Extended Data Fig. 8: Left-right-alternating internal direction during goal-directed navigation.
From: Left–right-alternating theta sweeps in entorhinal–hippocampal maps of space
a, Internal direction signals on an m-shaped spatial alternation task10. Panels show inferred internal direction (LMT based on MEC-parasubiculum population activity) on two outbound trials (left: outbound left; right: outbound right). Alternate theta cycles are plotted with different colours. The running trajectory from the trial segment is plotted in black, while the full trajectory across all trials is plotted in light grey. b, Polar histograms of internal direction inferred by LMT (top row) or decoded based on head-direction tuning curves (bottom row) during open field foraging (left column), on the central stem of the m-shaped maze preceding a left turn (middle column) or on the central stem preceding a right turn (right column). Same recording as in a. Note that the angular distributions look similar across conditions, indicating that internal direction is not biased towards the to-be-chosen path. c, Internal direction is not biased towards the current goal. Left: schematic of m-maze indicating maze segments and reward wells. Outbound running trajectories were linearized by defining centre well and side well as beginning and end, respectively. Reward direction is defined by the vector from the rat’s head to the to-be-chosen reward well. Right: internal direction averaged across all outbound right laps (n = 13 laps; same recording as in a). Solid lines show mean head-centred internal direction (inferred by LMT) following a left (red) or right (blue) internal direction in the preceding theta cycle. Shading shows 95% bootstrapped confidence intervals of the mean. Brown line shows the head-centred reward direction. Transitions between maze segments are indicated with vertical lines. Note that internal direction on alternating cycles is stable across the entire run despite substantial changes in reward direction. d, Polar histograms of head-centred internal direction in successive maze segments (centre arm, top arm, left or right arm) on outbound trials (left) and inbound trials (right). Note bimodal distribution of internal direction during all task phases. Credit: rat, scidraw.io/Gil Costa.
