Automaticity and attention in geometric visual illusions.

Abstract

Geometric visual illusions are automatic in the sense that their mechanisms seem unavailable to introspection and unaffected by volition. However, an additional quality that might be considered a criterion of automaticity is the independence of an illusion from any mental processing resources not fully dedicated to it. Little attention has been given to assessing that quality in visual illusions. Two independent variables were chosen to manipulate the level of nondedicated resources available for making judgments of visual illusions: concurrent workload in the form of a digit-search task, and stimulus duration. In the first three experiments, workload was manipulated while subjects made judgments of three illusions: the Ponzo, horizontal-vertical (inverted-T), and Muller-Lyer. For the Ponzo and horizontal-vertical illusions, deviations of subjects' points of subjective equality from the point of objective equality were greater when the digit search was required, indicating that the illusions were sensitive to the level of nondedicated resources available, and therefore were not fully automatic. A nonsignificant trend in the same direction was found for the effects of workload on the Muller-Lyer illusion. A fourth experiment required subjects to make judgments of the horizontal-vertical illusion while the illusion configuration was presented at four different durations (33, 100, 350, and 5000 msec). The magnitude of the illusion decreased with increasing duration, providing further evidence for the involvement of nonautomatic processes in the illusion. A fifth experiment investigated the relationship between automaticity of processes and experience. Subjects were given extensive experience with either the horizontal-vertical or Muller-Lyer figure while trials were presented with stimulus durations of 33 or 5000 msec. The horizontal-vertical illusion was again found to be smaller with longer stimulus duration, but neither the illusion nor the effect of stimulus duration on it changed with experience. The Muller-Lyer illusion was affected by stimulus duration early in experience, being greater with briefer stimuli, and both the illusion and the effect of stimulus duration on it decreased throughout the experimental session. The results of these experiments suggest that illusion magnitude is greater under conditions that limit the influence of flexibly allocated mental resources. Illusions, as they are typically measured, are therefore not mediated by fully automatic processes.