Abstract
Visual crowding refers to the deleterious effects of closely spaced stimuli (flankers) on the ability to identify targets. Increasing spatial separation between flankers and targets reduces crowding, as does increasing their featural differences. Rosen and Pelli (2015, JOV) showed that a reduction in crowding caused by contrast-polarity differences between target and flankers is nearly eliminated when outer flankers that match the target are added to the display. Importantly, these outer flankers alone cause little crowding. This pattern crowding effect is similar to effects in which the perceptual organization of a display, rather than the immediately proximal stimulation near the target, seems to determine crowding (Manassi et al., 2015, JOV). Two alternative classes of explanation have been offered for these kinds of effects. Texturization accounts propose that crowding is caused by imprecision in the representation of image features following early filtering (Balas et al, 2009, JOV). A defining characteristic of texturization is that scene structure (e.g., figure-ground organization) is irrelevant in determining crowding. Segmentation accounts propose that crowding is caused by post-perceptual-organization confusion of target and flankers (Francis et al., 2017, Psy Rev), and display characteristics that facilitate target-flanker parsing will reduce crowding. A defining characteristic of segmentation is that scene structure is critical in determining crowding. We sought to test between texturization and segmentation accounts of pattern crowding by presenting opposite contrast-polarity flankers at close, medium, and/or far distances from the target while also manipulating the presence or absence of same contrast-polarity outer flankers. This created displays with varying target-flanker proximities and configural structures. Texturization predicts that the density and proximity of stimuli will determine crowding, whereas segmentation predicts that configural structure will determine crowding. The results are consistent with separate contributions from texturization and segmentation, suggesting that crowding effects derive from separate early and mid-level visual processes.
Meeting abstract presented at VSS 2018