Anthony Herzog¹ (), Kyra Warmuth¹, Sami R. Yousif¹; ¹Ohio State University

Visual memory is biased. Even when remembering something as simple as the location of a dot within a circle, people make small but systematic errors — errors that reveal the underlying processes that structure spatial memory. Two prominent accounts attempt to explain these biases. On one account, spatial memory is described using a two-system framework in which a coarse, categorical representation of space interacts with a fine-grained, metric representation, with biases arising primarily from the categorical system (Huttenlocher, Hedges, & Duncan, 1991). More recently, an alternative account has proposed that spatial memory is biased towards regions of high perceptual acuity (Langlois et al., 2021). Here, we present evidence that challenges key predictions of both accounts. Across seven experiments, participants completed variants of a classic visuospatial memory task in which they viewed a single dot inside of a shape and later recalled its location. The task was performed in square environments with various salient structures (e.g., grid lines, landmarks). Three patterns emerged. First, categorical spatial biases proved remarkably stable in some cases: The presence of gridlines, for instance, did not eliminate such biases, with participants continuing to recall dots closer to the centers of their quadrants. Second, and in contrast, local structural changes produced global distortions. The introduction of a small landmark in one quadrant of the space dramatically reshaped error patterns across the entire environment, even disrupting categorical biases in the opposite quadrant. Third, regions of high perceptual acuity produced repulsive rather than attractive distortions (contradicting recent accounts). Together, these findings reveal several highly general regularities in spatial memory that are not predicted by existing theories, suggesting that environmental structure interacts with spatial representations in complex and unintuitive ways.

Acknowledgements: Special thanks to the Spatial Cognition Lab at the Ohio State University for all the support they've given.