Martin Constant¹ (martin.constant@unige.ch), Dirk Kerzel¹; ¹University of Geneva

We follow up our recent work that demonstrated the massive impact of salience on visual working memory performance (Constant & Liesefeld, 2021; https://doi.org/gjk9jh) and that this robust effect explains most of the performance differences, even when contradictory top-down information or cues are employed (Constant & Liesefeld, 2023; https://doi.org/gr6xzr; Constant & Kerzel, 2024; https://doi.org/mvnk). In the present series of experiments, we re-employed this design, asking participants to encode three tilted bars (12°, 28° and 45°) presented among 33 vertical filler-bars for 350 ms. Following a 1000 ms retention interval, the response display appeared. Here, the key difference with the previous research is that participants had to recall the color of all three targets in a seemingly random order (i.e., whole-report procedure) instead of reporting the color of a single target. We wanted to examine whether salience could prevent or reduce the performance cost affecting targets that are recalled later. We hypothesized that this could occur through the storage of the most salient target in a special active state in visual working memory. Contrary to our expectations, the order effects’ performance cost was comparable across all levels of salience, and therefore more salient targets were not shielded. We replicated this finding in a second experiment, and show that this is not prevented when targets of a given display share the same tilt. Finally, in a third experiment, we used a spatial pre-cue to indicate which of the three targets would be recalled last in the sequence. Preliminary results indicate that this advance information on the recall order does protect the cued target to some extent. However, this benefit appears to come with a performance cost for earlier targets, suggesting a performance trade-off.