关键词: 视觉搜索, 目标检测, 方位维度, 平行搜索, 系列搜索

Abstract: Objective Treisman and Gelade put forward A Feature-Integration Theory of Attention in 1980. The theory included concepts such as visual search and illusory conjunction, among others. Visual search contained feature and conjunction searches. They suggested that search for the target defined by single features should be in parallel processing without attention limits in feature search. In conjunction search, they assumed that search for the target defined by more than one features’ conjunction should be serial processing with focal attention. The crucial factor of slopes (the functions of reaction time ´ set size) in visual search was an important research question. Duncan and Humphreys found that search was faster when differences between the target and distractors were larger and when differences among distractors were smaller. Their results provided qualitatively crucial factors for the slopes. This paper further quantified these crucial factors for the slopes in orientation dimension.
Methods The crucial factor (g (T, D, n)) was hypothesized. 12 adult subjects (6 female and 6 male) with normal or corrected-to-normal vision took part in the experiment. The experiment included a control and three experimental conditions. Stimuli in the control condition were randomly selected. Stimuli of the three experimental conditions were selected according to their experimental purposes. The g (T, D, n) values of all conditions were computed. All stimuli were lines for orientation dimension. Four conditions were feature searches of heterogeneous distractors. Set sizes (the number of target and distractors) of four conditions were 5 and 13. Whether g (T, D, n) values decided the slopes of visual searches was examined by comparing the control condition with three experimental conditions. The experiment adopted a within-subjects design.
Results The statistical results showed that difference between the control and first experimental condition in reaction times was insignificant, when their g (T, D, n) values were same. When the g(T, D, n) value of the second experimental condition was larger than that of the control condition, its reaction times were significantly short (F (4, 19) = 7.89, p < 0.01). Its slope was small. It shows that the degree of the difficulty in visual search in the second experimental condition was lower than that of the control condition. When the g (T, D, n) value of the third experimental condition was smaller than that of the control condition, the difference (F (4, 19) = 5.74, p < 0.01) between them in reaction time was also statistically significant, suggesting that the degree of the difficulty in visual search of the third experimental condition was higher than that of the control condition. The slope of the third experimental condition was large.
Conclusion When the g (T, D, n) values of the two conditions was same, their slopes was almost equal. The slope of the larger g (T, D, n) values was smaller. The slopes of the smaller g (T, D, n) values were larger. Therefore, the g (T, D, n) values might decide the slopes of visual search

Key words: visual search, detection target, orientation dimension, parallel search, serial search