Behavioural Assessment of Coding Strategies in Spatial Working Memory

Abstract

Introduction

Spatial working memory (sWM) allows temporary retention of spatial information. To support this function, the brain uses different coding strategies. Sensory coding preserves the target as a perceptual representation, whereas motor coding encodes it as a planned movement [1]. Likewise, allocentric (object‑centered) frames reference external landmarks, while egocentric (self‑centered) frames use the observer’s body and gaze [2].

This study investigates how distinct spatial coding strategies support the encoding and maintenance of locations in working memory. To address this, we developed a behavioural paradigm that systematically promotes each strategy, combined with a self-report questionnaire to assess participants’ encoding approaches. The aim was to determine whether these strategies yield measurable differences in recall precision and variability.

Methods

Thirty-eight participants (31 women, 7 men; mean age 20 ± 1 year) completed five tasks: simple, sensory, motor, allocentric and egocentric. In the simple task, participants viewed a stimulus at various screen locations and, after a 10‑second delay, moved their eyes to its remembered position. In the sensory condition, they first shifted gaze to a new point before responding, which disrupted motor planning and relied on sensory memory. In the motor condition, participants first looked at the remembered location, enabling direct motor execution. Allocentric and egocentric tasks required recalling the target based on either absolute screen coordinates or relative to a shifted fixation point, respectively. Eye movements were recorded via EyeLink 1000. 

Results

Participants' spatial recall precision differed across task conditions, as measured by the Euclidean distance between target and response locations. Smaller distances indicated more accurate spatial recall. The motor condition yielded the smallest median error and lowest variability, indicating the most precise responses. The simple condition showed slightly higher errors but remained relatively consistent. The sensory condition produced moderately increased errors and variability. In contrast, the allocentric condition resulted in larger errors and greater response spread, while the egocentric condition showed the highest overall error and variability across participants.

Discussion

These results suggest that different task conditions promoted distinct spatial coding strategies, reflected in systematic differences in recall precision. Highest accuracy in the motor condition indicates that planned movements support more effective encoding. In contrast, allocentric and egocentric conditions, which required spatial transformations, led to greater errors and variability. This supports the interpretation that increased cognitive demands reduce recall precision. While behavioural data alone limit insight into underlying mechanisms, future fMRI research will help clarify the neural basis of these strategies.

References

[1] N. Purg, M. Starc, A. S. Ozimič, A. Kraljič, A. Matkovič, and G. Repovš, “Neural evidence for different types of position coding strategies in spatial working memory,” Frontiers in Human Neuroscience, vol. 16, 2022. [Online]. doi: 10.3389/fnhum.2022.821545. [Accessed: May 9, 2025].

[2] A. D. Ekstrom and E. A. Isham, “Human spatial navigation: representations across dimensions and scales,” Current Opinion in Behavioral Sciences, vol. 17, pp. 84–89, 2017. [Online]. doi: 10.1016/j.cobeha.2017.06.005. [Accessed: May 9, 2025].