: they support powerful dynamical computations, but by filtering sequences of input instead of generating sequences. Using two-photon optogenetics, we measure neural responses to natural images and play them back, finding that responses are boosted when inputs are played back during the correct movie dynamic context-when the preceding sequence corresponds to natural vision. This sequence selectivity depends on a network mechanism: earlier input patterns produce responses in other local neurons, which interact with later input patterns. We confirm this mechanism by designing sequences of inputs that are boosted or attenuated by the network. These data suggest that recurrent cortical connections perform predictive processing, encoding the statistics of the natural world in input-output transformations.