Below is the unedited preprint (not a quotable final draft) of:
Saunders, B. A. C. & van Brakel, J. (1997). Are there nontrivial constraints on colour categorization? Behavioral and Brain Sciences 20 (2):167-228.
The final published draft of the target article, commentaries and Author's Response are currently available only in paper.
For information about subscribing or purchasing offprints of the published version, with commentaries and author's response, write to: (North America) or (All other countries).


B.A.C. Saunders & J. van Brakel

Centre for Social and Cultural Anthropology
University of Leuven
Tiensestraat 102
3000 Leuven (Belgium)

Institute of Philosophy
University of Leuven
Kardinaal Mercierplein 2
3000 Leuven (Belgium)


colour, categorization, universals, unique hues, opponent processes, linking propositions


In this target article the following hypotheses are discussed: (1) colour is autonomous: a perceptuo-linguistic and behavioural universal; (2) it is completely described by three independent attributes: hue, brightness and saturation; (3) phenomenologically and psychophysically there are four unique hues: red, green, blue, yellow; (4) the unique hues are underpinned by two opponent psychophysical and/or neuronal channels: red/green, blue/yellow. The relevant literature is reviewed. We conclude: [i] psychophysics and neurophysiology fail to set nontrivial constraints on colour categorization; [ii] linguistic evidence provides no grounds for the universality of basic colour categories; [iii] neither the opponent hues red/green, blue/yellow nor hue, brightness and saturation are intrinsic to a universal concept of colour; (iv) colour is not autonomous.

1. Introduction

The term ekphrasis refers to a device used by Greek and Roman writers to describe works of art that did not necessarily exist. Such accounts were not mere literary exercises but lavished detail on the most desirable, eloquent and moving qualities a work might possess. Not quite wishful thinking nor ideal type, these descriptions tell us much about the values towards which practitioners strove, which ideologies were applauded, and what kind of rhetorical strategies were persuasive within those socio-historical milieus. When linking propositions between theories of language, vision and biology are taken to set non-trivial constraints on colour categorization, we think there may be an analogy with ekphrasis. The following quotations exemplify the sort of links that have been claimed. "[H]ue boundaries, as well as hue foci are given by the biological sensitivity of the organism ... there ... exist natural divisions of the spectrum" (Bornstein 1987, p. 291; 1985); "beyond the shadow of doubt, the senses of basic colour terms are determined by the perceived characteristics of their denotata" (Allen 1986, p. 120); "[t]he epigenetic constraints in color perception are reflected in the verbal color classifications employed in the languages of all cultures thus far studied" (Lumsden and Wilson 1981, p. 45). In short "the basic linguistic categories themselves have been induced by perceptual saliences common to the human race ... biology determines phenomenology and, in consequence, a piece of semantic structure" (Hardin 1988, pp. 168, 156). We therefore examine the following hypotheses:
  1. Colour is autonomous in the two senses of being subserved by distinct neural mechanisms and being readily abstracted, by all human beings, as a property of objects independent of other properties. It is therefore a perceptuolinguistic and behavioural universal, subserved by a distinct physical domain;

  2. Colour is completely described by three independent attributes: hue, saturation, brightness.

  3. There are four unique hues: red, green, blue, yellow;

  4. The unique hues are underpinned by two opponent psychophysical and/or neuronal channels: red/green, blue/yellow.
In reviewing these assumptions the following terminology is used. A unique hue (Urfarbe) cannot be described by hue names other than its own. Unique, unitary, principal, simple, purest, and primitive are synonyms. It is generally accepted that there are two pairs of opponent unique hues (red/green, blue/yellow). Often two unique brightness or lightness terms, black and white, are added. Unique colours should be distinguished from invariant, primary and basic colours. Invariant hues are defined as those not affected by changes in luminance (luminance being the physical correlate of phenomenal brightness/lightness). A set of primaries is a minimal set of coloured lights, mixtures of which can match any other colour perceptually. Basic colours are the referents of Basic Colour Terms which are identified by linguistic and/or psychological criteria (see further section 2.1).

Although not the focus of concern, a central problem in reviewing evidence for the four assumptions is the relation between language and vision. Linguistic evidence has often been the starting point for research on colour. For example a unique hue is defined by international convention as a "[p]erceived hue that cannot be further described by the use of hue names other than its own" (CIE 1987; cf. Wyszecki and Stiles 1982). This definition is derived from Hering's suggestion that: "language has long since singled out red, yellow, green, and blue as the principal colors" (Hering 1964 [1878], p. 48). In a cross-cultural context the point was first made by Ladd-Franklin (1901, p. 400): "the acute tribe of Eskimo examined by Mr Rivers [1902] have discovered for themselves that red, yellow, green and blue (and no other colors) are of a unitary character." Although colour vision scientists tend to stress non-linguistic evidence for the unique hues, throughout the literature unique hues are introduced in linguistic terms (as in the CIE definition). See Hardin (1988, p. 66), Hurvich (1981, pp. 1-11, 53), Lennie & D'Zmura (1988, p. 337), Kuehni (1983, p. 39), Quinn et al. (1988), Pokorny et al. (1991, p. 44). This has methodological ramifications. A distinction must be made between non- linguistic evidence being consistent with the received view and it providing independent evidence for it.

In section 2 the evidence most frequently cited for the universality of basic colours is reviewed. In sections 3 to 6 empirical evidence and justification for the four assumptions are examined from most specific (opponent pairs) to most general (autonomy of colour). Some evidence is relevant to more than one assumption. We conclude there is inadequate support for all four assumptions.

To avoid misunderstanding we emphasize that we do not argue for the following hypotheses:

  1. all researchers publishing on colour naming, categorization or vision accept one or more of the four hypotheses;

  2. there are no mechanisms concerned with colour vision;

  3. in English more than four colour names are needed to describe the character of bright colour viewed in a dark surround ;

  4. relativism and unconstrained plasticity should prevail; the right approach is hermeneutics and/or social constructivism.
In this target article we examine the propositions listed above and conclude there is little evidence for them. They turn out in fact to be quite soft targets. In each subfield, although we make our own evaluations, there are respected local experts who have already made devastating criticisms. What we seem to have here is a case of an apparently coherent story of the type that most scientists would like to believe in, but which is seriously entertained only because people over-charitably assume that the parts in which they are not experts are sound.

2. The evidence for universality

2.1 Linguistic universals

In many disciplines Berlin and Kay's (1969 [1991]) Basic Color Terms is cited as support for the cross-cultural universality of a fixed number of basic colour categories. The exact meaning of basic colour terms (henceforth BCTs) has never been spelled out. Here is how BCTs might be understood in different contexts. The referent could be a set of colour chips (colour-in-the-world), a set of neurons in the brain or functionally defined term in a language-of-thought (colour-in-the-head), words in different languages labelling basic colours (colour semantics), or the experience or sensation associated with basic colour categories (phenomenal colour). As basic colours are claimed to be universals, the exact reference of a BCT is irrelevant because all levels are connected by linking propositions standing in one-to-one correspondence. A BCT names it all.

Berlin and Kay's (1969) goal was to refute relativism. Drawing on an established research tradition (Lenneberg 1953; Brown and Lenneberg 1954; Lenneberg and Roberts 1956),[1] they aimed to show that colour naming was not linguistically constrained. Testing 20 languages experimentally they showed that the BCTs of different languages were congruent when identified by their 'best examples' or 'foci'. They proposed that all languages had words referring to the same two to eleven foci (labelled by BCTs). The claim to universality was based primarily on the clustering of these foci across the 20 languages. These experiments were further amplified by data on seventy-eight languages gathered from dictionaries, ethnographies, and personal communications. A fixed, unilinear evolutionary sequence prompted by environmental triggers explained why all languages did not automatically have 11 BCTs. A typology of seven evolutionary stages was proposed: WHITE/BLACK, RED, GREEN/yellow or YELLOW/green, blue, brown, purple/pink/orange/grey. A slash indicates equal probabilities of evolution; upper case terms refer to composite categories. For example in a language with three BCTs, the composite category RED agglomerates red, yellow, orange, pink and purple. WHITE includes all light hues. BLACK includes blue and green. In a language with five BCTs GREEN covers greens and blues until the BCT for blue evolves.

In general the theory received enthusiastic peer acclaim. Berlin and Kay's results were assimilated in a variety of disciplines. See for example Allen (1986), Bickerton (1981), Billmeyer (1992), Brown (1991), Franzen (1990), Gardner (1987), Hardin (1988), Holenstein (1985), Miller and Johnson-Laird (1976), Pokorny et al. (1991), Sahlins (1976). It prompted the claim that the evolutionary sequence constituted "a primary epigenetic rule serving color category development" linking genes, neurons and the evolutionary development of macro-cognitive behaviour (Lumsden 1985, p. 5808). There are many problems with this theory however.[2]

Berlin and Kay presented their results as an empirical discovery. They proposed linguistic universals defined by the existence of 11 BCTs, the clustering of foci and the evolutionary sequence. Although the theory was thus received in the cognitive sciences and elsewhere, doubts can be raised about the empirical status of this theory. We find we can only understand this work on the assumption that Berlin and Kay had a strong a priori belief that just as "biological foundations of...language...must exist for syntax and phonology" so "basic color lexicons suggest such connections are the realm of semantics" (Berlin and Kay 1969, p. 109f). In support of this contention we offer the following.

