Are eye movements planned as sequences?
When you try to find an object, your gaze location will rapidly shift from one location to another. Such rapids shifts in gaze are accomplished by saccades, ballistic movements of the eye from one point of focus to another. In between these saccades, the gaze position remains (relatively) steady for a short period of time (a fixation) to allow your visual system to analyze the light falling on the retina. In researching what determines where the gaze position will be shifted to next, most of the literature works with the assumption that this upcoming location is determined during the current fixation. However, we know from other motor systems that often multiple movements are coordinated ahead of time. For instance, the coordination of my fingers as I'm typing this post. Previously it has also been shown that a subsequent saccade can be prepared in parallel to the execution of a current saccade (typically referred parallel programming). However, only limited research has focused on whether this parallel programming of saccades occurs in active search and to what extent (for some extra background, scroll to the bottom).
We conducted two experiments to evaluate whether saccades are planned ahead and to what extent. The general rationale behind both experimental procedures was that if any form of a plan exists we should be able to find some trace of this plan: If a path is planned deviating from this path should be more costly (i.e. slower) than continuing on this path. In the first experiment participants were presented with displays containing three disks, two white and one black (as depicted in the animation to the left). Participants had to locate the white disk that also contained a small horizontal line in the middle. Considering the difference in brightness, few eye movements were made to the black disk, as it could be rejected as the target without making an eye movement. As such, knowing the destination of the first saccade allowed us to predict where the second saccade would go. Distinguishing whether the horizontal line was present, however, was only possible once the eye landed on a white disk. Whether a second eye movement was required, was only known after the completion of the first. Now to evaluate whether the destination of second saccade was planned, on a proportion of the trials, the location of the other white disk and the black disk would be switched during the first eye movement. To the participant this only becomes apparent after the eye lands (during a saccade visual input is not analyzed). If the destination of the second saccade was already planned, it was expected to continue erroneously towards the old position of the potential target, rather than the new. On a large proportion of trials this is exactly what we find, demonstrating that even in a search relying on explorative eye movements, more than one eye movement destination is planned ahead.
We conducted two experiments to evaluate whether saccades are planned ahead and to what extent. The general rationale behind both experimental procedures was that if any form of a plan exists we should be able to find some trace of this plan: If a path is planned deviating from this path should be more costly (i.e. slower) than continuing on this path. In the first experiment participants were presented with displays containing three disks, two white and one black (as depicted in the animation to the left). Participants had to locate the white disk that also contained a small horizontal line in the middle. Considering the difference in brightness, few eye movements were made to the black disk, as it could be rejected as the target without making an eye movement. As such, knowing the destination of the first saccade allowed us to predict where the second saccade would go. Distinguishing whether the horizontal line was present, however, was only possible once the eye landed on a white disk. Whether a second eye movement was required, was only known after the completion of the first. Now to evaluate whether the destination of second saccade was planned, on a proportion of the trials, the location of the other white disk and the black disk would be switched during the first eye movement. To the participant this only becomes apparent after the eye lands (during a saccade visual input is not analyzed). If the destination of the second saccade was already planned, it was expected to continue erroneously towards the old position of the potential target, rather than the new. On a large proportion of trials this is exactly what we find, demonstrating that even in a search relying on explorative eye movements, more than one eye movement destination is planned ahead.
In Experiment 2 we extend the first experiment, by evaluating whether also a third destination is automatically planned ahead. We again asked participants to search for a white disk, now distinguished from other white disks by a small vertical line at its center. However, in the displays used in this experiment more white and black disks were included (3 white, 3 black). Placing the disks in columns from left to right allowed us to predict the order in which elements would be fixated. Again we rely on the notion that if a path is in preplanned we should be able to see some cost of deviating from this plan. To test this, in a subsection of the trials, the white and dark disks disappeared during the first saccade and instead participants had to follow a jumping red disk. The red disk either followed the predicted path of the participant, or it would deviate on the third eye movement by moving towards the previous location of the distractor in the third column. We find that participants were quicker to follow the red disk when it stayed on the predicted path on the third saccade, rather than deviating from it.
Thus, it appears that some form of planning of upcoming saccade destinations takes place and this holds even for a third destination. The fact that three saccade destinations can be planned in a single fixation is quite stunning considering the little time there is to plan this path: In the current experiment participants executed their initial saccade after only some 155ms (averaged over all participants). Despite the fact that the disks disappeared from the screen upon the execution of the first saccade, participants were still be faster in executing the planned path, and would even mistakenly make an eye movement to the old location of the white disk on the third eye movement occasionally.
Thus, it appears that some form of planning of upcoming saccade destinations takes place and this holds even for a third destination. The fact that three saccade destinations can be planned in a single fixation is quite stunning considering the little time there is to plan this path: In the current experiment participants executed their initial saccade after only some 155ms (averaged over all participants). Despite the fact that the disks disappeared from the screen upon the execution of the first saccade, participants were still be faster in executing the planned path, and would even mistakenly make an eye movement to the old location of the white disk on the third eye movement occasionally.
|
de Vries, J.P., Hooge, I.T.C., & Verstraten, F.A.J. (2014). Saccades toward the target are planned as sequences rather than as single steps. Psychological Science, 25(1), 215– 23.
|
Extra Background: Some previous studies had already demonstrated that a subsequent eye movement can be programmed in parallel to the first: When the first saccade erroneously moves the eye to the wrong destination a second saccade will quickly follow to correct the first. However, the focus of our study was not on the correction of a mistake, but on whether planning occurs in sequences of explorative eye movements. When you need to find your bright red car in a parking garage filled with grey and blue cars you can spot yours even without making an eye movement. If your first eye movement accidentally moves towards any other car, the system may not be able to prevent this movement, but information is present that this movement is going to go in the wrong direction. However, typically you are confronted with the opposite situation: you have to find your grey car in a garage filled with many other cars that have similar colors. Only after you have shifted your gaze to a specific car your visual system can determine whether this is your car. In such cases an eye movement to a grey car that looks like yours is not in error, but serves an exploratory purpose. Thus, in Experiment 1 we evaluated whether previous findings on saccade planning also hold when saccades are not made in error, but to evaluate potential targets. |
