Eyes wide open: enhanced pupil dilation when selectively studying important information

Abstract

Remembering important information is imperative for efficient memory performance, but it is unclear how we encode important information. The current experiment evaluated two non-exclusive hypotheses for how learners selectively encode important information at the expense of less important information (differential resource allocation and information reduction). To evaluate these hypotheses, we measured changes in learners' pupil diameter and fixation durations while participants performed a selectivity task that involved studying lists consisting of words associated with different point values. Participants were instructed to maximize their score on a free recall task that they completed after studying each list. Participants' pupils dilated more when studying high-valued than low-valued words, and these changes were associated with better memory for high-valued words. However, participants fixated equally on words regardless of their value, which is inconsistent with the information reduction hypothesis. Participants also increased their memory selectivity across lists, but changes in pupil diameter and differences in fixations could not account for this increased selectivity. The results suggest that learners allocate attention differently to items as a function of their value, and that multiple processes and operations contribute to value-directed remembering.

Figure 1

Mean proportion recall and mean selectivity index (SI) across lists. Error bars represent within-subject standard error of the mean (Loftus and Masson 1994).

Figure 2

Mean task-evoked pupillary response (TEPR) in millimeters (left bars) and mean proportion recall (right bars) for low value (1 to 4 point), medium value (5 to 8 point), and high value (8 to 12 point) words collapsed across lists. Error bars represent within-subject standard error of the mean (Loftus and Masson 1994).

Figure 3

Mean task-evoked pupillary response (TEPR) in millimeters (left bars) for low value (1 to 4 point), medium value (5 to 8 point), and high value (8 to 12 point) words in early lists, where participants displayed low selectivity (M = .26, SE = .04) and later lists where participants displayed higher selectivity (M = .37, SE = .04). Error bars represent within-subject standard error of the mean (Loftus and Masson 1994).

Figure 4

Mean task-evoked pupillary response (TEPR) in millimeters for low value (1 to 4 point), medium value (5 to 8 point), and high value (8 to 12 point) words as a function of selectivity quartile. The first quartile consists of individuals with the lowest selectivity index and the fourth quartile consists of individuals with the highest selectivity index. Error bars represent between-subject (quartile) standard error of the mean.