Berlin and Kay assumed that the perceptuolinguistic basic colour system is innate, biologically constrained and (semi-) automatic. In the absence of any reason to suspect members of other speech communities having different automatisms, they felt justified in taking the American English colour lexicon as a standard. Experiments were set up in such a way that performance could be transposed into competence through a generating or translation rule. This revealed that at the meta-level, as in American English, there were exactly eleven BCTs. Although it is suggested that BCTs were the result of cross-cultural empirical research, this lexicon was in fact derived from the most popular American- English colour terms in Thorndike's Teacher's Handbook (via Brown and Lenneberg [1954]).

Of the 20 languages for which Berlin and Kay gathered data, 19 were represented by one bilingual speaker only, leaving out of consideration from the start any true cross-cultural significance. (The language studied in the field, Tzeltal, was being studied by Berlin for ethnobotanical classifications.) Methodologically one bilingual speaker per language cannot be considered empirically adequate.

Despite the general acclaim for the theory, most detailed reviews of Berlin and Kay (1969) were critical of their methods of gathering and/or presenting data. There is an appearance of sloppiness which cannot but reduce one's confidence in their conclusions. See Durbin (1972), Hickerson (1971), Newcomer & Faris (1971).[3] For example, apart from many printing errors and mislabelled colours in the mapping diagrams, there were also ethnographic errors and phonemic mistranscriptions. No straightforward information on the informant sample was provided and the choice of languages was not justified. In their use of data from the literature, Berlin and Kay seem to have used whatever came to hand. In reviewing the methodology and nature of the data, Durbin (1972, p. 259) concluded "the reliability and validity of the experiments are zero".

Several commentators pointed out that the restrictions on the BCTs remove most of the world's colour vocabulary. See Panoff-Eliet (1971), Sahlins (1976), Shweder and Bourne (1984, p. 160). Moreover the defining criteria for BCTs were extremely plastic. See Bousfield (1979), Durbin (1972), Hickerson (1971), Kim (1985), Mervis & Roth (1980), Moss (1989a), Snow (1971), Wescott (1970), Zimmer (1982), Zollinger (1984), and Hamp's (1980) response to Branstetter (1977). For example a whole industry developed to determine how many BCTs Russian has for blue (1 or 2?) and purple (0, 1, or 5?). See Corbett and Morgan (1988), Davies et al. (1991), Morgan & Corbett (1989), Moss (1989b), Moss et al. (1990). Using Berlin and Kay's BCT criteria, Dournes (1978) reported that Jrai (Vietnam) has 23 BCTs (though no word translates as 'colour'). Because of the fluidity of the rules, data manipulation is easy, as Hickerson (1971) illustrates. For example where RED is concerned, and no appropriate field gloss is forthcoming, a correction is made, as in Poto: eyeyengo (field gloss: yellow) becomes RED. As GREEN and YELLOW must precede BLUE in evolutionary emergence, so by the logic of the scheme, for Pukapuka, yenga (blue or yellow) must be YELLOW. By the same token, another correction is made for Daza: zede (jaune, bleu, vert) must be GREEN. See Berlin and Kay (1969, pp. 58, 72, 78).

Alternative explanations of tendencies to basic colour categories across languages were not considered. For example Tornay (1978a, p. xxxi) proposed the history of the progressive domination of the West and its values accounts for apparent universality. This seems a plausible suggestion with respect to what is often quoted as Berlin and Kay's most solid result: the clustering of foci. To repeat, this clustering was observed for speakers of 20 languages, 19 of which were represented by one bilingual speaker living in the San Francisco Bay area. According to Rosch Heider (1972a, p. 11) these were foreign students, being therefore literate and westernized.

Finally, although Berlin and/or Kay published various emendations to their theory, in particular to introduce more possible evolutionary sequences (Berlin and Berlin 1975; Kay 1975; Kay et al. 1991), they have never addressed issues raised by their critics. In sections 4.3,

5.3, and 6.2 we present a range of empirical evidence that further

undermines the validity, not only of Berlin and Kay's (1969) theory, but of their whole approach.

2.2 Perceptual universals

It was quickly established that many linguistic communities do not lexicalize eleven BCTs. This offered the opportunity to check whether BCTs were nevertheless in the head awaiting their evolutionary triggering. The work of Rosch (Rosch Heider 1971, 1972a,b, 1973a,b) is the most sustained exploration of this idea. Rosch hypothesized the foci of colour categories to be so perceptually salient as to more quickly draw attention and be more easily remembered. They would thus become more readily lexicalized. Learning colour names would merely be a matter of attaching labels to salient foci. These would be generalized to similar instances and accumulate in language and behaviour. (These foci were "'natural prototypes' (rather like Platonic forms)" [Rosch 1974, p. 114].) It is this work on colour recognition, memory, and learning which led to the development of prototype theory (Rosch 1975, 1978). Our comments apply only to Rosch's work on colour.

Cross-cultural support for these natural prototypes was drawn from Rosch's work with the Dani people of Western New Guinea (Irian Jaya) who apparently used only two colour words mili and mola. The purpose of Rosch's investigation was to discover whether the foci of innate colour categories were learned faster and remembered better than non- focals (Rosch Heider 1972a,b, Rosch Heider and Olivier 1972). She hypothesized that the Dani might undergo a learning process recapitulating the Berlin and Kay evolutionary order (Rosch Heider 1972a, p. 13). Teaching the Dani people to label these innate colour categories, she concluded that categories containing the eleven prototype foci were the most easily remembered and the ease with which they were learned was roughly the same as Berlin and Kay's evolutionary sequence of BCTs.[4] However these experiments raise a number of questions. See Ratner (1989); Saunders (1992; 1995); Saunders and van Brakel (1988).

First Rosch did not inquire into what kind of words mili and mola were, assuming an unproblematic reference. However as this was not confirmed, she interpreted her results as requiring a modification of Berlin and Kay's Stage I. Mili and mola could not be translated as BLACK and WHITE, or DARK and LIGHT or WARM and COOL as each term seemed to refer to both sides of each dichotomy (Rosch Heider 1972b).[5] Consequently the division of the spectrum was neither one of hue nor of lightness. Furthermore, the most popular focus of mola was not white, but divided over two sorts of red. However Rosch raised no further questions about how such words were learned or used. Though linguists and ethnographers have pointed to complexities of the Dani language (van der Stap 1966; K.G. Heider 1979) suggesting that mili and mola are evaluatory words (K.G. Heider 1970, p. 175f), Rosch herself merely noted that the "Dani Ss tended to 'chant' the two names at a constant rate" (Rosch Heider 1972a, p. 16). This suggests there was more to their utterance than labels for the spectrum.[6]

Second she noted that a problem in setting up the learning experiments was that the "Dani would not learn nonsense words" (Rosch 1974, p. 114). The experiments could only be carried out when indigenous words were used. If the theory were correct it would be difficult to explain why the Dani were not willing to learn new labels for salient, natural, universal, ideal-types "reflecting the perceived world structure" (Rosch 1978, p. 29), samples of which were presented repeatedly.

Third she notes that Dani people "were unwilling to designate one of the color chips as the most typical member" of three chips of related hues, one of which was a focal colour (Rosch Heider 1973a, p. 340). This would seem to undermine the universal salience of both focality and prototypicality. Rosch then attempted to show the universality of the colour space and its focal structure by carrying out memory experiments. These experiments did show that the Dani remembered focal colours better than non-focal ones, as did Americans. However when asked to point out a focal colour shown 30 seconds before in an array of 160 colours, Dani people were mistaken 75% of the time, Americans 34%. If humankind has a biological sensitivity to focals, it is difficult to understand how this level of error, or the difference between the Dani and Americans, can be explained.

Fourth what was confirmed in Rosch's experiments (with Dani people and others) was the primacy of focal colours defined by saturation: "the most saturated colors were the best examples of basic colour names both for English speakers and for speakers of the other 10 languages represented" (1972a, p. 13 italics added) and "focal colors are all of the highest saturations available for that hue and value" (Rosch 1972a, p. 19; see also Ratner 1989, p. 366). Given a particular hue category it would seem self-evident that the best example is the most saturated, because 'most saturated' means 'having most colour.' At least one meaning of 'being the best example' of something is having the most of whatever it is that it is. Unless experiments are carried out in which hue foci are identified independently of saturation, no conclusions can be drawn about the universality of foci.[7]

Fifth a related weakness in Rosch's results was the pregiven nature of her colour categories. The most saturated exemplar of a colour was chosen not from a random array, but from shades of that colour fixed by herself. Participants were not asked to choose the best example of their own hue categories but only the best example of a hue category provided by Rosch.

Finally although all Rosch's results are presented within the rhetoric of discovery none escapes her circular reasoning. For if the premise is that colour is both in the head and in the world then however that is experimentally realized will confirm (with more or less noise) that premise.

