Figure 3.4. Schematic of Katz’s theory of color constancy. The stimulus color produced by abnormal illumination is cognitively transformed in the direction of its genuine color.
However, when a surface lies in abnormal illumination, such as a spotlight or a shadow, the stimulus color it produces is displaced away from its genuine color (the color it would appear in normal illumi-nation) in the direction of the abnormal illumination. In that case a central transformation process, rooted in the observer’s past experi-ence but triggered by the visual context, modifies the stimulus color back toward its genuine color, resulting in the percept. This transfor-mation process is not engaged when the object appears in normal illumination. These components are illustrated in Figure 3.4.
Four terms are essential to Katz’s account of constancy: normal il-lumination, genuine color, stimulus color, and transformation. Ironi-cally, with the possible exception of genuine color, these concepts were not revealed to Katz through phenomenology. They were merely the available tools of the day. Perhaps in an effort to ground these con-cepts in phenomenology, Katz suggested that both the stimulus color and the transformation process were revealed by the use of the re-duction screen in his basic light/shadow constancy experiment. The reduction screen, by occluding the visual context, reveals the stimulus color of a target surface: “the various procedures of reduction, how-ever, give us some information concerning whether the retinal pro-cesses corresponding to various impressions are alike or not” (Katz, 1935, p. 83). Thus, when a target in the lighted region and a target in
the shadowed region are viewed through a reduction screen and ad-justed to appear identical, we can say they produce equal local stim-ulus and have equal stimstim-ulus colors. When the reduction screen is removed, the two targets appear different. But, as Katz notes, only one of the two targets appears to change, and that is the target in the abnormal illumination. This change reveals the transformation process in action, and it further reveals that only the target in the abnormal illumination is transformed.
Katz did not reject the duality between local peripheral stimulation and central interpretation, shared by Helmholtz and Hering. He re-jected only the claim that both contrast and constancy could be ex-plained by a single approach, either sensory or cognitive. As Gelb wrote (1929, quoted in Ellis, 1955, p. 203), “The principal difference between Katz and his predecessors is his claim that colour contrast and colour constancy require essentially different explanatory princi-ples.” Von Fieandt (1966, p. 217) has written, “Already in his first edi-tion Katz aimed at a synthesis of the dualistic view shared by both Hering and Helmholtz. Just as Hering included both physiological and ‘higher-order’ psychological (memory colors and learning) among his explanatory principles, so Helmholtz too referred both to tradi-tional physiological determinants and to something reminiscent of cognition and reasoning, his famous ‘unconscious inferences.’ ”
Katz’s influence in the prewar period was so great that his term transformation came to be used as a virtual synonym for the phenom-enon of constancy, even though Katz himself used it to refer to the psychic process that produces constancy. MacLeod (1932, p. 37) ob-jected, “Inasmuch, however, as the transformation of experience im-plies the pre-existence of a specific untransformed experience, it is best to use the term only in the discussion of a point of view which pos-tulates a transformation process, as was the case in Katz’s original work.”
Katz’s Failure to Escape the Zeitgeist
Katz criticized Helmholtz’s theory of constancy as overly intellectual, yet it is difficult to see how his own theory is different. Transforma-tion, like unconscious inference, is seen as a central process and de-pends on the observer’s past experience. Both concepts are vague and speculative, and both run into the same difficulties. More concretely, neither concept is workable unless the illumination level can be de-termined.
Indeed, Katz’s theory is undermined by his own experimental work. It is not clear how Katz’s theory of constancy can be reconciled with his own research showing strong constancy in animals. And his findings on the importance of articulation and field size for constancy have no comfortable place in his theory. No explanation is given for
why high articulation or large field size should produce a more suc-cessful transformation process or allow a better determination of the illumination level. Ironically, it was Kardos who later found a place in lightness theory for these two concepts.
But the biggest problem for Katz’s theory came from an experiment by Gelb that, while disarmingly simple, was devastating to Katz’s theoretical edifice: Gelb’s paradoxical experiment.
SUMMARY
The publication of Katz’s book on color in 1911 launched a 25-year period of vigorous research in lightness. Katz took a phenomenolog-ical approach, describing the various modes in which colors appear.
In particular, he drew a sharp distinction between surface colors and film colors. He also noted that lightness and brightness are separate phenomenal dimensions of surface colors. Most importantly, though not widely appreciated, it was Katz who gave us our basic psycho-physical methods for the study of lightness, including asymmetrical matching in side-by-side fields of illumination and shadow. Using these methods he demonstrated that lightness constancy is not com-plete. He created the first measures of the strength of constancy, mea-sures that were later improved upon by Brunswik and Thouless. Katz found that the degree of lightness constancy correlates with both the size of a field of illumination and with its degree of articulation.
Katz himself did not break new ground theoretically. The work he inspired grappled with several issues disputed by Helmholtz and Her-ing, such as the role of central and peripheral factors and whether constancy can be reduced to contrast. To this end, many lightness constancy experiments were conducted with animals. Lightness con-stancy in the absence of contrast factors was tested (and found) using a method I have called countershaded backgrounds: targets in high illumination were presented in front of dark-gray backgrounds while targets in low illumination were presented in front of equi-luminant light-gray backgrounds.