Abstract
Previous research indicates that symmetry is especially ‘good’ when it is defined by a single object while it is weakened when symmetrical information is distributed over multiple objects (Baylis and Driver 1994,2001). Our current research investigates the robustness of symmetry to temporal offset. Temporal offset can serve as a cue for figure/ground segmentation (Leonards, Singer & Fahle 1996). By means of two experiments we investigate whether temporal segmentation into multiple objects impairs symmetry perception.
In our first experiment subjects had to discriminate patterns that were perfectly symmetrical from patterns that were partially symmetrical (i.e. stimuli that were composed of a symmetrical part and an asymmetrical part). In the second experiment subjects had to discriminate stimuli that were completely asymmetrical from partially symmetrical stimuli. Temporal offset of the two stimulus parts ranged from 0 to 100 milliseconds.
The results show that the identification of completely symmetrical stimuli is not affected by temporal offset while the identification of completely asymmetrical stimuli is. This suggests that symmetry is robust to temporal offset.
For the partially symmetrical stimuli the results were as follows. When the symmetrical part is presented last, performance decreases when subjects had to discriminate these patterns from perfectly symmetrical patterns (Exp. 1), while they showed improved performance when discriminating these patterns from completely asymmetrical patterns (Exp. 2). When the symmetrical part was presented first there were no effects of temporal offset. We argue that these results can be explained by an order-dependent salience effect of symmetry.