2.3 Infant and child development

According to one often-quoted study (Bornstein et al. 1976), four-month-old infants naturally partition the spectrum in four categories: red, green, yellow and blue. As there are many problems involved in establishing just what an infant or other language-less creature is responding to (Brown 1990, Teller and Bornstein 1987), caution is needed in concluding that the spectrum is divided into four unequivocal bands. In support of this caution we offer the following circumstantial evidence.

First special explanations have been invoked to explain why children do not easily learn correct colour words (Bornstein 1985). Correct usage begins between 3 and 8 years. Darwin (1877, p. 376) worried about this, observing the slow development of his own children. Bornstein mentions Boynton's 12-year old son (Boynton & Gordon 1965) who "named a number of the same wavelengths blue, yellow, or some combination of blue and yellow" (Bornstein 1985, p. 77) In general the correctness of colour names improves after children go to school. Attempts at colour-to-name training before the age of four seem ineffective. Reviewing a large number of publications Bornstein says that although one would expect that "linguistic identification simply overlays perceptuocognitive organization," paradoxically the supposed perceptuocognitive organization does not facilitate semantic development (Bornstein 1985, p. 74). Moreover the order of acquisition and use of colour terms is reported to be "wholly idiosyncratic" (Bornstein 1985, p. 87). Nonetheless he claims that the terms which go with "the basic fourfold color-name organization of the spectrum" are acquired first. Preuss (1981) however, in a study with 2-year olds at a university day-care, found orange headed the list of both comprehension and production tasks. Orange also came first as the colour with the highest memory accuracy for both Dani and American adults (Rosch Heider 1972a,b).

As further support for categorical perception Bornstein claimed that not only do infants categorize the spectrum as (American) adults do, they also display the same colour preferences: red/blue, green/yellow (Bornstein 1975). But adult colour preference is far from established: Is the preference for pleasantness, arousal, or conspicuousness (Wiegersma and van Loon 1989)? Is it for hue or saturation? Does the Subject go for a strong, serious, or warm colour? Other studies reveal a variety of adult preference orderings both intra- and interculturally. See Davidoff (1991, pp. 115-9), Garth (1922), Helson and Lansford (1970), Martindale and Moore (1988), Wiegersma & van Loon (1989), Zld et al. (1986). Although most studies report either blue or red as heading the preference order, no other study repeats Bornstein's ordering.

3. Opponent colours

3.1 The notion of opponency

Instead of the trichromatic theory of Young and Helmholtz (Helmholtz 1911), Hering (1878) proposed six primitives: red/green, yellow/blue, black/white. He claimed these three experiential opponent pairs reflect human neurophysiology. That is, there is a linking proposition between perceptual opponency and a mechanism at the neurological level (which is reflected in psychophysical models). Hering's theory was revived in psychophysics in a series of publications by Jameson and Hurvich in the 1950's. See Jameson and Hurvich (1955) and Hurvich and Jameson (1955).

Hering however was not the first to propose this theory. The concept of polarity had previously been formulated by Goethe. He claimed certain colours reciprocally evoke each other. In Goethe's colour circle there were three opponent hues: red/green, orange/blue and yellow/purple (Goethe 1976; Sepper 1988). Yet if the unique red/green, blue/yellow opponencies really were the timeless primitives of colour vison, why was so careful an investigator as Goethe unable to introspect them correctly?

Early research on colour opponent cells was carried out by De Valois et al. (1966, 1968) at Berkeley where Berlin, Kay and Rosch were working. Although not in personal contact (MacLaury, personal communication), colour research was clearly on the broader research agenda. Integration of its various strands might be suggested to have occurred with Kay and McDaniel (1978). Though such integration was intimated in Berlin and Kay (1969), Rosch (1971, 1972a,b, 1973a,b) did much to bring macro-colour naming and micro-reductive opponency together. Others who supported this move were Bornstein (1973a,b, 1975); Ratliff (1976); Zollinger (1972, 1976) and Miller and Johnson-Laird (1976).

Using Hering and DeValois as validation, Kay and McDaniel (1978) claimed the distinctive properties of the semantic categories black, white, red, green, yellow, blue correspond precisely to the properties of fuzzy response functions describing the neural mechanisms which underlie colour vision. All other semantic colour categories derive either from fuzzy unions or fuzzy intersections among the six fundamental neural response categories. Intersections produce categories such as purple (red and blue) or brown (yellow and black), while unions give the composite colours such as blue-with- green. To reconcile opponent processes and basic colour terms/categories, Kay and McDaniel insisted that the categorization of the opponent colours developed according to the Berlin and Kay evolutionary rule. Where unions were found, the brain had not yet been prompted to produce the proper primitives of colour vision (evolutionary Stages I - IV); where intersections were found, the brain was well on its way to the full lexicalization of the colour space (Stages V - VII).

Although Mervis and Roth (1980) showed that Kay and McDaniel's fuzzy sets cannot differentiate basic from non-basic colour categories, the neurophysiological underpinnings of Berlin and Kay (1969) by Kay and McDaniel's (1978) appeal to opponent unique colours is often quoted as received wisdom (for example Boster 1986; Lakoff 1987, pp. 26-30). In this way the belief in linking propositions between macro- colour names and the micro-firing of neurons was made explicit.

3.2 Evidence from folk psychology

According to Hering (1878) appeal can be made to tradition and folk psychology for the existence of opponent colours. Hardin (1988, pp. 40-5) quotes several experiments based on qualitative similarity judgements to support the existence of two pairs of opponent unique hues. But the interpretation of these results is not unequivocal (van Brakel 1993). Moreover it is dubious methodology to suppose that the phenomenological intuitions of twentieth century English speakers justify timeless generalizations.

The standard psychophysical explanation for opponency claims that the red/green channel either gives a positive or negative response (see section 3.3). There can thus be no red-green experience. However, the results of Crane and Piantanida (1983) suggest that whether or not this is so is an empirical matter. Teller (1984, p. 1239) shows the in- principle neurophysiological possibility of reddish-green. Moreover according to the colour circle, there are an infinite number of complementary colours with the same opponent property as blue/yellow and red/green.

Occasionally the naturalness of opponency is explained with reference to an intuitive synaesthesic warm/cool dichotomy, for which there is alleged psychological/phenomenal evidence (cf. Kay et al. 1991, Hardin 1988, pp. 129-21). But, although the warm/cool opposition is well-known in folk psychology, support for it is tenuous. Davidoff (1991, pp. 113-5) reviewing recent literature, cautions against associations with particular emotive values as they have no inter- or intra-cultural validity.

There is also empirical folk evidence against natural opponency. Because of a common root in Indo-European meaning 'spring up', implying both 'grow' and 'glow', the same word may in different contexts have associations with red, golden, and green; in Sanskrit for example hari is translated as 'reddish, golden, greenish' (Wood 1902, p. 37f). In medieval Christianity red and green were thought of as interchangeable, as equal in value or as dual components of natural or mystical light. Similarly the Latin and French terms glaucus, ceruleus and bloi could signify either blue or yellow (Gage 1993, p. 90) and there are 'yellow, blue' words in other Indo-European languages (for example plav in old Serbo-Croatian [Herne 1954, p. 73f; Kristol 1978, p. 226]). If the phenomenal self- evidence of opponency were true, there should be stronger historical evidence for it.

In conclusion there appears to be no hard evidence for the phenomenal pervasiveness of colour opponency. However it can be argued that all this is irrelevant as opponency figures in models of colour vision. It is therefore to these models we turn.

3.3 Psychophysical evidence

Generally it is thought there are three types of cones in the retina, each with a different photopigment, and consequently different spectral sensitivity. The three cone types are maximally responsive to short (S), middle (M) and long (L) wavelengths of light. Monochromatic light at the maximum of the absorption curves for S, M, and L cones looks violet, blue-green, and yellow-green respectively. However individual cones are 'colour blind', preserving no details of wavelength. Only from the outputs of different types of cones (responding to the same part of the field of vision) can information be extracted about the activating wavelength.

While the properties of the three types of cones are relatively uncontentious, the evidence for the opponent process theory of colour is less clear. The psychophysical story is that there is one achromatic and two chromatic channels (Hurvich 1981). The first channel processes overall luminance within the broad range of spectral frequencies that excite L and M conesthis is the L+M channel (also called the brightness channel). The second channel processes the relative intensities of long and mid-spectral light, but is insensitive to absolute levels of illuminationthis is the L/M channel (also called the red/green channel). Activity in the third channel is proportional to the difference between the activation of S cones and the combined activation of L and M conesthis is the S/L+M channel (also called the yellow/blue channel). But this simple psychophysical model is at odds with many experimental results.[8]

Discrepancies display themselves in psychophysical experiments primarily in the form of non-linearities which are then attributed to various non-standard interactions between cone types. Nonlinearities are very prominent in the yellow/blue channel and also occur in the red/green and brightness channels. See for example: Ayama & Ikeda (1986), Boynton (1988), Burns et al. (1984), Ejima and Takahashi (1984, 1985, 1986), Elsner et al. (1987), Lee et al. (1989), Lennie & D'Zmura (1988), Shevell and Humanski (1988), Suppes et al. (1990), Mollon & Sharpe (1983). There is no agreement whether there is S cone input to the L+M channel (i.e. whether the S cones contribute to the processing of luminance contrast), and it is unclear whether yellow input comes from the L or M cones or both. See Gouras (1984), Hess et al. (1989), Rabin and Adams (1992), Shevell (1992), Stockman et al. (1991), Stromeyer et al. (1991), and several contributors in Mollon and Sharpe (1983).

