Journal of Mind and Behavior, 1989, 10, 361-372
A Sociohistorical Critique of Naturalistic Theories of Color Perception
Carl Ratner
An awkward truce prevails between naturalistic and sociohistorical theories of color perception, with each position dominating different aspects of the phenomenon. By mutual agreement, the boundaries of color categories are explained as deriving from social experience. Congruent with Sapir, Whorf, Vygotsky, and Luria's conception of socially mediated psychological processes, perception of color boundaries is construed as being shaped by language and other social practices. Parents literally teach children color boundaries by referring to certain colors with the same linguistic code, while other colors are designated by other codes. When an American parent asks her child the name of blue and green objects, and the child answers with the same word "green," the parent rebukes the child and readjusts his categorization system by insisting that "no, that object is blue, not green." Psychologists falling within the rubric of sociohistorical psychology maintain that individuals come to perceive (experience) colors according to this kind of socially mediated experience. In addition, color perception will manifest significant cultural variation insofar as different societies emphasize different color categories. Sociohistorical psychology emphasizes the factthat sensory information is selected, interpreted, and organized by a social consciousness. Perception is thus not reducible to, or explainable by, sensory mechanisms, per se. Sapir, Whorf, Vygotsky, and Luria all maintained that sensory processes are subordinated to and subsumed within "higher" social psychological functions. (Luria, 1932; Whorf, 1956, pp. 267-268; Vygotsky, 1978. Cf. also Brown and Lenneberg, 1954, p. 465; Schaff, 1973, p. 191).
While psychologists generally accept a social account of color boundaries, they invoke entirely opposite concepts to explain the perception of the mid-points of color categories, known as focal colors. Perception of focal colors is construed as determined by a natural, species-wide, physiological, sensitivity to these colors independent of social experience. Socially mediated consciousness does not override natural processes in this domain. Instead, natural processes determine focal color perception without interference from social consciousness.
Although this resolution of naturalism and sociohistorical psychology commands mutual wide respect, it cannot produce lasting peace. For the two theories invoke different premises and processes which cannot both be true. It is illogical to expect that the perception of color boundaries depends upon social learning and manifests cultural variation, while perception of category mid-points is governed by natural, universal mechanisms independent of social experience. How could the single act of color perception be composed of two disjointed, incompatible processes? How could they operate together? What would the switching mechanism be that invoked one process for boundary perception and another process for focal color perception? How could conscious mediation be called up in one case but not the other? Or, alternatively, suppressed in one case but not the other? What functional advantage would be gained by such an awkward juxtaposition of dichotomous operations?
In view of this problematic it is necessary to re-open the debate between naturalistic and sociohistorical theories in order to discover a more principled resolution. There are several ways to embark upon such a resolution. One way is through reviewing the premises of each position with the aim of finding an advantage of one over the other or a possible new synthesis of them both. Because both positions have been clearly articulated in the literature, I shall resist the temptation of taking this theoretical tack, leaving the reader to pursue it on his own. Instead, I shall analyze the empirical research that is offered as support for the naturalistic position. I shall demonstrate that the research on focal colors fails to justify the universalist theory and actually indicates that perception of focal colors follows sociohistorical principles. This empirical critique suggests that all aspects of color perception -- including the perception of color boundaries and focal colors -- can be parsimoniously explained by a unitary set of sociohistorical psychological principles.
The research on focal colors attempts to establish a natural foundation by demonstrating universal sensitivity to focal colors. The assumption is made that universality of responsiveness reflects natural processes. We shall have occassion to argue later on that this assumption is fallacious. However, for the moment, we shall let the assumption pass and confine our examination to ascertaining whether focal color perception is universal in the first place. A review of the research on focal colors must necessarily begin with the founding work of Berlin and Kay. While their methodology has received severe criticism, these critiques have done little to slow down acceptance of naturalistic conclusions. Consequently, a brief review of these critiques is warranted. Berlin and Kay's data will then be freshly re-analyzed with the intention of reinforcing the earlier methodological critiques. We shall then proceed to analyze other naturalisticresearch on focal colors.
