Jointly looking to the past and the future in visual working memory
archive: archived pipeline: cataloged verified
Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)
Summary
This study investigates how visual working memory integrates past sensory information with future behavioral demands. While working memory inherently bridges past inputs and future actions, conventional laboratory tasks often conflate these components by encoding and testing stimuli at the same spatial location. The authors address the underexplored question of whether the brain retains and utilizes both the past (encoding) and future (testing) locations of memory contents when they differ, as is common in dynamic real-world scenarios. To dissociate these attributes, the researchers developed a visual working memory task where encoding and testing locations were orthogonalized. Twenty-five healthy human volunteers memorized two colored gratings presented either vertically or horizontally. After a delay, a retro-cue prompted participants to select the relevant memory item. Crucially, the test stimuli appeared on the axis orthogonal to the encoding axis, with the specific relevant test location determined by a predictable "future rule" stable within each session. This design ensured that participants had to rely on both the encoded location (past) and the rule-defined test location (future) to perform the task. The authors tracked spatial biases in gaze behavior, specifically analyzing microsaccades, to independently monitor the utilization of past and future memory attributes. The results demonstrated that participants jointly considered both past and future locations during mnemonic selection. Gaze shifts were significantly biased toward both the encoded location and the to-be-tested location, with these biases emerging at overlapping time windows starting approximately 200 ms after the cue. Critically, single-trial analysis of individual saccades revealed a dependency between these biases: saccades biased toward the future location were exclusively those that were also biased toward the past location. This interaction indicates that past and future attributes are not processed serially or independently across trials, but are truly jointly activated at the single-saccade level. These findings reveal a fundamental property of working memory: the simultaneous retention and utilization of multiple spatial attributes associated with a single memory item. The study challenges models suggesting that memory contents are either remapped solely to future-relevant locations or that past and future attributes are accessed sequentially. Instead, the data support a model where working memory representations are rich and flexible, retaining copies of both past and future locations to serve evolving behavioral demands. By leveraging the discrete nature of microsaccades, the authors provide robust single-trial evidence for this joint consideration, advancing the understanding of how vision, eye movements, and memory interact in dynamic contexts.
Provenance
The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | Crossref | — | — | 1 | 2026-06-19 |
| archive | success | canonical_url | — | — | 1 | 2026-06-26 |
| extract | success | cached | — | — | 2 | 2026-06-26 |
| clean | success | clean | — | — | 1 | 2026-06-20 |
| chunk | success | chunk | — | — | 1 | 2026-06-20 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-20 |
| promote | success | — | — | — | 1 | 2026-06-19 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 1 | 2026-06-26 |
| tag | success | vector_similarity | — | — | 6 | 2026-06-20 |
| verify | success | — | — | — | 1 | 2026-06-26 |
Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.
Topics
Ranked by relevance to this paper. Hover a topic for its definition.