Consequently there have been many proposals for adjustments and modifications to the model. For example on a functional level it has been suggested that colour vision proceeds by a mixture of opponent and non-opponent channels (cf. section 6.1) and that different mechanisms process short- and long- wavelength parts of the spectrum that look reddish (Ingling et al. 1978; Paulus and Krger-Paulus 1983; Zrenner 1983, p. 83). As tuning to wavelength may be drastically changed by size and duration, the degree of opponency becomes a variable. Multiple L/M channels (presumably corresponding to multiple cell types) have been proposed. See Finkelstein et al. (1990), Finkelstein and Hood (1984), Hood and Finkelstein (1983, p. 37), Webster and Mollon (1991, p. 238), Zrenner (1985). But this implies that the idea of a fixed spectral sensitivity function has to be given up and with it we have lost a simple explanation of why a light cannot appear red and green at the same time (Shevell and Handte 1983). In addition, it has been suggested that at some level, either there are more axes than three, or the cardinal axes are more mutable than was previously thought (where 'three axes' corresponds to one brightness dimension and two opponent hue channels). See Albright (1991), D'Zmura (1991), Flanagan et al. (1990), Hurlbert (1991), Krauskopf et al. (1982, 1986), Lennie et al. (1990), Webster and Mollon (1991).

One problem in assessing these anomalies is that describing standard summation and differencing of cone outputs requires the empirical curve-fitting with a number of free parameters. The coefficients of the opponent-processes are calculated from absorbance data for the three cone types combined with chromatic-response curves. The latter are determined in experiments in which subjects are presented with spectral lights. First the four unique hues are fixed: the Subject is asked which spectral light is pure blue, etc. This sets the zero level for the L/M and S/L+M channels. The Subject is then asked to mix particular combinations - say a red/yellow light - with so much of an opponent hue - say blue - that a unique hue is obtained (in this case red). This is what is called a cancellation or differencing experiment: the blue cancels the yellow. Similar cancellation experiments are carried out for other combinations. Then the L/M and S/L+M cancellation curves are matched at the point of balanced orange. They are further matched with an achromatic response curve to calculate saturation. Let us pass over the problem of polymorphism and anomalous receptors (Jordan and Mollon 1993, MacNichol et al. 1983, Neitz and Jacobs 1990, Neitz et al. 1993), the complexities of determining the achromatic response curve, the fact that only spectral lights are used, or that a Subject is presented with carefully selected (hence unnatural) viewing conditions (Hurvich 1981, 1985, Werner and Wooten 1979). The crucial question is: Why start with the four unique hues in the first place?

The evidence does not support the postulate of exactly three opponent channels of a specified sort. As argued in the next section, there is no neurophysiological support for a fixed number of opponent neural pathways. Psychophysically distinct channels might well emerge in particular types of well-determined experimental conditions, but these channels do not necessarily exist outside those conditions. Psychophysically speaking there might be many opponent channels while neurophysiologically speaking there are none.

3.4 Neurophysiological evidence

Single-opponent cells among retinal and LGN ganglion cells are considered to display spatial opponency, wavelength opponency or both (Gouras 1984). Such cells have receptive fields consisting of an on-or-off circular centre and an annular surround of the opposite sign, where the centre and surround are connected to different parts of the retina. Such cells do not respond to the absolute level of illumination, but rather to differences in illumination in different regions of their receptive field. If the same cone type subserves centre and surround, the cell is spectrally non-opponent. If centre and surround are connected to different cone types, the cell is also wavelength-opponent. Note that wavelength opponency might be a trivial consequence of the receptive field centre being driven by a single cone and the surround by more than one cone (Lennie and D'Zmura 1988, p. 372).

The suggestion has been made that some ganglion cells are L/M opponent and others are S/L+M opponent. These would thus form neural pathways corresponding to the psychophysical channels introduced above. But again, it is not that simple. As De Valois and Jacobs (1968) already noted there are no neat correspondences between the psychophysical channels and neurophysiological pathways (cf. De Valois and De Valois 1993). It appears that the neutral point of the L/M cells (i.e. the cross-over point from excitatory to inhibitory responses) is not fixed in one spectral region. Even with identical spatial and temporal stimuli and identical adapting conditions, there can be considerable variation in the neutral point among individual cells. Zrenner (1983, 1985) gives a range of 420 to 650 nm for the L/M cross-over point (virtually the whole spectrum). If the results for many L/M cells are averaged, the 'average cell' does give opponent signals when stimulated by an L and M light respectively and there are good grounds for talking about opponent cells. But that is a far cry from there being individual red/green opponent cells. For one thing, whether a cell behaves as if responding differently to a red or green light may depend on characteristics of the stimulus other than being red or green (see section 6.1). For another, it leaves unclear what the purpose of a variety of cells higher up the visual system is, which seem to duplicate the procedure.

If we look at the possibility of a blue/yellow neural pathway the situation is still more confused. As many as nine different ways have been reported for cones and rods to be connected to retinal ganglion cells (Gouras 1984). L/M cells predominate, but a small number of ganglion cells are definitely connected to S cones. They have been called blue/yellow opponent cells, but this means no more than that they are connected to S cones as well as to the L and/or M cones. They respond positively to blue/violet light, but it is not clear how they respond to a yellow signal (Gouras 1984, Zrenner 1983).

The suggestion that there is no neurophysiological evidence for the existence of exactly two pairs of opponent hues is not new. Gouras and Eggers (1984) deny Hering's opponent colour channels in the primate retinogeniculate pathway. D'Zmura (1991, p. 951) says "observers [also] possess chromatic detection mechanisms tuned to intermediate hues such as orange." Teller (1991, p. 530) suggests: "[t]he retinal coding scheme requires further recoding if neurons fully worthy of the name red/green and (particularly) yellow/blue are to emerge. Such neurons have not yet been seen in primate visual systems, and no one knows where or whether they will ever be seen." Similarly Mollon (1992) reviewing Davidoff (1991), claims the latter's appeal to "chromatically opponent neurons that signal redness and greenness or blueness and yellowness" is pseudo-physiology because "the neurons he requires to substantiate his view are not those that have so far been found electrophysiologically in the visual pathway."

None of this denies the existence of a variety of opponent and other types of cells. That however is not the issue. The issue is that the alleged evidence for exactly two pairs of opponent hues is not well-grounded.

We recognize that neurophysiological data are inherently error-prone due to fluctuations in the organism's level of arousal, alertness, anaesthesia, etc., and that variability in the neutral point of a putative chromatic opponent channel may arise either because of measurement error or because two opponent channels simply do not exist. Nonetheless the neurophysiological evidence for red/green and blue/yellow opponent pathways is reviewed because the cautions and hesitancies of the neurophysiologist are frequently lost when adjacent disciplines adopt their findings. Nothing we have said diminishes the importance of neurophysiological research or the functional characterization of the visual tract.

4. Four primitive hues?

4.1 Intuitive phenomenal categorization and its dependence on surrounding culture

Textbooks state that Newton established the fact of seven colours in the natural spectrum or rainbow: red, orange, yellow, green, blue, indigo, violet. How reliable is this scientific fact? Newton himself did not see seven colours: "I held the Paper so that the Spectrum might fall upon this delineated figure ... whilst an Assistant, whose Eyes for distinguishing Colours were more critical than mine, ... noted the Confines of the Colours, that is ... of the red ... orange ... yellow ... green ... blue ... indigo ...and ... violet" (Newton 1952, p. 126; see also Gage 1993, ch. 9; Topper 1990). As Campbell (1983) notes: "If Newton's assistant had not been so eager to please his master our current textbooks would be different." Newton's assistant saw seven colours probably because the harmonic series of Pythagoras still dominated mathematical thinking. Newton himself (and many before and after) studied the analogy between the colour spectrum and musical chords (Gage 1993, ch. 13). Not all historians of science agree the analogy fixed the number of colours in the spectrum (Topper 1990). But one thing is clear: the reported number of colours in the spectrum was not determined by what Newton or anyone else saw.

What other evidence from direct observation do we have? Thomas Young at one time divided the spectrum into three primaries: red, yellow, blue. In 1802 Wollaston reported to the Royal Society (Sherman 1981): "the colours into which a beam of white light is separable by refraction, appear to me neither 7 ... , nor reducible ... to 3 ... but ... [to] 4", viz. red, yellowish-green, blue, and violet. Young at once rejoined that red, green, and violet were the primitives. In 1822 Brewster claimed to have shown conclusively that the spectrum contains yellow too. But his experiments were unrepeatable (Sherman 1981; cf. Mach 1919, p. 53). According to Helmholtz a spectrum short enough to be viewed in its entirety consists of four colours (red, green, blue, violet), but added (1911, p. 117): "there are no real boundaries between the colours of the spectrum. These divisions are more or less capricious and largely the result of a mere love of calling things by name."

