Contributed Talk Sessions | Poster Sessions | All Posters | Search Papers

Poster Session A: Tuesday, August 12, 1:30 – 4:30 pm, de Brug & E‑Hall

Head-Direction Cells in Postsubiculum Show Systematic Parallax Errors During Visual Anchoring

Sven Krausse¹, Friedrich Sommer², Emre Neftci³, Alpha Renner⁴; ¹Forschungszentrum Juelich GmbH, ²University of California Berkeley, ³Foschungszentrum Juelich and RWTH Aachen, ⁴Forschungszentrum Jülich

Presenter: Sven Krausse

Spatial navigation relies on the head-direction (HD) system, which integrates angular head velocity (AHV) to track orientation. Since integration accumulates drift over time, visual landmarks provide corrective cues. However, whether the HD system explicitly accounts for the apparent shift in proximal objects' position when viewed from different angles, remains unclear. These shifts are caused by the parallax effect, where closer objects move more strongly on the retina than distant ones. Here, we analyzed postsubicular HD cell activity in mice navigating with a single visual cue. We discovered a systematic parallax bias in the decoded HD, indicating that the HD system misinterprets the cue's position dependent on the viewing angle. The observed error was smaller than predicted by a pure vision model, which we show can be explained by the combination of AHV integration with simple visual anchoring. Notably, each animal exhibited a unique anchoring angle — the direction at which the cue was associated with head direction — suggesting that the HD system maintains a relatively stable and possibly learned mapping between the cue angle from visual input (bearing) and head direction. These results provide evidence that the HD system, at least in simplified environments, does not perform explicit parallax correction but may instead attenuate errors passively through AHV integration and simple anchoring to multiple cues. This highlights a fundamental trade-off in neural coding between computational efficiency and positional accuracy, with implications for biological and artificial navigation systems.

Topic Area: Visual Processing & Computational Vision

Extended Abstract: Full Text PDF