Berlin and Kay's (1969) pioneering study established focal colors as the universally agreed upon, best prototypical examples of certain colors. The colored sets from which the focal colors were selected were dubbed "basic colors." Now, any claims about the universality of focal colors depend upon the universality of the basic colors from which they were selected. After all, if the basic colors are not universal, then the focal colors certainly cannot be either. Berlin and Kay's procedures for deriving the basic colors make it extremely unlikely that these colors are universally salient. Instead, they seem to be salient primarily to Western enculturated people, which taints the focal colors with the same ethnocentric bias.
Suspicion about the universality of the basic colors starts with the subjects who provided the information for generating the basic colors. The subjects were asked to state the names of colors most salient to them, and to identify these on a color chart. Although supposedly representing diverse cultural groups, they, in fact, were bilingual foreign residents of San Francisco. Their obvious social and linguistic enculturation would be expected to make their color perception continuous with native Americans, and any similarity cannot be presumed to be due to universal biological processes. Compounding this cultural restriction in the subject population is Berlin and Kay's homogenizing of the data they produced. From the numerous colors the subjects selected as salient, Berlin and Kay arbitrarily decreed that what qualified as a basic color were only those colors whose verbal description was a single, abstract color word. Colors whose names were also object names were excluded. This arbitrary criterion excluded 90% of the colors with which subjects are familiar. Moreover, the restriction contained a crucial cultural bias: abstract color names (like all abstract stimuli and concepts) are characteristic of modern society, which means that Berlin and Kay's criterianecessarily biased the selection of colors to those that are noticed in our culture. Berlin and Kay's calling these colors "basic" is therefore quite ethnocentric. It also creates the false impression that they are naturally salient when, in fact, they were simply imposed by the authors. It cannot be emphasized too strongly that the basic colors from which focal colors were later selected, were arbitrarily selected by Berlin and Kay from a larger list of colors provided by the subjects.
The "epistemological chauvinism" (Bousfield, 1979) that haunts Berlin and Kay's methodology has yet another expression. The stimulus colors were further homogenized by arbitrarily giving them certain English names which are not necessarily accurate. For example, the Tiv language has one color category that includes "warm colors" from red to yellow to brown. When Tiv subjects indicated the boundaries to this category, as well as the best example (focal color), their choices were arbitrarily glossed as "red." This obscures the possibility that they actually perceived the category and the best example as brown, yellow, or even some color that is quite foreign to Americans (Hickerson, 1971, p. 267). Berlin and Kay never investigated the psychological meaning or experience that the color had for the subjects. They unjustifiably assumed that the Tivs and Americans had the same perceptual experience (Bousfield, 1979, p. 206).
Given such sanitized stimulus materials and such homogenized subjects, it would not be surprising had they agreed on the best example, or focal color, of each basic color category. What is surprising is that the subjects' choices of focal colors varied dramatically. Berlin and Kay's Figure 3 on page 9 of their book shows an enormous distribution of focal choices across subjects of different cultures. For example, focal green varies from our green-yellow through our green to our blue. In other words, certain cultures pick the best example of "green" where we see blue. Thedistributions of focal orange and yellow also overlap, which means that what some cultures see as the best example of orange corresponds to what other cultures select as the best example of yellow. In view of this data, it is misleading for Berlin and Kay to claim that "the foci of basic color terms are similar in all languages" (p. 10). Just as "basic color terms" are not at all basic, neither are focal colors really focal in the sense of being a consensual converging point of color judgments. Instead, focal colors evidence dramatic cultural variation indicating that their perception is shaped by social experience.
Lumsden's (1985) attempt at proving humans' natural, universal sen-sitivity to focal colors is equally unsatisfactory. Lumsden found that 16-week old infants, presented with a spectrum of colored lights, habituated to the repeated presentation of the lights and stopped paying attention to them. They only renewed interest when discriminably different lights were presented. Lumsden construed these boundary points as color boundaries. He found a high correspondence between these color boundaries and the distribution of focal colors that Berlin and Kay plotted in their Figure 3. In other words, Berlin and Kay's distribution of red focal colors ranged over the same wavelengths that comprised one of the infants' color boundaries. Altogether, Lumsden identified four color categories separated by three boundaries for children which corresponded to similar ranges and boundaries in adult focal color perception. Lumsden concludes that the congruence is due to the fact that infants' natural sensitivity is the basis of adult perception of focal colors.