There is still disagreement about how many colours can be seen in the spectrum. The majority opinion seems to be that there are five: red, yellow, green, blue, violet. This is also the hue family of the Munsell system. However Biernson (1972) claims that orange must be added, and Campbell (1983) doubts whether yellow can be seen between green and red (see also Duck 1987 and Kidder 1989). As to the colours of the rainbow Gage (1993, p. 93) suggests "the very delicacy of the transitions of the bow ... makes it extremely hard to number and name the colours. This has made the phenomenon especially apt for interpretation according to any number of prevailing schemata."

But which colours one sees in the spectrum/rainbow is irrelevant to the question of how many unique hues there are. Unique hues would cause lines, not graded bands in the spectrum and no reason exists why people should use only unique hues to report what they see. The purpose of the example is to illustrate the difficulty of separating what is seen from theoretical presuppositions and prejudices. It would not be too far-fetched to suggest that the observation of seven colours in the spectrum was fixed by a prevailing number-7-cosmology, comparable to cosmologies elsewhere.[9] Probably many will agree that the suggestion of Paritsis and Stewart that (1983, p. 109): "at the cortical level, colours are classified into seven classes of cells" is nonsense. But if in the twentieth century some scientific encylopdiae illustrate a spectrum by the rhetorical seven bands of colour, then diagrams in current textbooks depicting the opponent pairs as red/green and yellow/blue must also be considered rhetorical.

4.2 Psychophysical evidence for four unique hues

A considerable range of wavelengths are identified as unique green, blue or yellow (whereas unique red, not being in the spectrum, raises questions about the operational meaning of unique hue): unique blue is found between 462-496 nm, unique green between 488-545 nm, and unique yellow between 566-588 nm (Dimmick and Hubbard 1939, Schefrin and Werner 1990). These points cover the humanly visible spectral range apart from 21 nm of yellow-green. Moreover since Hering (1878, p. 58), uncertainty abounds as to whether or not brown is unique (Bartleson 1976, Fuld et al. 1983, Quinn et al. 1988). It is true that if one averages the data into a Standard Observer the unique hues become more constantthat after all is what averaging does. But the issue is not whether English speakers roughly agree on what green is or whether brown is, or is not, unique. The issue is whether asking people to point out the unique hues reveals anything more than their command of English (defined by the average speaker). Therefore quoting Sternheim and Boynton (1966) or others as having shown that English speakers can describe all spectral colours with the four unique-hue names misses the point. It is also unclear how to deal with people who do not agree with the average observer. According to the CIE definition, green is unique if it cannot be further described by hue names other than its own. But in the Euro-American tradition many people are trained to see green in terms of blue and yellow. They will say green should not count as an Urfarbe. If green is an Urfarbe 'intuitively' then all arguments against the uniqueness of purple, orange, aqua, and so on, fall away.

Although there is a considerable amount of non-linguistic evidence for trichromaticity and some sort of opponent processing, it is less clear what exactly the non-linguistic support is for four unique hues. There are many conceptual unclarities here which would justify a separate review. First most measurement techniques rely on some sort of threshold detection, not representative of ordinary colour vision at suprathreshold levels. Second many experiments are carried out with spectral lights (as distinct from coloured surfaces), which are not representative of the chromatic world humans live in. (Neither of course are colour charts, as Klee [1961] amongst many others has shown.) Third it is a matter of controversy whether the chromatically opponent channels suggested by additivity experiments are the same as account for the results of cancellation experiments. Finally in introductions to the subject the Bezold-Brcke phenomenon of invariant hues is usually quoted recycling data from Purdy's 1929 dissertation (Purdy 1931, 1937). These invariant hues are then assimilated to the unique hues (Hurvich 1971, p. 73). But Purdy (1931) himself noted that the invariant hues of the Bezold-Brcke effect are not the same as the unique hues. Cf. Pokorny et al. (1991, p. 45), Ejima and Takahashi (1984, 1985), Suppes et al. (1990), Paulus and Krger-Paulus (1983), Vos (1986), Zrenner (1985). It seems therefore premature to conclude that psychophysical evidence for four well defined unique hues or simple colours has been established.

4.3 Problems identifying the analytic mind

If four unique hues were a universal human perceptual grounding, cross-cultural research would confirm it. But empirical evidence for a fixed number of primitives (whether four or any other number) is utterly evanescent. Of the Munsell colour chips commonly used in cross-cultural experiments, 60-80% often remain unnamed (see Berlin et al. 1991). When presented with chips people get confused and give inconsistent answers. Tougher- minded subjects find the naming and/or categorization tasks absurd, not self-evident. For example Berlin and Berlin (1975, p. 85n.50) recount their difficulties with monolingual Peruvian Aguaruna informants. Many would simply stare at the array. Others who could bring themselves to place the pen to the plastic would begin drawing individual black circles around chip after chip often moving along some level of brightness, completely ignoring hue. Several attempted to provide a different name for each perceptually different chip, employing terms which later proved to be the names of trees, plant dyes, and parrot feathers. One informant, when asked to show where all the red chips were, took the pen and very carefully circled the entire board. Similar examples can be found in Berlin et al (1991), Bulmer (1968), Conklin (1955), Friedl (1978), Kuschel and Monberg (1974), Luria (1976), Rivers (1903), Saunders and van Brakel (1995).

The common solution to such problems is to conclude that lack of abstract colour categories is the result of evolutionary backwardness (see comments of field workers in Berlin et al. 1991). For example, in the nineteenth century, drawing on the absence of a word for blue in Ancient Greek, Hebrew, Aramaic, and Akkadian (Brenner 1982), it was argued that colour vocabularies must have had an evolutionarily determined physiological basis (Gladstone 1858, 1877, Geiger 1871, 1872, Magnus 1877). Although refuted (Allen 1879, Kirchhoff 1879, Krause 1877, Magnus 1880, Virchow 1878, Woodworth 1905, 1910, Rivers 1901, 1905, Titchener 1916), Berlin and Kay (1969) nevertheless revived the idea in terms of evolutionary stages (though insisting on universal physiology). Kay (1977, p. 30) for example, stated the broad goal: "The direction of linguistic evolution is toward the precise and explicit speech of the analytic philosopher, the scientist and the bureaucrat." Using this analytic outlook, the highest stage of chromatic evolution is perhaps to divide the Munsell chart in four vertical bands (corresponding to the four unique hues). (Cf. Lumsden's [1985] presentation of an epigenetic rule for colour.) Some speakers of Mixtec (MacLaury and Galloway 1988) and Shuswap (MacLaury 1987) divide the Munsell chart into three vertical hue bands and thus approach this level of analyticity. But what about horizontal rather than vertical bands? MacLaury (1992, p. 151) reports data from speakers who divide the chart into three horizontal bands. Are vertical bands more analytic than horizontal?

Why is calling the sky 'blue' more analytic than calling it 'light blue' (for example celeste in Mesoamerican languages; see Bolton 1978, Harkness 1973, MacLaury 1986, 1991, Mathiot 1979). Why is it more analytic than calling it 'bright' as in Mursi (Turton 1978, p. 366), 'clear, serene' as in Sanskrit (Hopkins 1883; cf. Wood 1902), 'whitish' as in Batak (Magnus 1880), verde 'green' as in Tlapenec (Dehouve 1978), or nothing at all (some Italian dialects: Kristol 1980, p. 142)? Many languages appear to differentiate between light and dark blue.[10] This converges with observations of western painters (e.g. Cezanne) who felt that chromatic blue has a double nature, related to light and dark. Dark blue and black are also frequently considered a unified entity. See Almquist (1883), Berlin et al. (1991), Forge (1970, p. 283), Gage (1978, p. 110), Gatschet (1879), Hopkins (1883), Jochelsen (1908), Rivers (1901), Tornay (1978b), Wierzbicka (1990), Wood (1902). Within the Berlin and Kay tradition however this is considered an earlier stage of evolution. But what makes unifying dark blue and black, or separating light blue and dark blue an inherently primitive habit? That is, is hue discrimination more analytic than brightness/lightness discrimination? The burden of proof rests with those who wish to argue the point.

In many languages blues and greens are mapped together under one BCT. See Berlin et al. (1991), Bornstein (1973a,b, 1975), MacLaury (1986). Less common is the problem of a yellow-with-green category. This was submerged in Berlin and Kay (1969), though long known. Of the literature to which they refer, Ray (1953), Rivers (1901) and Magnus (1880) mention it.[11] The yellow-with-green category is common in North America, in particular on the Northwest Pacific coast. See Gatschet (1879), Holmer (1954/5), Kinkade (1988), Proulx (1988). But yellow-with-green terms in Indo-European languages have also been discussed since the nineteenth century. See Schulze (1910), Schwentner (1915, p. 68), Weise (1878, p. 281). Proto-Indo-European *ghel- yellow, green, grey, blue, from which yellow is derived, has cognates for green in daughter languages, for example Lithuanian zelvas (Pokorny 1959, p. 429-30). Some etymologies provide the gloss 'yellow, green' for older Indo-European languages: for example, Sanskrit harita, Greek chloros and old-Slavic zelenu. See Pokorny (1959, p. 429), Filliozat (1957, p. 305f), Hopkins (1883, p. 175), Schulze (1910, p. 800), Irwin (1974, p. 77), Rowe (1974, p. 341), Schwentner (1915, p. 68).

