Abstract
Many natural objects possess distinctive surface textures as well as shapes, both of which may serve as cues to object identity. Previously, we showed that the presence of natural chromatic texture and shape cues improved the accuracy and precision of memory color for natural objects [Vurro et al., VSS 2008]. Here, we investigate the effect of the diagnosticity and congruency of chromatic texture and shape on color discrimination of natural objects. We performed two experiments; subject's task was to match a test stimulus from two alternatives, presented either simultaneously (Experiment 1) or successively (Experiment 2) relative to the test stimulus. Six familiar fruits and vegetables were employed to generate the stimuli: apple, banana, carrot, cucumber, lime, and potato. For each object, three chromatic surfaces were used: (1) natural surface texture; (2) spatially scrambled version of the texture preserving its chromatic content; or (3) uniform surface with the texture's mean color. Each chromatic surface was combined with each of three 3D solid shapes: congruent (e.g. apple texture on apple shape); incongruent (e.g. lime texture on banana shape); or generic (e.g. carrot texture on rhomboid shape). On each trial, alternatives varied only in the mean hue angle of the surface chromaticity distribution. Simultaneous angular discrimination thresholds were significantly higher for chromatically variegated surfaces relative to uniform surfaces, and for incongruent shapes relative to congruent or generic. Successive discriminations were easier and faster for natural textures compared to uniform surfaces or scrambled textures, as were natural shapes compared to generic shapes. The results indicate a low-level mechanism, responsible for chromatic variegation generally degrading simultaneous hue discrimination, as well as a high-level mechanism, responsible for diagnostic chromatic texture and congruent shape improving successive hue discrimination. The latter effect suggests that memory color is linked to shape and incorporates representations of the natural spatio-chromatic texture.
Supported by EPSRC (Project EP/D068738/1).