This conclusion is invalidated by Lumsden's highly questionable assumptions and procedures. In the first place, it is not obvious that infantile habituation to a group of colored lights is equivalent to adult color perception. We do not know that the range of lights to which an infanthabituates is what an adult sees as being one color. In fact, Lumsden's own evidence suggests that this is not the case. The range of colored lights to which the infants habituated and which Lumsden called yellow is the same range in which Berlin and Kay's adults typically distinguished three colors -- yellow, orange, and brown. Proclaiming these three adult colors to be identical with the infants' one is unjustified at best.
Secondly, Lumsden confuses the focal colors that adults perceive with color categories that infants purportedly perceive. Focal colors are obtained by asking adults to choose the best example of a range of colors. These best examples are not equivalent to the full color category which includes marginal colors as well. Infants obviously have no concept of a best example and their spontaneous habituation to colored lights indicates, if anything, color categories, not focal colors. Lumsden's data do not support claims regarding focal color perception because focal colors were not investigated.
Eleanor Rosch-Heider further overstated the universality of focal color perception. She further homogenized Berlin and Kay's colors by selecting the most commonly chosen focal color within each of their categories (Heider, 1972). These eight colors were presented to American and bilingual foreign students residing in the U.S. -- another extremely homogeneous sample. The hues were presented in varying saturations and subjects were asked to pick the saturation (purity, richness) that was the best example of the "basic color term". To reiterate, subjects were not asked to choose the hue that best represents the color, they were only given one hue for each color and asked to choose the best saturation (Although Rosch calls this choice "focal color," the term is a misnomer because subjects were selecting saturation, not hue. A more appropriate term is "focal saturation.") Approximately 95% of both foreign and American subjects chose the highest saturation levels asthe best representative or focal color.
Rosch's construal of these findings as proof of a physiologically conditioned, universal sensitivity to certain natural prototypical colors is quite gratuitous. In the first place, universality has been precluded through a narrow homogeneous sample of subjects as well as stimuli. It is quite inappropriate for Rosch to restrict her sample to Americans and foreign students residing in the U.S. and then proclaim her results as universal. Rosch did attempt to establish universal perception of the highly saturated focal colors in a genuine cross-cultural study. She compared Americans with Dani subjects from Indonesian New Guinea whose language only contains two color terms, light and dark. Although Rosch concludes that the two groups responded similarly to focal colors (saturations) her data do not support the naturalist claim. One indication of perceptual differences was the Danis' "unwillingness" to distinguish focal colors from nonfocals. When presented with a set of one focal and two nonfocal colors (saturations), the Dani subjects "were unwilling to designate one of the color chips as the most typical member of the three-chip category" (Rosch, 1973, p. 340). It is not clear whether this unwillingness was due to the Dani's failure to distinguish the ethnocentrically-derived focal colors; but whatever the reason, the fact remains that they did not select the supposedly universal focal colors.
Passing over this disturbing anomaly, Heider employed a more circuitous approach to demonstrate universal sensitivity to focal colors. She hypothesized that the focal colors' salience would make them more easily remembered than nonfocal colors, and this should hold true even for cultures lacking the basic color terms. In other words, contrary to Sapir-Whorf, focal colors are salient independent of language and should be remembered even in the absence of appropriate color terminology. In one experiment (Heider, 1972) Dani and Americans were presented eight highly saturated focalcolors as well as 16 nonfocal colors of different hues and saturations. A color was presented for 5 seconds, followed by a 30 second unstructured waiting period, after which an array of 160 colors was shown. Subjects had to identify the stimulus color in the array. Results were as follows: both Americans and Dani recognized more focal colors than nonfocal, however the differences are complex as the following summary table indicates.