Current linguistic interest in yellow-with-green was stirred by MacLaury (1987) who asserted that its presence in Shuswap "contradicts present physiological knowledge." Kwakw'ala (spoken on Vancouver Island and the adjacent mainland) also has one word for yellow-with-green: lhenxa (Saunders 1992). On the grounds of theory one would expect the anomaly to evaporate if speakers were reminded that yellow and green are two different unique colours, two of the four built-in opponent hues. But though most current speakers of Kwakw'ala are bilingual and know perfectly well the difference in English between yellow and green, they stick to lhenxa. The yellow-with-green category is particularly intransigent compared to say, ayendzis 'orange' a recognized loan word from Chinook jargon, and pinkstu an obvious neologism (Saunders 1992). The counter- suggestion is to say that lhenxa supervenes on the union of two Urfarben at a deeper level. But then why do the innate categories always coincide with twentieth century American English? One reason might be that linguistic research tends to confirm the cross-cultural validity of Euro-American categories by imposing them on non-written languages when first inscribed. Boas for example, invariably glossed lhenxa as 'green' (Boas 1892, 1931, 1934, Boas and Hunt 1905), as did Curtis (1915). But Dawson (1887), Grubb (1977) and Lincoln and Rath (1980) provide the more nuanced yellow-with-green gloss (Saunders and van Brakel 1996). Similar problems arise with the history of European languages. But recognition of the contingency and socio-history of current colour categories in English fails to be appreciated in the colour science literature.

In conclusion, before deciding there is scientific evidence for four unique hues (or any other number of unique or basic colours), it is necessary to be sure one is not simply fitting one's data to modern English.

5. Hue, brightness, and saturation

5.1 Attributes of colour

It is generally assumed that colour has three independent psychological dimensions: hue, brightness, and saturation (Munsell's hue, value and chroma). We take this assumption to be an empirical statement about the properties of colour. If colour is defined as hue, brightness and saturation, our comments dispute the autonomy of colour.

Narrowly construed brightness tends to be used both for lights and for surface colours viewed in aperture mode. The corresponding term for the appearance of surface colours is lightness. In ordinary language, words like shade, tint and colour tend to be used for hue. Alternative adjectival forms for lightness are light, bright, pale, colourless, shaded, whitish; similarly for saturation: deep, dense, intense, vivid, pure, lustrous, faded, lacklustre, permeated, infused, full, vibrant, dull. But it is notoriously difficult to separate saturation and lightness, and even quite technical words tend to overlap in these dimensions; for example faded, bright and lacklustre might be considered both lightness and saturation terms. In addition, the status of the concept of saturation when applied to lights is unclear, particularly in the case of negative contrast.

There are various standardized systems of colour classification and measurement (Derefeldt 1991); for example Munsell, NCS, DIN, OSA/UCS, and CIE. The Swedish NCS is most in line with Hering's ideas of opponency, unique hues, and a natural order. Munsell however is the most widely used system in the English-speaking world. The chips representing the colour space eliminate all aspects of the location of objects, their surfaces and their relations in the worldthe difference between related and unrelated colours. They remove such features as duration, size, texture, glossiness, lustre, fluctuation, flicker, sparkle, glitter, shape, insistence, pronouncedness, brilliance, fluorescence, glow, iridescence, colourfulness, nuance, background or surround colour - all of which have been proposed as specific attributes of particular coloured surfaces or volumes. See Beck (1972), Evans (1974), Hunt (1977), Pokorny et al. (1991), Pointer (1980), Gibson (1979, p. 31). Preliminary research on colour phenomenology from satellites suggests there may yet be more differences (Vasyutin and Tishchenko 1989). As Pokorny et al. (1991, p. 45) say: "No comprehensive theory of colour appearance can be based only on the properties (hue, saturation and brightness) of unrelated colours." Further, there is much unclarity about the (inter)relation of surface colours (as modelled by the Munsell system) and colours in other modes of appearance (Beck 1972; Katz 1935; Nickerson and Newhall 1943).

What experimental support is there for the assumption that colour in daily life consists of three psychologically salient components: hue, brightness and saturation? Burns and Shepp (1988) argue there are serious problems about a three dimensional spatial metrics as the proper psychological dimension of colour vision. They review evidence that physical attributes of colour do not independently affect the psychological dimensions. Chang and Carroll (1980) suggest that the psychological colour space has seven ( 1) dimensions. Work on the OSA system has shown it impossible to represent colour in Euclidean space (Man and MacAdam 1989, Nickerson 1981). None of the existing systems of colour classification achieves the goal of uniform perceptual intervals between any two adjacent colours (Derefeldt 1991, Indow 1988). Therefore hue, brightness and saturation notwithstanding their usefulness for particular technical purposes, can only be claimed to describe the Munsell and similarly artificial colour spaces.

5.2 Interdependence of hue, brightness, and saturation

Because of the Bezold-Brcke and Abney effects there seems little doubt that hue, brightness and saturation are interdependent. This undermines the appeal to linking propositions connecting the phenomenal and physical description of colour. There is no consensus on the relation between luminance, brightness, lightness and/or whiteness. See Beck (1972), Boynton (1988), Heggelund (1992), Pokorny et al. (1991), Whittle (1994). Definitions of non-hue attributes differ in different systems (Derefeldt 1991). Two stimuli matching for CIE-defined luminance may differ radically in brightness (Ikeda and Nakano 1986; Wyszecki and Stiles 1982, pp. 411-20). Boynton (1988, p. 82) quotes a report of Ware and Cowan in which 63 studies of the brightness-luminance phenomenon are fitted to a fourth-order polynomial to calculate the brightness/luminance ratio for any chromaticity. The brightness/luminance discrepancy depends on both saturation and dominant wavelength (Uchikawa et al. 1984). In addition two separate neural mechanisms for brightness induction have been proposed (Shevell et al. 1992). Similarly purity (the physical correlate of psychological saturation) fails to dovetail with saturation in any clear way (Hunt 1977, Pridmore 1990). A saturated colour is perceived as brighter than a desaturated one when the two are equated in luminance (Yaguchi and Ikeda 1983).

Lockhead (1992) reviews an impressive range of studies showing how brightness judgements are subject to successive, simultaneous and other contextual variables, raising doubts about a psychophysical law for brightness. According to Davidoff (1992, 1991) perceiving hue, lightness and saturation requires elaborate training to separate their interactions with any success: "for the naive [subjects] lightness and saturation are not independent" (Davidoff 1974, p. 79; cf. Burns & Shepp 1988). Being integral (that is unified as a whole in vision), it is unsurprising that hue and lightness, hue and saturation, and lightness and saturation cannot easily be separated (Pokorny et al. 1991, Saunders 1992).

Further problems arise with the psychological difference between black/white, dark/light, and dull/bright. For example chromatic surface colours may appear equally greyish, although their lightnesses (or brightnesses) differ (Hering 1878, Katz 1935, Beck 1972, Derefeldt 1991). Heggelund (1992, 1993) reminds us that Hering had already noted the bi-dimensionality of grey colours. No consensus exists about whether the ordinary words 'black' and 'white' refer to colour, brightness, or something else. There is difficulty in imagining a context-independent scale for brightness or whiteness comparable to say, length (Smith and Shera 1992). It has been suggested that brain grey, Eigengrau, is "the intrinsic basal sensation associated with the equilibrium condition of the entire visual system" (Hurvich 1985, p. 66). This would suggest that the natural zero for brightness is not black or absolute darkness, but brain grey. Psychophysically the status of (induced) blackness is unclear (Lee at al 1989, Volbrecht et al. 1990). A white surface may be defined as a surface that equally reflects at every wavelength across the spectrum. But that offers no help for the psychological differences here under review and leaves out for example the luminous-whitish properties seen on white light sources (Heggelund 1993).

5.3 The primacy of hue

At the cross-cultural level things are no better. At various times a dominance of brightness (as distinct from hue) classification in tropical areas has been proposed (Simon 1951, Van Wijk 1959). It has been suggested that people in those areas have different cone pigments (Rivers 1901, Bornstein 1973a). But this hypothesis has no physiological basis. Recently MacLaury (1992) has revived brightness classification as an early evolutionary stage, though he disclaims physiological implications.

People in the Euro-American world are trained to distinguish hue. Cross-cultural research reveals the distinction to be contingent - hue, we must conclude, not being naturally salient. This is supported by numerous translation problems. (Note however that current translational choice is limited to hue, brightness and saturation, these being the only parameters recognized under the Munsell scheme.) For example, hue/brightness problems arise in translating Sanskrit (Hopkins 1883) and Arabic (Fischer 1965, Gtje 1967). Cases such as Sudanese and Arabic green or blue skin (Bender 1983) must be considered metaphors. Homeric Greek, discussed in numerous publications, presents intractable problems for the hue/brightness distinction (Hickerson 1983, Irwin 1974, Maxwell-Stuart 1981). Skard (1946) gives more than 50 sources discussing these problems in pre-1940 literature, and Maxwell-Stuart (1981) needs 200 pages to discuss the uses of glaukos. Ancient Greek colour terms are problematic because they have more to do with brilliance and lustre than with hue. Sensitivity to, and the importance of gloss and glitter descriptions in the Homeric poems should alert any reader to dimensions of temporality and movement as distinct from stasis in the use of these terms (and the same would seem to apply to Sanskrit: Hopkins 1883). This goes much further than saying the ancient Greeks were more interested in brightness than hue.

