Diverse long range configural judgments use a single map of object locations

Abstract

Purpose. To test whether configural judgments across large distances in visual space share a common, abstract representation of object locations. Method. Let T be a perceptual task with dependent variable v (e.g., percent correct in a 2AFC judgment) in which performance is limited by the salience of targets on a background, and let S₁, S₂,…, S_N be stimulus properties (e.g., brightness, darkness, greenness, etc.) that can be used to define targets and adjusted in intensity to control performance in task T. Then for any level p of performance in T let Q_T(p) be the N-dimensional vector whose k^th component is the intensity of P_k that yields performance p in task T. The range of Q_T (i.e., the locus of N-dimensional vectors visited by Q_T) is called the task T equisalience curve for S₁, S₂,…, S_N. If two tasks share the same equisalience curve for a set of properties, this supports the hypothesis that they are informed by a single, shared representation. We used stimuli comprising widely separated (all by at least 10 deg.) texture-defined discs on a textured background. Discs were defined by being either brighter, darker, greener, higher in contrast variance or lower in contrast variance than the background. Equisalience curves were measured for (1) three different configural tasks, requiring judgments about the relative locations of three discs, and (2) a task in which the observer judged which of three (widely separated) discs was different in size from the other two. Results. The equisalience curves for the three configural tasks were identical (within measurement error) but differed significantly from that for the size-judgment task. Conclusion. These results support the idea that long range judgments of spatial configuration make use of a shared, abstract map of object locations.