Visual Context Modulates Systematic Biases in Human Spatial Working Memory

Poster Presentation 56.406: Tuesday, May 19, 2026, 2:45 – 6:45 pm, Pavilion
Session: Visual Working Memory: Spatial

Danny Lim1, Thomas Sprague; 1UC Santa Barbara

Working memory (WM) maintains information to guide behavior, recruiting circuits across visual, parietal, and prefrontal cortex (Christophel et al., 2017; Curtis & Sprague, 2021; Kiyonaga & Serences, 2025). Yet spatial WM shows systematic recall biases (Jiang & Leung, 2021). Parallel research in humans and nonhuman primates shows that medial temporal lobe structures, including entorhinal and hippocampal cortex, use allocentric information to modulate gaze position (Killian et al., 2012; 2015; Meister & Buffalo, 2018; Nau et al., 2018; Julian et al., 2018; Wilming et al., 2018), suggesting that environmental structure influences spatial coding beyond retinotopic maps. We tested whether such geometry-anchored representations also shape human spatial WM. Participants performed a spatial WM task in which a stable screen aperture (circle, square, wide rectangle, tall rectangle) remained visible throughout each trial. They briefly viewed a target presented within the aperture (randomly chosen on each trial), and then reported its location either with a memory-guided saccade or a mouse click to the remembered position. If allocentric scaffolding contributes to WM, geometric context should bias recall. Our results reveal interactions between aperture geometry and target position across response modalities. Circular and square apertures produced relatively small overall biases, whereas wide and tall rectangular apertures yielded pronounced anisotropies aligned with their long axis. Within the square aperture, however, responses to diagonal targets were systematically displaced toward the corners, indicating that even nominally symmetric geometries impose structured distortions. These effects demonstrate that spatial WM is not encoded solely in retinotopic coordinates but is shaped by allocentric structure in the surrounding visual environment. Together, these findings show that geometric context exerts a strong, systematic influence on spatial WM representations. They provide behavioral evidence that the visual system integrates allocentric cues to stabilize and organize memory, offering a principled framework for probing the neural mechanisms underlying reference-frame transformations.

Acknowledgements: Funding: UCSB Academic Senate Research Grant