Fixation duration surpasses pupil size as a measure of memory load in free viewing

Abstract

Oculomotor behavior reveals, not only the acquisition of visual information at fixation, but also the accumulation of information in memory across subsequent fixations. Two candidate measures were considered as indicators of such dynamic visual memory load: fixation duration and pupil size. While recording these measures, we displayed an arrangement of 3, 4 or 5 targets among distractors. Both occurred in various orientations. Participants searched for targets and reported whether in a subsequent display one of them had changed orientation. We determined to what extent fixation duration and pupil size indicate dynamic memory load, as a function of the number of targets fixated during the search. We found that fixation duration reflects the number of targets, both when this number is within and above the limit of working memory capacity. Pupil size reflects the number of targets only when it exceeds the capacity limit. Moreover, the duration of fixations on successive targets but not on distractors increases whereas pupil size does not. The increase in fixation duration with number of targets both within and above working memory capacity suggests that in free viewing fixation duration is sensitive to actual memory load as well as to processing load, whereas pupil size is indicative of processing load only. Two alternative models relating visual attention and working memory are considered relevant to these results. We discuss the results as supportive of a model which involves a temporary buffer in the interaction of attention and working memory.

Keywords: attention; eye movements; fixation duration; memory load; processing load; pupil size; visual search; working memory.

Figure 1

A sample display with 40 items: 5 targets (“T”) and 35 distractors (“L”).

Figure 2

Cumulative target fixation scores calculated for each second of search interval. The means across 19 participants are shown. Standard errors are not displayed because they were so small that they did not exceed the mean markers in size.

Figure 3

Average fixation duration and pupil size. Fixation duration (A) and pupil size (arbitrary units) (B) in 1-s intervals for the entire search interval. Along the time axis the target conditions having significant difference are indicated as found from post-hoc tests. The difference between 3- and 5-target conditions is not shown because it occurred from the 2nd to the 10th second for fixation duration and from the 3rd to the 10th second for pupil size. All plots show the data averaged across 19 participants. The error bars indicate the standard errors of the means.

Figure 4

Pupil size series (arbitrary units). (A) Pupil size series averaged across all target conditions for entire duration of search interval. (B) Pupil size series for the 3 target conditions after omitting the initial pupil response to luminance change. In (B) the bar along the time axis indicates the interval of significant difference between target conditions as found in the point-wise ANOVA. All plots show the pupil size series averaged across 19 participants. The shaded areas indicate standard errors of the means.

Figure 5

Fixation duration and pupil size for targets and distractors. Fixation duration (A) and pupil size (arbitrary units) (B) for fixations on successive targets and their preceding distractors in the order of target fixations. Only for fixation duration, a significant difference between targets and distractors as found from post-hoc tests was seen for all target visiting orders. All plots show the data averaged across 19 participants. The error bars indicate standard errors of the means.