One Whole or Two? The Flexible Role of the Observer in Perceptual Grouping

Abstract

The flexibility of perceptual grouping was examined from a functional perspective. If particular wholes are good for particular tasks, we can bend the rules of Gestalt organisation in order to see them? Evidence was gained by observing the effects of Gestalt factors on task performance, in the context of subjects' experience and expectation. Ellipsoids were split into two semi-ellipses by a central gap of variable width, which controlled the relative dominance of a one-whole percept (Narrow-Gap) or a two-parts percept (Wide-Gap). In a sequential comparison task, a reference ellipsoid was paired either with another ellipsoid (Whole/Whole), or with a single semi-ellipse (Whole/Part). Subjects responded “different” to a change in the shape of one semi-ellipse. Stimuli appeared for 50 ms and the reference stimulus was masked with stimulus onset asynchronies (SOAs) of 50 to 250 ms. Subjects were first trained either on Whole/Whole or Whole/Part only; both comparisons types were then mixed randomly. A loss of discriminability was predicted for Whole/Whole with Wide-Gap, and for Whole/Part with Narrow-Gap, but only for untrained and unexpected comparisons. The predicted interaction was significant. The effects were particularly strong with 50 ms SOA, suggesting an early involvement of task-contextual factors in perceptual grouping, and showing no evidence of global priority. Subjects appeared to readily learn flexible grouping strategies, abstracting only the behaviourally relevant relationships, and thus offsetting the effects of stimulus structure on perceptual grouping.