关键词: 多目标追踪, 抑制, 视觉标记, 运动方式

Abstract: Previous research on stimulus inhibition has found that there are two types of inhibition in terms of what is inhibited. One is the feature-based inhibition, which holds that it is the unattended objects which share the same property with the attended ones that are inhibited by the cognitive system (s) (see the research of visual marking by Juang, Chun, & Marks, 2002; Olivers & Humphreys, 2003; Watson & Humphreys, 2002). The other is the object-based inhibition which states that the individual tokens of nontargets are inhibited, and that the inhibition of the tokens remains even when those nontargets are moving around. The inhibition theory was used to account for the results of probe-dot detection with the Multiple Object Tracking (MOT) task. That is, the difference in performance of dot detection is caused by the inhibition of the nontargets. However, no evidence has shown whether the nontarget inhibition in the MOT task is feature-based or object-based. The present research aimed to answer this question by employing the same task (dot detection in the MOT task).
Participants in two experiments performed either the tracking or no-tracking task first, and then they de-tected the presence of a dot at a location occupied by the randomly-moving target, by the randomly-moving nontarget, by the static (or regularly-moving) nontarget, or by nothing. The correct percentage of tracking and probe detection were measured. As a major concern, the difference in accuracy of dot detection in each location condition between the tracking and no-tracking tasks was computed as an index of inhibition, because all the stimuli in the no-tracking task were processed in the same way, and it is assumed that there is the same amount of inhibition or no inhibition to all the stimuli. The only difference between the first and the second experiment was the feature similarity (in terms of moving pattern) of nontargets to targets. Experiment 1 included ran-domly-moving targets, randomly-moving and static nontargets, whereas Experiment 2 included randomly- moving targets, randomly-moving and regularly-moving nontargets.
The results of Experiment 1 showed more inhibition to the randomly-moving and static nontargets, relative to the targets and the space. There was no significant difference in inhibition between the randomly-moving and static nontargets. However, in Experiment 2, more inhibition was obtained to the randomly-moving targets than to the regularly-moving targets. Still, the inhibition to nontargets was larger than to the targets and the space. The t-tests for both experiments showed the only significant difference in inhibition to randomly-moving non-targets between the first and the second experiment. More inhibition to the ramdomly-moving targets was ob-served in the second experiment than in the first one. In general, the findings of the present study suggested that both the feature-based and the object-based inhibition can occur in the MOT task. When the target and the non-target are easy to distinguish from each other (e.g., the moving target and the static nontarget), the token of nontargets is inhibited (as shown in Experiment 1). Whereas, when it becomes more difficult to distinguish the target from the nontarget (e.g., the moving target and nontarget share more same features), feature-based inhibi-tion begins to play an important role (i.e., more inhibition to the nontargets that are more similar to the targets).

Key words: multiple object tracking, inhibition, visual marking, moving pattern