Closed-loop control and recalibration of place cells by optic flow

Abstract

Understanding the interplay between sensory input, endogenous neural dynamics, and behavioral output is key toward understanding the principles of neural computation. Hippocampal place cells are an ideal system to investigate this closed-loop interaction, as they are influenced by both self-motion (idiothetic) signals and by external sensory landmarks as an animal navigates its environment1-9. To continuously update a position signal on an internal "cognitive map", the hippocampal system integrates self-motion signals over time10,11. In the absence of stable, external landmarks, however, these spatial correlates of neuronal activity can quickly accumulate error and cause the internal representation of position or direction to drift relative to the external environment1,5. We have previously demonstrated that, in addition to their known roles in preventing and/or correcting path-integration error, external landmarks can be used as a putative teaching signal to recalibrate the gain of the path integration system6. However, it remains unclear whether idiothetic cues, such as optic flow, exert sufficient influence on the cognitive map to enable recalibration of path integration, or if instead an unambiguous allocentric frame of reference, anchored by polarizing landmark information, is essential for path integration recalibration. Here, we use principles of control theory12,13 to demonstrate systematic control of place fields by pure optic flow information in freely moving animals by using a neurally closed-loop virtual reality system that adjusts optic flow speed as a function of real-time decoding of the hippocampal spatial map. Using this "cognitive clamp", we show that we can not only bring the updating of the map under control of the optic flow cues but we can also elicit recalibration of path integration. This finding demonstrates that the brain continuously rebalances the influence of conflicting idiothetic cues to fine-tune the neural dynamics of path integration, and that this recalibration process does not require a top-down, unambiguous position signal from landmarks.