Can Vision Replace Text in Working Memory? Evidence from Spatial n-Back in Vision-Language Models

TL;DR

This study investigates whether vision-language models can perform working memory tasks similarly to text-based models, revealing differences in accuracy and underlying processes when using visual versus textual information.

Contribution

It provides the first systematic comparison of working memory-like behavior in vision-language models using spatial n-back tasks with visual and textual stimuli.

Findings

Models perform better with text than vision in n-back tasks.

Model responses often reflect recency effects rather than true lag-based memory.

Stimulus grid size influences interference and error patterns.

Abstract

Working memory is a central component of intelligent behavior, providing a dynamic workspace for maintaining and updating task-relevant information. Recent work has used n-back tasks to probe working-memory-like behavior in large language models, but it is unclear whether the same probe elicits comparable computations when information is carried in a visual rather than textual code in vision-language models. We evaluate Qwen2.5 and Qwen2.5-VL on a controlled spatial n-back task presented as matched text-rendered or image-rendered grids. Across conditions, models show reliably higher accuracy and d' with text than with vision. To interpret these differences at the process level, we use trial-wise log-probability evidence and find that nominal 2/3-back often fails to reflect the instructed lag and instead aligns with a recency-locked comparison. We further show that grid size alters recent-repeat structure in the stimulus stream, thereby changing interference and error patterns. These results motivate computation-sensitive interpretations of multimodal working memory.