Abstract
A typical multiple object tracking task (MOT) requires participants to track some number of targets among a set of featurally identical nontargets. When targets and nontargets approach close to one another they can become confused, leading to errors. A simple model of MOT might expect that lost targets are lost completely —participants track the wrong item and now ignore the former target as though it is a nontarget. We investigated the possibility that participants may keep track of lost targets, hypothesizing that they are aware when hazardous close encounters lead to errors, and therefore, that they may track items not simply as targets or nontargets, but on a likely-target continuum. To test this hypothesis, we asked participants to first identify as many targets as originally assigned, and then to identify two more items —i.e. to make two additional guesses. Thus with four targets assigned, participants made six responses. On trials where an error was made in the set of the original four responses, the crucial question was whether additional guesses were made accurately at a significantly higher rate than would be expected by chance. They were. We also found that guessing was significantly better than chance when excluding trials in which participants may have used a proximity heuristic, i.e. selecting the nearest object to a previously selected one. Overall, our results suggest that observers may at times ‘keep tabs’ on targets that they suspect they may have confused with nontargets. These results have important implications for theories of tracking, generally supporting probabilistic mechanisms, and also implying that participants may sometimes track more objects than initially assigned.
Meeting abstract presented at VSS 2013