Abstract
"Features" of a stimulus object are often specified by certain salient parts of that object (e.g., "eyes" constitute a facial feature). Existing work has provided evidence that "emergent features" such as orientation or proximity may be more salient than "elementary" features like location when two dots are configured in an appropriate fashion using the odd-man-out paradigm. We adapted the original task to a new sequential change detection task. This task had two main advantages. Participants could glean all of the information necessary to complete the task from a single fixation and "false pop-out" effects, induced by configuration of the dots across quadrants, were avoided. In one condition of the double-dot display, the "emergent feature" (i.e. configuration) changed along with location; in another condition the amount of location change was exactly the same, but there was no change in the configuration. In order to replicate the results of previous experiments, we also included uninformative context trials in which one dot is fixed while the other changes location. This will always induce a change in configuration. We successfully replicated the results of previous experiments using our new task. Additionally, we applied workload capacity statistics, which assess the performance when the emergent feature is (potentially) present against what can be expected from ordinary parallel processing of the constituent elementary features. The measures are nonparametric and have been shown to be robust and reliable in many other tasks. Using these measures, we found that efficiency is superior to ordinary parallel processing when an emergent feature is contained in the stimulus. Conversely, capacity was hampered when the emergent feature was absent, even though the information given by elementary features was the same in both cases. These results support orientation and proximity as emergent features and rigorously quantify the effect of a change in those features.
Meeting abstract presented at VSS 2012