Biphasic neural response properties, where the optimal stimulus for driving a neural response changes from one stimulus pattern to the opposite stimulus pattern over short periods of time, have been described in several visual areas, including lateral geniculate nucleus (LGN), primary visual cortex (V1), and middle temporal area (MT). We describe a hierarchical model of predictive coding and simulations that capture these temporal variations in neuronal response properties. We focus on the LGN-V1 circuit and find that after training on natural images the model exhibits the brain's LGN-V1 connectivity structure, in which the structure of V1 receptive fields is linked to the spatial alignment and properties of center-surround cells in the LGN. In addition, the spatio-temporal response profile of LGN model neurons is biphasic in structure, resembling the biphasic response structure of neurons in cat LGN. Moreover, the model displays a specific pattern of influence of feedback, where LGN receptive fields that are aligned over a simple cell receptive field zone of the same polarity decrease their responses while neurons of opposite polarity increase their responses with feedback. This phase-reversed pattern of influence was recently observed in neurophysiology. These results corroborate the idea that predictive feedback is a general coding strategy in the brain.

For many neurons in the early visual brain the optimal stimulation for driving a response changes from one stimulus pattern to the opposite stimulus pattern over short periods of time. For example, many neurons in the lateral geniculate nucleus (LGN) respond to a bright stimulus initially but prefer a dark stimulus only 20 milliseconds later in time, and similar changes in response preference have been found for neurons in other areas. What would be the computational reason for these biphasic response dynamics? We describe a hierarchical model of predictive coding that explains these response properties. In the model, higher-level neurons attempt to predict their lower-level input, while lower-level neurons signal the difference between actual input and the higher-level predictions. In our simulations we focus on the LGN and area V1 and find that after training on natural images the layout of model connections resembles the brain's LGN-V1 connectivity structure. In addition, the responses of model LGN neurons are biphasic in time, resembling biphasic responses as found in neurophysiology. Moreover, the model displays a specific pattern of influence of feedback from higher-level areas that was recently observed in neurophysiology. These results corroborate the idea that predictive feedback is a general coding strategy in the brain.