Detecting transient changes in dynamic displays: The more you look, the less you see

Abstract

A series of experiments was conducted examining the detectability of transient changes in a dynamic visual search paradigm. Participants viewed displays with numerous moving objects and were asked to detect when an item was added to the display (onset) or when an item changed color. Consistent with the attention capture literature, onset changes were detected better than color changes. This was true even when participants were highly motivated to detect color changes over onset changes. These results are consistent with contingent capture theory which holds that the attention system can be set to respond to either static or dynamic discontinuities, but cannot be set to selectively respond to different change types within each of these categories. Individual scan strategies could account for a large proportion of variance in detection performance in all experiments (up to 50%). Participants who made few eye movements performed best while participants who actively scanned the display performed worst. A surprisingly large number of participants engaged in this later maladaptive scan strategy. Saccadic suppression was ruled out as an explanation for poor performance. Additionally, improved performance for participants who did not move their eyes was not due to these participants fixating an optimal location in the display. We argue that the act of making eye movements hinders detection performance due to a difficulty in forming a stable representation of the display. When poor performers were instructed to reduce the number of eye movements made during search their performance matched that of the best performers. Conversely, good performers who were instructed to actively scan showed a large decrease in accuracy. These results indicate that the optimal strategy for detecting transient display changes is covert rather than overt search. Although intuition tells us that the harder we look for something the more likely we will see it, this may not be true for transient display changes.