The perceived color of a light can depend strongly on the context in which the light is seen, but previous work oversimplifies the contribution of individual elements within the context by treating all elements of context as the ground in figure/ground segmentation. In natural viewing, most scenes have many objects, some with stronger ties to the focal test object than others. The color appearance of an object may depend on elements with shared properties, but what elements or shared relations are most important have been largely unexplored. The proposed research will develop a more refined understanding of how the various components of context differentially affect color perception. In particular, color appearance changes will be measured with some parts of the ground more strongly linked to the test than other parts.
Aim 1 compares color appearance shifts with elements separated in different depth planes. The degree of chromatic induction and chromatic adaptation from stimuli varying in S-cone stimulation will be measured with some elements linked to the figure within the same depth plane and others further removed. Perceived color shifts will be quantified with color matches to the appearance of a test object. The influence of depth will be evaluated by the changes in matches in conditions that vary the chromatic properties of the context at various depths.
Aim 2 will explore the degree of chromatic induction with elements linked to the figure by the direction of motion. The hypothesis to be tested is whether context in two-dimensional planar or three- dimensional depth motion will dominate chromatic induction cues when the test is perceived to move in the same direction.
Aim 2 will use similar methods as for Aim 1 to quantify perceived color shifts.
Aim 3 is concerned with chromatic adaptation mechanisms and their sensitivity to three-dimensional scene representations. Perceived color shifts will be measured following adaptation to low-frequency chromatic noise varying along a particular direction in color space (for example, the LM or S direction), with and without scission. Scission occurs when the photometric and/or geometric relations between the figure and ground support a percept of transparent layers. Perceived color shifts will be compared in conditions that vary the percept of scission. These experiments are guided by the overall hypothesis that all elements within a scene do not contribute equally to the color appearance of a figure, and that their differential influences depend on object relations, not simply retinotopic proximity. The experiments will provide new knowledge about the neural mechanisms of color perception that are active in natural viewing.
In natural viewing, each object we see is embedded within complex content of other objects simultaneously in view. While context is known to influence color perception, this research will provide a more refined understanding of how each part of the context alters the colors we see. This work is an essential foundation for understanding abnormal functioning within higher levels of visual neural processing, which are critical for seeing colored objects in natural situations.
|Elliott, Sarah L; Cao, Dingcai (2013) Scotopic hue percepts in natural scenes. J Vis 13:15|