Abstract
Many visual tasks require that we monitor scenes that change over time, looking for critical events that can occur (or not) at any time. We use the term visual surveillance to capture these task characteristics. Lifeguarding and driving, for example, are real-world tasks that require visual surveillance. Here we explored the effect of variable target-onset time in visual surveillance using a flickering change-detection task. Observers saw fields of vertical and horizontal bars. The bars were presented for 300 ms, followed by a 200-ms blank screen, followed by bars for 300 ms and another 200-ms blank screen. This sequence cycled continuously for the duration of the trial. The task was to find a single bar that alternated between vertical and horizontal as quickly as possible. There were either 16 or 32 bars in the display (set size), and the change was delayed relative to the onset of the cycling display by 0, 4 or 12 seconds. Change detection times were faster for setsize-16 (5409 ms) than for the setsize-32 (8286 ms). They were also faster in the 4-s (6164 ms) and 12-s (6129 ms) delay conditions than in the 0-s delay condition. There was no reliable difference between the 4- and 12-ms delay conditions, however. The longer delays might have yielded slower performance as observers fatigued with searching the unchanging displays. For at least up to 12 seconds, however, there is no evidence of this. Additional experiments will seek to resolve these ideas.
Meeting abstract presented at VSS 2012