Trajectory scanning as a predictive coding mechanism for goal-directed navigation, obstacle avoidance and episodic memory.

Planning of future trajectories through the environment may rely on our capacity to encode and recall episodic memories that include spatial representations. Here, we describe potential mechanisms for behaviour based on episodic memory that would use neural trajectory scanning to explore and select trajectories towards remembered goal locations. This model is consistent with rodent neurophysiological data recorded in the medial entorhinal cortex that demonstrated alternating left-right scans of space and data recorded in the hippocampus of place cell firing that showed alternating scans, or scans directed towards a goal location. In a related model, sequential activation of different directions along trajectories would allow encoding and retrieval of complex sequences of vectors that correspond to episodic memories of events at a specific time and place. Trajectory scanning may also detect and avoid potential collisions with representations of visible or remembered objects and barriers in allocentric or egocentric reference frameworks to plan trajectories. Moreover, the same trajectory scanning in egocentric space could generate egocentric neural responses such as egocentric boundary cells and egocentric bearing cells consistent with recent human imaging experiments. Together, these theoretical mechanisms provide a predictive coding framework linking allocentric navigation, episodic memory retrieval, and egocentric obstacle representations. This article is part of the theme issue 'The role of hippocampal predictions in cognition: bridging perception and memory'.