Hence the inherent independence and/or salience of hue (or brightness) does not seem a well-supported conclusion in the cross-cultural data. This brings us to the more general issue of the autonomy of the colour domain.

6. The autonomy of colour

6.1 Cross-modal neurophysiology

In this section we briefly review neurophysiological evidence. Our conclusion is that it fails to support the autonomy of colour. As noted in section 3.4 most ganglion cells (in the retina and the Parvocellular layers of the LGN) are connected to both L and M cones. These L/M-cells cannot distinguish between a large coloured and a small white spot. Such cells can switch instantly to nonopponent processing e.g. if an object is presented only briefly. They show a surprising functional plasticity: in a way, they adapt to the purposes of looking. It has been realized that if in the simple opponent-process model both summing and differencing signals (both brightness and chromaticity) must go through a single neural pathway then "[t]he absence of a distinct achromatic' pathway is the most troublesome physiological finding" (Lennie and D'Zmura 1988, p. 372). This is especially so if one expects chromaticity and brightness to be neatly separated and in one-to-one correspondence at the psychophysical and neurophysiological level. See also Derrington et al. (1984), Estvez and Dijkhuis (1983), Gouras (1984), Gouras and Eggers (1984), Finkelstein (1988), Hood and Finkelstein (1983), Tansley et al. (1983), Zrenner (1983).

Recently more emphasis has been placed on two distinct visual pathways from retina through LGN to visual cortex. See Hubel and Livingstone (1990), Lennie and D'Zmura (1988), Livingstone and Hubel (1984, 1988), Mollon (1989), Zeki (1985). According to Shapley (1990), achromatic vision is identified with the M pathway (via the Magnocellular LGN) and chromatic vision with the P pathway (via the Parvocellular LGN). There are however, a number of reasons why this is implausible (see in particular Mollon 1989). As the PLGN is much bigger than the MLGN, it is doubtful that it would only be implicated in wavelength detection; evidence exists that it also responds to motion and acuity (texture, fine pattern and fine stereoscopy). See Albright (1991), Boynton (1988, p. 81), Gouras (1991), Hubel and Livingstone (1990, 1991), Ingling and Grigsby (1990, 1991), Ingling and Martinez (1983), Kooi and De Valois (1992), Krauskopf and Farell (1990), Merrigan (1989), Mollon (1989, 1990), Ohmura (1988), Schiller et al. (1990). Even Shapley (1990), a strong advocate of a version of chromatic/achromatic dualism and parallelism, admits that "(c)ooperative and suppressive interactions ... demonstrate that these pathways may start in parallel but ... converge". Hubel & Livingstone, often quoted for their claim that colour and form are processed separately in the visual cortex, do not deny that "colour information" may be used to "detect borders" and that "some opportunities for cross-talk exist" between the P- and M-pathways (Hubel and Livingstone 1990, p. 2223). In addition too little is known about the significance of the massive back-projections from all areas of the cortex to the thalamic nuclei. See Barlow (1990), DeYoe and Van Essen (1988), Steriade and Dechenes (1985), Zeki and Shipp (1988).

Psychophysical evidence suggests that space and wavelength are intricately linked, an issue related to the existence of non-linearities in all three psychophysical channels (see section 3.3). There seems to be consensus that the chromatic channels contribute to brightness, at least above threshold levels (Ingling and Martinez 1983; Cole et al. 1990; Yaguchi and Ikeda 1983). Inputs originating from wavelength differences may go to the "motion channels" (cf. discussion on M- and P-pathways above), chromatic channels also displaying "orientation sensitivity". See Bradley et al. (1988), Dobkins and Albright (1993), Flanagan et al. (1990), Javadna and Ruddock (1988), Shapley (1990). In addition there are rod-cone interactions (Ingling 1977; Montag and Boynton 1987; Zrenner 1983).

There is strong evidence that between retina and cortex, processing of wavelength is intricately mixed with luminosity, form, texture, movement response and other environmental change. It is sometimes suggested that the value of colour vision is to pick up survival information from the environment. But why pick up colour? Answers could be: because colour contributes to object recognition; or: it contributes to identifying edible fruits; and so on. However, to arrive at the conclusion that the fruit is ripe, or over there, there is no unique need for colour (cf. Akins and Lamping 1992). For that we would need an antecedent argument for pre-ordained cosmic harmony (cf. Shepard 1991; Saunders 1995a)

At the level of individual cells there is no evidence for anything that might be called, even metaphorically, a "colour-coded" cell. Of the many cells that seem to contribute to colour vision, double-opponent cells in area V1 of the visual cortex are most sensitive to simultaneous presentation of stimuli of two different wavelengths, one covering the centre of the cell's visual field, the other illuminating the surround. What are called green off centre cells give a maximum response to a red spot surrounded by a green annulus. Such a cell is not influenced by a large homogeneous (chromatic or achromatic) light spot covering the entire receptive field. So double-opponent cells are claimed to be sensitive to wavelength differences only. But it would be premature to conclude that we have now found the locus of "colour-coded" cells.

First there is a "bewildering variety of colour-coded cell types" (Livingstone and Hubel 1984, p. 348; cf. Lennie and D'Zmura 1988) and little is known about the organization of the receptive field of double-opponent cells, or about how they connect to cells in the LGN (Billock 1991; Zeki 1985). Experimental evidence is as disputable as it is difficult to isolate centre from surround (Shapley 1990, p. 647). This may explain why there is no agreement on the spatial properties of colour-opponent cells (Lennie and D'Zmura 1988, p. 376) or on the number of such cells in different areas of the visual cortex (Michael 1985; Zeki 1985). In general wavelength sensitivity varies with the particular conditions under which the measurement is made.

Second not only has the existence of double-opponent cells in area V1 been disputed (Lennie and D'Zmura 1988; Ts'o and Gilbert 1988), but there are also single-opponent cells, cells with single spectral sensitivity curves and cells which only respond to wavelength in conjunction with lines or edges of particular orientations. Some researchers report columns or blobs in the visual cortex which respond solely to colour stimuli and not to white light; others dispute their existence. See Boynton (1988, p. 92), Livingston and Hubel (1984, 1988), Michael (1983, 1988), Tanaka et al. (1983), Zeki (1985).

Third Zeki (1984, 1985) doubts that the double-opponent cells in V1 "code for colour." He believes double-opponent cells merely respond to a predominance of a signal from one cone type. Only in area V4 would we find the first truly "colour-coded cells" (i.e. cells that respond not merely to wavelength, but to context-effects and such like). In this respect he has suggested that his account converges with Land's retinex theory (Land 1986). But Zeki's conclusion that neurons in V4 are chromatically more selective than those at lower levels is disputed by many. See for critical discussion of Zeki's work: Andersen et al. (1983), Boyd (1988, p. 93), Desimone et al. (1985), Gouras (1991, p. 189), Krger and Fischer (1983), Lennie and D'Zmura (1988, p. 390), Schein et al. (1982). Moreover, Livingstone and Hubel (1988, p. 744) refer to a "dozen or so areas north of the striate cortex" of which the (colour) vision properties still need to be investigated (cf. Lennie et al. 1990).

By what criteria then should a cell be qualified as "colour-coded"? This already poses a problem on a purely technical level (Krger and Fischer 1983, p. 295; Lennie and D'Zmura 1988, p. 382). For example it is a common fallacy to say that a cell has a special coding role for a stimulus which makes it fire faster (see Estvez and Dijkhuis 1983, Hardin 1988, p. 56, Martin 1988, p. 383, Teller 1984), as we cannot exclude the possibility that the cell might respond similarly or even better to a stimulus for which it was not tested. Furthermore neurophysiological support for theories of human colour vision has drawn primarily on experiments with monkeys. These monkeys tend to be anaesthetized, meaning that in addition to feeling no pain they are seeing nothing at all (Boynton 1988, p. 93, cf. Haenny et al. 1988, p. 245). This means that "the extra-retinal inputs to the prelunate cortex ...would be reduced or eliminated and the whole state of control of the excitability of this cortical area might be different, leaving the cells with their pure afferent sensory inputs that originate from the retina" (Krger and Fischer 1983, p. 293). This touches on the more fundamental problem of how to make a principled division between active colour see-ers and passive wavelength responders.

6.2 The disintegration of the colour concept

Examples from non-western languages where hue plays a less prominent role than it does in English have already been given. The imperative of a clear-cut colour category disintegrates further when aspects of hue are referred to in ways that are vaguely chromatically individuated in some contexts. Examples in English are gold, silver, rosy, and blond. Similarly there are many languages which seem to refer mainly or solely to colour change (cf. Zaidi and Halevy 1993), or to indicate something of the interrelationship of growth and maturity, or to the interaction of colour with evaluation.

