Abstract
We investigated causes of failures in a visual surveillance task. Observers searched for and reported the orientation of an oblique bar among vertical and horizontal distractor bars. Under standard (static) conditions, this is an easy task. However, adding frames prior to and after the target display changes the task into search for a single state of the bars within an ongoing dynamic display. We distinguish this as a surveillance task because it involves monitoring objects that change over time. Targets were black 45° (left or right) bars and distractors were black 0° or 90° bars. Static displays consisted of a single 150 ms frame in which the target appeared. Dynamic displays consisted of multiple 150 ms frames in which each bar rotated clockwise or counter clockwise, passing through the critical 45°/0°/90° orientations in the middle frame. Left/Right accuracy was 85% for static displays and 51% (near chance) for dynamic displays (Experiment 1). This failure in surveillance may depend on having to report a changing feature (orientation). To test this we altered the displays, which were otherwise as before, so that during the critical 45°/0°/90° frame, bars were 3 of 4 possible colors and the task was to report the color of the target bar (Experiment 2). Performance was much better (static: 94%, dynamic: 80%). Similarly, reporting the orientation of the target bars in the color-change displays (Experiment 3) also yielded improved performance (static: 92%, dynamic: 72%), though accuracy in this Left/Right task was still lower than when people reported color. The color-singleton status may guide attention to the target item, facilitating the extraction of the relevant information from temporally surrounding information more efficiently than when less guidance was available. Attentional guidance may be especially critical in surveillance because both spatial and temporal surrounding information can introduce distraction.
Meeting abstract presented at VSS 2012