Rapid updating of spatial working memory across saccades

Abstract

Each time we make an eye movement, positions of objects on the retina change. In order to keep track of relevant objects their positions have to be updated. The situation becomes even more complex if the object is no longer present in the world and has to be held in memory. In the present study, we used saccadic curvature to investigate the time-course of updating a memorized location across saccades. Previous studies have shown that a memorized location competes with a saccade target for selection on the oculomotor map, which leads to saccades curving away from it. In our study participants performed a sequence of two saccades while keeping a location in memory. The trajectory of the second saccade was used to measure when the memorized location was updated after the first saccade. The results showed that the memorized location was rapidly updated with the eyes curving away from its spatial coordinates within 130 ms after the first eye movement. The time-course of updating was comparable to the updating of an exogenously attended location, and depended on how well the location was memorized.

Figure 1

Experimental paradigm of Experiment 1. (a) The participants fixated the fixation dot and remembered the location of the memory cue (a white square). After a retention interval two saccade targets appeared, and participants had to make a sequence of two saccades. At the end of the trial participants had to judge whether a test stimulus was presented at the same or a slightly different location as the remembered cue. (b) The white squares indicate possible locations of the memory cue in a trial with a left/downward saccade sequence. The white error bar indicates possible locations of the fixation dot relative to the first saccade target.

Figure 2

Experimental predictions. Participants performed a sequence of two saccades while holding a location in memory. The memorized location could be presented either counterclockwise (orange) or clockwise (blue) from the second saccade target (a). Trajectories of the second saccade were used to measure the effect of the first saccade on the representation of the remembered location. The curved lines illustrate the predicted curvature away from retinotopic (b) and spatiotopic locations (c). If the memorized location is rapidly updated, we expected the second saccade to curve away from its spatiotopic location at short intersaccadic intervals. However, if the formation of a spatiotopic representation is a slow and effortful process, saccades should only curve away from this location at longer intersaccadic intervals.

Figure 3

Results of Experiment 1. (a) Curvature difference for all four latency bins (counterclockwise minus clockwise memory location). The diamond shapes represent the within-subjects SEM calculated over curvature (vertical dimension) and the between-subjects SEM calculated over the mean intersaccadic latencies for that latency bin (horizontal). The grey line at the bottom represents the interval for which data was smoothed. (b) Average of the smoothed curvature of all participants (counterclockwise minus clockwise memory location). Shaded error bars indicate the within-subjects SEM calculated over curvature. The grey line indicates a cluster that did not survive the cluster-based permutation testing. (c). Average saccade trajectories while memorizing locations at counterclockwise (orange) and clockwise (blue, see inset) locations for the three latency bins. Error bars indicate the within-subjects SEM.

Figure 4

Saccade curvature as a function of memory performance in Experiment 1. (a) The data from the best (blue) and worst performing participants (orange) on the memory test. The diamond shapes represent the within-subjects SEM calculated over curvature (vertical dimension) and the between-subjects SEM calculated over the mean intersaccadic latencies for that latency bin (horizontal). The best performing participants showed spatiotopic curvature already at the shortest intersaccadic intervals. For the worst performing individuals it takes much longer before saccades start to curve away from the spatiotopic location. The grey line at the bottom represents the interval for which data was smoothed. (b). Difference in curvature between the smoothed data in the counterclockwise and clockwise conditions for the best performing participants (blue) and worst performing participants (orange) on the memory test. Shaded error bars represent the within-subjects SEM calculated over curvature. The blue horizontal line indicates significant clusters for the best performing individuals.

Figure 5

Experimental paradigm of Experiment 2. Participants fixated on the fixation dot and remembered the location of the cue. After a retention interval a saccade target appeared, and participants had to make a saccade. After a random interval between 0 and 300 ms after landing of the first saccade, another saccade target appeared and participants had to make a second saccade. At the end of the trial participants had to judge whether a test stimulus was presented at the same or at a slightly different location as the remembered cue.

Figure 6

Results of Experiment 2. (a) Curvature difference for all four latency bins (counterclockwise minus clockwise memory location). The diamond shapes represent the within-subjects SEM calculated over curvature (vertical dimension) and the between-subjects SEM calculated over the mean intersaccadic latencies for that latency bin (horizontal). The grey line at the bottom represents the interval for which data was smoothed. (b) Average of the smoothed curvature of all participants (counterclockwise minus clockwise memory location). Shaded error bars indicate the within-subjects SEM calculated over curvature. The black horizontal lines indicate significant clusters. (c). Average saccade trajectories while memorizing locations at counterclockwise (orange) and clockwise (blue) locations for four latency bins.

Figure 7

Saccade curvature as a function of memory performance in Experiment 2. (a) Calculated difference in curvature between counterclockwise and clockwise conditions for all four latency bins. Data from the best performing participants (blue) and worst performing participants (orange) on the memory test. The diamond shapes represent the within-subjects SEM calculated over curvature (vertical dimension) and the between-subjects SEM calculated over the mean intersaccadic latencies for that latency bin (horizontal). The grey line at the bottom represents the interval for which data was smoothed. (b) Difference in curvature between the smoothed data in the counterclockwise and clockwise conditions for the best performing participants (blue) and worst performing participants (orange) on the memory test. Shaded error bars represent the within-subjects SEM calculated over curvature. The blue horizontal line indicates significant clusters for the best performing individuals.