Color constancy, lightness constancy, and the articulation hypothesis.

The light that reaches the eye has a history. Its spectral distribution at each point of the retina depends on the illumination present in the scene, the spectral properties of surfaces it encounters on its path to the eye, and other factors (Nassau 1983). The human visual system assigns colors to surfaces. If the assigned color of a surface patch is (approximately) determined by the spectral properties of the patch itself, then the visual system is (approximately) color constant. It is sometimes said that human color vision is `̀ approximately color constant'', but this claim is very misleading. Under some circumstances, constancy fails dramatically (Helson and Judd 1936); under others, it holds up remarkably well (Brainard et al 1997; Brainard 1998). We do not know what factors lead to more or less color and lightness(1) constancy in human vision. In the 1930s, Gestalt psychologists introduced the term articulation as a label for those scene attributes that enhanced the stability of perceived color and lightness. These included the number of distinct surfaces present in the scene (ie numerosity) and the degree of depth variation (ie three-dimensional structure), as well as other factors. As Gestalt psychologists used the term, `articulation' was both ill-defined and ambiguous. Discomfort with the resulting confusion in terminology led Henneman to conclude that `` ... this rather vague term [articulation] is badly in need of clearer definition and explanation'' (Henneman 1935). Sixty-five years later the vague term `articulation' continues to confuse. In 1999, a special symposium on articulation and lightness perception was held at the 22nd European Conference on Visual Perception in Trieste, Italy. There was little agreement among the speakers as to what the term `articulation' could or should refer to, or even whether it is a useful tool to organize our thinking about human color vision. One of us (LTM) argued that the term should never be used again. Yet, underlying the terminological confusion, there remains the scientific issue: what factors contribute to the stability of surface color perception? For this special issue, we invited researchers to explore candidate factors and to discuss their results in the context of the term `articulation'. The papers accepted for the special issue divide neatly into four groups, each emphasizing a different factor. The factors are numerosity (Gilchrist and Annan; Linnell and Foster), configural cues (Schirillo and Shevell; Logvinenko; Neumeyer et al; Spehar et al), variability (Mausfeld and Andres; Brenner and Cornelissen), and three-dimensional structure (Bloj and Hurlbert; Kraft et al). We next summarize the papers by group. Color constancy, lightness constancy, and the articulation hypothesis Perception, 2002, volume 31, pages 135 ^ 139