Abstract
To reduce redundant information, the visual system may create statistical representations of sets, discarding information about individual items. Ariely (2001) found that observers could not reliably determine whether a test circle appeared in a set of circles presented immediately beforehand, but could determine whether it corresponded to the mean size of the set. Here we investigate whether the average size of a set of circles presented in a rapid sequence is statistically represented. In Experiment 1, an RSVP display of circles was followed by a test circle, and observers determined whether the test circle was the mean or a member of the set. Observers could determine average size, but could not identify individual circles. Furthermore, the test circle was erroneously identified as the mean with increasing frequency as the difference in size between the target and mean decreased. To determine whether performing the mean identification task depleted central attentional resources needed to perform concurrent processing, in Experiment 2 observers were presented either with a stream of circles (and identified the mean) followed by a stream of shapes, or with a single circle within a stream of shapes (and identified the size of the circle). The target shape followed the circle(s) at one of five lags. Shape identification was better at short lags following the mean task compared to the single circle task. Further, shape identification was lag-dependent following the single circle task (an attentional blink; Raymond, Shapiro, & Arnell, 1992), but unaffected by lag following the mean judgment task. This suggests that identifying a single circle, but not judging the mean of the set, depleted central attentional resources necessary to perform the subsequent shape task. We conclude that rapid averaging over sequentially presented displays is accomplished by an efficient, possibly unavoidable process, that does not require central attentional resources.