------------------------------------------------------------
Americans Danis
Stimulus colors Stimulus colors
8 focal 16 nonfocal 8 focal 16 nonfocal
No. correct 5.25 5.73 2.05 1.18
------------------------------------------------------------
Figure 1. American and Dani recall of focal and nonfocal colors
(from Heider, 1972)
While the data might be taken as support for Rosch's hypothesis about the superior memorability of focal colors, Lucy and Shweder (1979 ) argue that Rosch's finding is an artifact of the array of test colors in which focals are more easily recognized than nonfocals. Using a modified array, but otherwise following Rosch's procedure of a 5 second presentation followed by a 30 second waiting period, memory for nonfocals improved to where it was equal with memory for focals. Garro (1986) rejects this critique and contends that if the 30 second waiting period is endured silently -- instead ofallowing conversation as Lucy and Shweder did -- focal colors evidence a memory advantage over nonfocal colors. Leaving aside the question of whether memory for focal colors is or is not superior to nonfocal color recall, Rosch's data itself can be interpreted as demonstrating enormous cultural differences in memory for focal colors. After all, Americans remembered 5 of the 8 focal colors in contrast to the Dani who only remembered 2 of 8. And this inter-cultural difference in memory of focal colors (5.25/8 - 2.05/8 = 3.2/8) is far greater than the intra-cultural difference between memory of focal and nonfocal colors (for Americans: 5.25/8 - 5.73/16 = 2.38/8; for Dani: 2.05/8 - 1.18/16 = 1.46/8) Put differently, the cultural difference in memorability of focal colors is greater than the difference between focal and nonfocal colors. In other words, culture and language have a greater impact on memory than focality does!
Rosch does acknowledge the Danis' poor memory for focal colors, however she attributes this to their poor memory in general. Their simple, uneducated life lead the Dani "to have neither need for, practice with, nor any explicit training in the use of memory control processes" (Heider, 1972, p. 19). This explanation is preposterous in assuming that the failure to remember Heider's ethnocentrically derived focal colors reflects a general deficiency in memory. It is exactly the same kind of epistemological chauvinism that underlies the claim that low scores on IQ tests reflects low general intelligence. A far more plausible interpretation of the Dani's poor memory for the colors is the fact that the materials appeared strangely unfamiliar to them. Primitive people who conceive of color as profoundly implicated in specific objects would obviously be bewildered by individual, discrete color chips, divorced from any concrete object. Moreover, the arbitrary task of remembering such meaningless stimuli apart from any realpurpose -- as an abstract act of free recall -- would compound the bizarreness of the experiment for Dani subjects. Interestingly enough, although Rosch's explanation is incorrect, it is a cultural explanation which invokes the Danis' life style as the reason for their poor memory. This explanation contradicts the naturalistic hypothesis it seeks to support. Rosch's hypothesis, after all, is that focal colors are naturally perceived and remembered because of an intrinsic sensitivity to them. She frequently insists that this sensitivity is unconditioned by culture or language. Yet her explanation for the Dani's poor memory depends entirely upon the social stimulation of memory. Her own conclusion thus contradicts rather than confirms the naturalistic theory.
In addition, neither Rosch's social nor her biological explanation can account for the failure of Americans to remember a sizeable proportion of the focal colors. American's complex social life and sophisticated memory control should enable them to remember virtually all of the focal colors, especially if combined with a natural sensitivity to them. Yet Americans inexplicably forgot 3 of the 8 focal colors.
Harkness's (1973) purported confirmation of Rosch's and Berlin and Kay's universal focal colors is as self contradictory as theirs. Harkness commits numerous methodological and conceptual errors which are surprising because her other writings have expressed a deep sensitivity to the cultural character of psychology.
In brief, Harkness presented Berlin and Kay's "basic colors" to children and adults in two ethnic groups (Spanish and Mayan speaking) in Guatemala. She asked the subjects to name the colors and select the best examples of each. While color areas (boundaries) varied markedly between the two groups, "the best examples of red, green, and yellow adult samplescorrespond exactly" (p. 182). The results are interpreted as giving "strong support to Berlin and Kay's finding on the universality of best examples of basic color terms" (p. 199). The reason for such universality is that basic color terms represent `natural categories' as Heider asserted (p. 200).
Strangely, the data support none of these contentions. In the first place, the two groups were linguistically quite close which could plausibly account for any perceptual similarity. Assumptions about natural, universal perceptual sensitivity are therefore unwarranted. The entire Mayan community spoke a bastardized dialect that included many Spanish words. The Mayan adults even used some Spanish color terms during the test. In addition, the Mayan-speaking children were students in a bilingual class. In fact, most 11-12 year olds in the Mayan community were so strongly bilingual they were excluded from the analysis (p. 188).
