Abstract
The uncertainty paradigm has been used in vision research to evaluate whether stimulus components are processed independently, or whether dependencies exist. The paradigm consists of several experimental conditions from which sensitivity indices are estimated and combined to provide evidence for or against independence. The paradigm presents observers with a stimulus consisting of multiple components, such as a face. The values of these components differ across stimuli; for example, the size of the nose, or the distance between the eyes may differ. Observers are required to decide if the change in a component across stimuli constitutes an "increment" (i.e., the eyes are farther apart) or a "decrement" (i.e., the eyes are closer together) compared to a standard value. The task involves two conditions. In the certainty condition, the observer knows which component will contain the change. In the uncertainty condition, the component that differs from standard is unknown. Performance in each condition can be compared to that which is predicted by the independence of components. Previous applications of the uncertainty paradigm have not adequately described the foundations upon which performance indices can be understood as relating to component independence. We sought to clarify these concepts and demonstrate how to apply the results using exemplary data from a study requiring observers to make simple judgments about facial features. Specifically, we derived predictions for observer sensitivity in the uncertainty condition and implemented a relative measure of root-mean-square (RMS) that incorporates performance in both uncertainty and certainty conditions. This was carried out for three major signal-detection based decision models: a distance-classifier, an optimal decision model (ideal observer), and a decisionally separable "independent" decisions strategy. We also considered, in the context of these models, implications for sensitivity and RMS when stimulus components were perceptually correlated.
Meeting abstract presented at VSS 2012