Visual exploration reveals the precision of semantic and spatial memory

Poster Session C - Sunday, April 14, 2024, 5:00 – 7:00 pm EDT, Sheraton Hall ABC

Anikka G. Jordan¹ (anikka@uchicago.edu), Joel L. Voss¹, James E. Kragel¹; ¹University of Chicago

Memories range in precision from highly detailed to vague, but the neural origin of this spectrum is not yet known. Current theories attribute these differences to the anatomical organization of the human hippocampus, with increasing memory resolution in the anterior to posterior direction. However, it remains unclear whether semantic and spatial components of memory share a common anatomical gradient. To address this question, we created a sequence-memory fMRI task that independently manipulates the semantic and spatial attributes of repeating sequences. Participants (N = 9) learned visuospatial sequences of objects and detected mismatching test sequences that differed along spatial and semantic dimensions. Eye-movement tracking was used to differentiate semantic versus spatial memory influences on behavior with high temporal precision. Linear mixed-effects modeling indicated that after learning, participants made predictive eye movements to the next location in the sequence (F(1,8.8) = 8.8, p < 0.001), reflecting spatial memory. Fixation durations increased for unexpected objects but not for objects presented in unexpected locations (F(1,34.6) = 4.8, p = 0.04), reflecting semantic memory. Finally, accuracy on the task varied based on semantic and spatial similarity, with better detection of mismatching sequences that were either spatially or semantically far from targets (F(1,9.6) = 9.6, p = 0.008). These findings indicate that the precision of spatial and semantic memory can be differentiated in eye-movement behavior and recognition responses. We will describe fMRI results that utilize these behaviors to relate semantic and spatial gradients along the hippocampal long axis to memory precision.

Topic Area: LONG-TERM MEMORY: Episodic