The colour perception of natural objects :familiarity, constancy and memory

Abstract

Perceived object colour tends to stay constant under changes in illumination. This phenomenon is called colour constancy. Colour constancy is an essential component of colour perception and is typically studied in the laboratory via asymmetric colour matching experiments, in which the observer views two colours under two different illuminations side by side and makes matches between them. This situation is unlike colour constancy in the real world, which must typically involve a comparison between the colour one views and the colour one remembers - in other words, colour memory must be required. Furthermore, most colour constancy studies use twodimensional Mondrian images as experimental stimuli. These stimuli enable easy computer control of colour but exclude most of the natural perceptual cues such as binocular disparity, 3D luminance shading, mutual reflection, surface texture, glossy highlights, all of which may contribute to colour perception. My aim, in this project, is to study the colour perception of real objects in a more natural environment. To do so, I have developed an experimental setup which preserves the advantages conferred by easy computer-driven control of colour as well as the natural binocular and monocular cues to 3D shape. The setup also permits the use of real solid objects as stimuli, and the manipulation of their apparent surface colour as well as the background illumination. Thus, using this setup, I have been able to employ both 2D and 3D natural objects as stimuli and investigate aspects of colour perception related to colour constancy and colour memory as well as object familiarity. In developing and analysing these experiments, I have also introduced a new index of colour constancy which explicitly incorporates colour memory. My experiments reveal the following main principles: 1) colour constancy relies on colour memory, and is as good as colour memory allows; 2) colour and shape perception interact in both object similarity and discrimination tasks, indicating that colour and shape cannot be studied completely independently of each other; 3) object familiarity affects colour perception, for both foreground and background objects; 4) object familiarity also affects colour perception at perceptual levels, as measured by the reaction times and the range of appropriate colours accepted for an object.