Abstract
In studies of change detection, observers’ ability to detect changes to an item in a display declines as set size increases. While the bulk of research has investigated the role of encoding and storage limits in change detection performance, less attention has been given to the role of comparison limits; that is, limitations on the rate or number of comparisons that can be made between test items and memory representations. Our study tested observers’ change detection ability for 3, 6, or 9 coloured circles presented for 1000 ms, followed by a 500ms mask and subsequent 500ms blank screen. At test, 3, 6, or 9 coloured circles again appeared in the same locations as the sample display, and observers were required to report whether a change occurred to one of the circles (50% of trials). On some trials one or two cues were presented at test, indicating which coloured circles may have changed, thus reducing the number of comparisons needed between the test display and memory representations. Our results showed that detection ability for the one cue condition (d’ = 1.8, k = 2.7) was significantly better than for the no cue condition (d’ = 1.6, k = 2.3), t(22) = 2.9, p = .008 (d’); t(22) = 3.07, p = .005 (k). We also found that observers adopted a conservative change detection bias at set size 9 when not provided a cue (c = 0.25), which was reduced when a cue was provided (c = 0.1), t(22) = 2.19, p = .04. We conclude that comparison limits do contribute to the decline in change detection performance, and that when searching for changes at large set sizes, observers favour reporting no change due to increasing comparison uncertainty.
Meeting abstract presented at VSS 2013