Whenever the eyes move, they don’t always land where we intended them to land. The visual system can ‘clean up’ after itself if it notices an imperfect eye-movement by performing a small additional eye-movement. This task becomes more complicated if the visual scene becomes more cluttered: how can the visual system recover what you were targeting in the first place, rather than just correcting to whatever happens to be around? The visual system detects motor imperfections based on two systems: motor corrections (‘the state of my muscle tension was incorrect’) and visual corrections (‘I am definitely not looking at the thing I was interested in’). Therefore, the visual system needs to know what it is correcting to prior to the eye-movement, and to do so, the visual system stores a representation of the object the eye-movement is performed to. However, stored visual information mandatorily takes up capacity in the visual working memory system, and we set out to investigate how many visual working memory resources an eye-movement uses.
In the current study we used a working memory/eye-movement dual task. Participants remembered one, two or three shape morphs, that were either stretched vertically or horizontally. After the shape morphs were shown, an array of colored circles was shown. Participants were instructed to execute an eye-movement to one of the items. In two-thirds of the trials, the array rotated so that the participant landed in between their original target and a distractor, requiring a corrective eye-movement. After the eye-movement task participant reported the exact width/height of the shape morphs they saw at the start of a trial. Comparing trials in which participants did not perform any eye-movements to trials where they did, revealed that that the cost of making an eye-movement is actually equally large to remembering an additional item. For example, the performance of a participant was equal if they were remembering two items without an eye-movement, or three items with an eye-movement. Furthermore, this cost did not increase when a participant was required to make a corrective eye-movement (retrieving the stored representation).
The current findings indicate that the visual system can flexibly allocate the resources of visual working memory for either saccade tasks or actively committing items to working memory. We conclude that making an eye-movement compulsorily uses these visual working memory resources, similar to volitionally committing an extra item to memory.
Schut, M.J., Van der Stoep, N., Postma, A., & Van der Stigchel, S. (in press). The cost of making an eye movement: A direct link between visual working memory and saccade execution. Journal of Vision