It has been widely supported that visual information that receives foveal information would be automatically encoded and stored by VWM However this proposition has been challenged by the findings of t

The current study challenges the widely supported notion that visual information that receives foveal attention is automatically encoded and stored in visual working memory (VWM). The study aims to investigate whether voluntary saccades or fixations on an item ensure VWM encoding through three experiments.

In Experiment 1 and 2, a modified attribute amnesia (AA) paradigm was used. AA refers to the phenomenon where people fail to report specifically attended attributes of an item, even though they have just identified those attributes. In each trial, all items (one letter and three digits) were initially masked, and participants needed to gaze at the masked locations to temporarily reveal the items. If participants stopped gazing at an item's location, it would be masked again. Participants were asked to search for a target item and report its location in pre-surprise trials. Then, unexpectedly, they were asked to report the identity of the target item in a surprise trial and two control trials. The accuracy of their reports in the surprise trial was significantly different from that in the control trials, indicating the presence of the AA phenomenon in this task.

In Experiment 3, the study directly compared memory traces between saccade and non-saccade conditions by testing memory-driven attentional bias. Participants were instructed to look at a memory item either directly or after saccading towards it. They needed to selectively remember either the shape or color of the item, depending on whether it was a circle or not, and report it at the end of the trial. Relevant features (i.e., color) or key features (i.e., shape) might appear as distractors in a subsequent search task. In both saccade and non-saccade conditions, the appearance of relevant features significantly increased search reaction time (RT), indicating enhanced memory traces. However, the appearance of key features did not affect search RT, suggesting that memory traces of key features were not enhanced with saccades.

These findings demonstrate that voluntary saccades or fixations on an item are not sufficient conditions for automatic VWM encoding. They suggest a separation between the mechanisms of foveal attention and VWM encoding