Furthermore, the areas of agreement on color terms and focal colors were limited to red, yellow, green and white. Beyond this commonality, the Spanish speakers named three additional categories -- blue, brown, and purple -- for which they selected best examples. These categories and best examples were not selected by the Mayan speakers. Even more significant is the fact that the intercultural areas of agreement are more apparent than real. This stems from the unworkable definition of focal color that Harkness employed. Focal colors are simply the chip chosen most frequently as being the best example of each color term. Harkness's minimal criterion was that 15% of subjects must choose a single chip for it to be designated as focal. Of course, sometimes more than 15% selected the same chip, but the low criterion means that many (if not most) of the focal colors were only selected by a small proportion of subjects; they do not represent widespread agreement at all. Consequently, intercultural commonality of focal colors simply means that as few as 15% of individuals in both cultures chose thosechips as the best examples. Given this, it is absurd to conclude that intercultural commonality of focal colors demonstrates a universal agreement that is engendered by a universal perceptual process.
In fact, it is misleading to even suggest universality when only 15% of the subjects are in agreement. Furthermore, Harkness admits that the percentage of subjects choosing the focal colors differed between the two groups which is further evidence for disagreement. She says (p.193) that the Spanish-speaking subjects demonstrated lower agreement amongst themselves than the Mayan-speakers.
Not only was agreement within each culture low on choosing focal chips. Agreement was also low on naming the chips. Less than 70% of the Spanish-speaking adults named the same chips pink, purple, orange, and grey. This led Harkness to conclude that "for the Spanish sample not all of the `basic color terms' are equally basic" (p. 184). Actually, Berlin and Kay's arbitrary selection of color terms makes none of them basic.
An additional piece of evidence contradicts Harkness's conclusion about universal color perception and demonstrates its cultural character. This evidence concerns developmental changes in color perception. Children chose somewhat different focal colors than did adults. Spanish adults chose two focal browns, neither of which agreed with the one focal brown selected by the 11-year olds. Furthermore, children of both samples manifested lower agreement in naming most of the colors than adults did (pp. 184-185). Whereas 100% of the Mayan adults agreed on which chips were yellow, less than 90% of the children agreed. And whereas 60% of Spanish-speaking adults agreed on purple, only 15% of the children agreed. Since there is no known sensory difference between 8-years olds and adults that would account for these discrepancies, the likely explanation is that the more extensive experience of adults has taught them a common set of colorschemas which children have not yet learned.
Social experience, rather than sensory maturation, is evidently also responsible for the fact that Spanish adults indicated significantly more focal colors than did Spanish children (pp. 189, 191). The children selected five focal colors (according to the criterion of at least 15% consensus) -- white, black, red, green, yellow -- in contrast to the adults' ten. In contrast, Mayan children and adults chose the same number of focal colors -- four and five, respectively (pp.186-187). This is an additional cultural difference that Harkness overlooks in her advocacy of universal focal color perception.
In conclusion, Harkness's data, like that of other naturalistically oriented psychologists, demonstrate cultural variation, not universality, in focal color perception. Evidently, humans are not all equally sensitive to the same focal colors. Since no differences have been discovered in the visual apparatus of the subjects in the foregoing studies, the substantial variation in focal perception suggests a cultural rather than a natural underpinning.
All the preceding cross cultural research is premised on the idea that common responsiveness to focal colors reflects and demonstrates a natural, sensory determination of perception. For instance, Bornstein, in arguing for the primacy of perceptual mechanisms over culture, makes the extreme claim that "hue boundaries, as well as hue foci, are given by the biological sensitivity of the organism" (1975, p. 791; Cf. also Kay and McDaniel, 1978). This assumption is fallacious because similarities in focal color perception could be due to environmental factors. Focal colors might possibly be more prevalent than nonfocal colors in the environment, in which case preference for and sensitivity to saturated colors would hinge on familiarity with them. Until this possibility is ruled out, physiological sensitivity to saturated colors cannot be presumed. It is simply a logical non-sequitur to conclude naturaldetermination from universal response tendencies. Of course, the striking absence of universal responsiveness to focal colors makes this issue quite moot and challenges any notion of naturalism.
