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

Sichu Liang, Hongyu Zhu, Wenwen Wang, Deyu ZhouarXiv 2026 · DOI

BibTeX
@misc{liang2026can,
  author = {Sichu Liang and Hongyu Zhu and Wenwen Wang and Deyu Zhou},
  title = {{Can Vision Replace Text in Working Memory? Evidence from Spatial n-Back in Vision-Language Models}},
  year = {2026},
  doi = {10.48550/arxiv.2602.04355},
}

Loading related content...

Figure