A more direct challenge to the naturalistic theory of color perception is Lucy and Shweder's (1979) direct demonstration that color perception and memory are not natural but involve higher social-psychological functions. Taking the unconventional step of talking to their subjects to ascertain the psychological activity of color perception and recall, Lucy and Shweder found that subjects coded colors in preparation for recall by associating them with names of colors as well as with names of familiar objects. Some examples are "pea soup green but a little lighter" or "the color of my bedroom." The test colors in the array were then compared against this verbal code in order to match the stimulus color. Subjects generally mediated their color perception by culturally provided cognitive and verbal codes. Very few subjects reported using pure visualization without the verbal element (Lucy and Shweder, 1979, p. 593). Emotions also mediate color perception since subjects had strong emotional reactions such as "ugly," "pretty," and "dull" to specific colors. Color perception of focals and nonfocals alike is therefore definitely not a natural sensitivity predetermined by physiological processes. It is a psychological process intimately involving social thinking and language. Focal color perception is not an automatic response to sensory information. It is intentionally constructed by a social consciousness that selects, compares, and interprets sensory information in order to yield a resultant perceptual response. The interdependence of focal color recognition with their social familiarity is reflected in Lucy and Shweder's finding that interpersonal codability of the stimulus and array colors correlates more than 0.5 with recognition memory (Cf. also, Brown and Lenneberg, 1954; Lantz and Stefflre, 1964; Stefflre, Vales, and Morley, 1966).
Lucy and Shweder's findings strip away the veil of reification that impersonal, "objective" experimental procedures impose. Traditional positivistic procedures obscure the social psychological activity that constitutes a response, and this lends credence to the illusion of naturalness. But illuminating the activity that actually generates responses reveals the intentional, socially mediated effort that underlies color perception -- and psychological functions in general.
References
Berlin, B., and Kay, P. (1969.) Basic color terms, their universality and evolution. Berkeley: University California Press.
Bornstein, M. (1975). The influence of visual perception on culture. American Anthropologist, 77, 774-798.
Bousfield, J. (1979). The world seen as a color chart. In R. Ellen and D. Reason (Eds.), Classifications in their social context (pp. 195-220). New York: Academic Press.
Brown, R., and Lenneberg, E. (1954). A study in language and cognition. Journal Abnormal and Social Psychology, 49, 454-462.
Ervin, S. (1961). Semantic shift in bilingualism. American Journal Psychology, 74, 233-241.
Garro, L. C. (1986). Language, memory and focality -- a reexamination. American Anthropologist, 88, 128-138.
Hardin, C. L. (1988). Color for philosophers. Indianapolis: Hackett.
Harkness, S. (1973). Universal aspects of learning color codes: A study in two cultures. Ethos, 1, 175-200.
Heider, E. (1972). Universals in color naming and memory. Journal of Experimental Psychology, 93, 10-20.
Hickerson, N. (1971). Review of basic color terms. International Journal of American Linguistics, 37, 257-270.
Kay, P., and McDaniel, C. (1978). The linguistic significance of the meanings of basic color terms. Language, 54, 610-646.
Kay, P., and Kempton, W. (1984). What is the Sapir-Whorf hypothesis?" American Anthropologist, 86, 65-79.
Lantz, D., and Stefflre, V. (1964). Language and cognition revisited. Journalof Abnormal and Social Psychology, 69, 472- 481.
Lucy, J., and Shweder, R. (1979). Whorf and his critics: Linguistic and nonlinguistic influences on color memory. American Anthropologist, 81, 581-615.
Lumsden, C. (1985). Color categorization: A possible concordance between genes and culture. Proceedings of The National Academy of Science, USA, 82, 5805-5808.
Luria, A.R. (1932). The nature of human conflicts. New York: Liveright.
Ratner, C. (1990). Vygotsky's sociohistorical psychology: A paradigm for today. New York: Plenum.
Rosch, E. (1973). Natural categories. Cognitive Psychology, 4, 328-350.
Schaff, A. (1973). Language and cognition, New York: McGraw-Hill.
Stefflre, V., Vales, V., and Morley, L. (1966). Language and cognition in Yucatan: A cross-cultural replication, Journal of Personality And Social Psychology, 4, 112-115.
Vygotsky, L.S. (1978). Mind in society, the development of higher psychological processes. Cambridge: Harvard University Press.
Whorf, B. (1956). Language, thought, and reality. Cambridge: MIT Press.