Consider the languages of African pastoral cultures, whose colour vocabularies are claimed to have a large number of BCTs (see Berlin and Kay 1969: for the Bedauye, p. 83; Masai, p. 85; Bari, p. 87; Dinka, p. 93; Nandi, p. 98). Magnus (1880) had already noted that Xhosa people distinguished twenty six cattle colours, though there were no words for blue and green. Many subsequent studies have addressed the difficulty of separating colour and cattle idiom. See Evans-Pritchard on Ngok Dinka (1933-5) and Nuer (1940), Lienhardt (1970) on Dinka. Fukui (1979) on Bodi, Tornay (1973, 1978c) on Nyangatom, Turton (1980) on Mursi. Obvious questions are whether a colour idiom is applied to cattle or a cattle idiom to colour, and whether it makes sense to enforce a distinction between colour and pattern vocabulary. Similar complexities arise where horses are important. See Radloff (1871) for Kirgiz, Laude-Cirtautas (1961) for Turkish, Hamayon (1978) for Mongol, Centlivres-Demont and Centlivres (1978) for Uzbek, Hess (1920) for Bedouin Arabic.

Second common confusions arise when it is unclear whether a word is about colour appearance or aspects of growth. For example in Lokono (Arawak) there is imoroto unripe, immature, green, pale yellow, koreto ripe, mature, red, orange, deep yellow, and bunaroto overripe, overdone, brown, buff, tan, purple. Attempts to conclude that either colour is metaphorically extended to growth or vice versa, fail. Similar problems arise in many other languages. See Hickerson (1953, 1988) for Lokono, Conklin (1955) for Hanuno, Geddes (1946) for Fijian (Bauan), Juillerat (1978) for Amanab, Bulmer (1968) for Karam.

Third consider Bellona colour words. Kuschel and Monberg (1974) began their research (based on dictionary data) with the assumption that Bellona had seven BCTs. But at the time of investigation the language was in a process of rapid change due to globalizing forces. Carrying out fieldwork in 1971/2, it was nonetheless possible to draw on pre- contact memories. Though for the younger generation the BCTs had the approximate meanings of white, black, red, yellow, blue, green, and brown/violet, for the older generation the situation was different. None of these words (nor the great variety of other Bellona colour words) could be used independently of context. They were not like secondary colour words (lemon yellow, poppy red, etc.) which can be used to name a colour on any object, but were more like words such as blond (applicable to hair, wood or beer). In Bellona, such words were used for a range of properties, objects, situations or events (of varying degrees of abstractness or particularity) which to Western understanding appear disparate and unconnected. For example, one of the many words for blackness or darkness lalangi could be used of a dark night, black tattoos, flying foxes and Melanesians from the West Solomons, but not of hair, whales, or fish.

Often Bellona words did not refer to colour as the property of an object, but to a process of change, such words being also frequently evaluative (like Dani mili and mola).[12] It is therefore unsurprising that most terms could not be mapped on the Munsell colour chart. The old uses have now disappeared being replaced by Western ones. In this Bellona perfectly illustrates what Tornay (1978a, p. xxxi) described as the mistake of confusing the cultural evolution of mankind with the history of the progressive domination of Western practices.

7. Conclusion

The results of the preceding sections can be summarized as follows:
  1. Neither structural nor functional opponency is denied. But neither neurophysiology nor psychophysics supports exactly two pairs of opponent hues or three pairs of opponent colours.

  2. The history of western art and science and a cursory cross-cultural glance reveal that to rely on folk perception to establish four primitive hues, eleven basic colour categories, or any other perceptual categorization of colour held to be universal (due to salience, innateness, or whatever) is an unreliable procedure. There is no convincing evidence that particular colour primitives exist at any kind of pre-linguistic, phenomenal or biological level.

  3. Psychologically there is nothing natural about either the combination or separation of hue, brightness and saturation. There are no linking propositions between say, luminance and brightness.

  4. There is little if any concrete evidence for a one-to-one correspondence between physiological pathways and psychophysical channels. Different functional channels may be (partly) embodied in the same set (pattern) of anatomic cells and different contexts may require different sets of functional channels. But there is no convincing neurophysiological evidence for an autonomous colour pathway.
  5. There is no strong evidence for specific mechanisms corresponding to Hering's four unique hues.
We conclude that the evidence supporting the four hypotheses reviewed in this target article is at best inconclusive, at worst flawed.

The analogy of our title is not meant to suggest that colour research should be abandoned any more than painting and literature were in the face of ekphrasis. The allusion is intended as a reminder of the normative dimensions of colour research and of the deep problems of colour ontology, an indicator of the slippage between precept and practice. It would be presumptuous to answer the question: whither colour research? Our aim has been to address the plausibility of the four hypotheses. We do not conclude that research must cease because of messy data or lack of certainty. On the contrary, we suggest that colour scientists may wish to consider what consequences the implausibility of the four hypotheses may have for their own research.


Akins, K.A. & Lamping, J. (1992) More than mere coloring: The art of spectral vision, Behavioral and Brain Sciences, 15: 26-27.

Albright, T.D. (1991) Color and the integration of motion signals, Trends in Neuroscience, 14: 266-69.

Allen, G. (1879) The Colour Sense: Its Origen and Development, London: Trubner.

Allen, K. (1986) Linguistic Meaning, London and New York: Routledge.

Almquist, E. (1883) Studien ber den Farbensinn der Tschuktschen, in Die Wissenschaftlichen Ergebnisse der Vega-Expedition, vol. 1, Leipzig: Brockhaus.

Andersen, V.O., Guld, C. & Sj, O. (1983) Colour processing in prestriate cortex of vervet monkeys, in Mollon & Sharpe (1983) 297-302.

Andree, R. (1878) Ueber den Farbensinn der Naturvlker, Zeitschrift fr Ethnologie, 10: 323-334.

ASTM (1980) Standard Method of Specifying Color by the Munsell System, Designation D 1535-80, Philadelphia: American Society for Testing Materials.

Ayama, M. & Ikeda, M. (1986) Additivity of yellow chromatic valence, Vision Research, 26: 763-769.

Barlow, H.B. (1990): What the brain tells the eye, Scientific American, April, pp. 90-95.

Bartleson, C.J. (1976) Brown, Color Research and Application, 1; 181-191.

Baxter III, W.H. (1983) A look at the history of Chinese color terminology, Journal of the Chinese Language Teachers Association, 18: 1-25.

Beck, B.F. (1969) Colour and heat in South Indian ritual, Man, 4: 553-572.

Beck, J. (1972) Surface Color Perception, Ithaca: Cornell University Press.

Beffa, M.-L. (1978) Rfrence Directe et Connotation: Remarques sur les Noms des Couleurs en Turc et en Chinois, in Tornay (1978) 249-257.

Bender, M.L. (1983) Color Term Encoding in a Special Lexical Domain: Sudanese and Arabic Skin Colors, Anthropological Linguistics, 25: 19-27.

Berlin, B. & E. Berlin (1975) Aguaruna Color Terms, American Ethnologist, 2: 61-87.

Berlin, B. & P. Kay (1969 Universality and Evolution, Berkeley: University of California Press; reprint 1991.

Berlin, B., Kay, P. & W.R. Merrifield (1991) The World Color Survey, Photocopied data available from the Summer Institute of Linguistics, Dallas, Texas, USA.

Bickerton, D. (1981) Roots of Language, Ann Arbor. Karoma.

Biernson, G. (1972) Why did Newton see indigo in the spectrum?, American Journal of Physics, 40: 526.

Billmeyer Jr, F.W. (1992) Review of Berlin & Kay (1991 [1969]), Color Research and Application, 17: 368.

Billock, V.A. (1991) The relationship between simple and double opponent cells, Vision Research, 31: 33-42.

Boas, F. (1891) Second General Report on the Indians of British Columbia, Report of the British Association for the Advancement of Science 1890, London: Murray, pp. 562-715.

Boas, F. (1892) Vocabulary of the Kwakiutl Language, Proceedings of the American Philosophical Society (Philadelphia: Mac Calla), 30: 34-82.

Boas, F. (1931) Notes on the Kwakiutl Vocabulary, International Journal of American Linguistics, 6: 163-178.

Boas, F. (1934) Geographical Names of the Kwakiutl Indians, New York: AMS Press (1969).

Boas, F. & G. Hunt (1905) Kwakiutl Texts (The Jesup North Pacific Expedition, vol. 3; Memoir of the American Museum of Natural History, New York, vol.3; Leiden: Brill, New York: Stechert), New York: AMS Press (1975).

Bolton, R. (1978) Black, white, and red all over: The riddle of color term salience, Ethnology, 17: 287-311.

Bolton, R., A.T. Curtis, & L.L. Thomas (1980) Nepali Color Terms: Salience on a Listing Task, Journal of the Steward Anthropological Society, 12: 309-22.

Bornstein, M.H. (1973a) Color Vision and Color Naming: A Psychophysiological Hypothesis of Cultural Difference, Psychological Bulletin, 80: 257-285.

Bornstein, M.H. (1973b) The Psychophysiological Component of Cultural Difference in Color Naming and Illusion Susceptibility, Behavior Science Notes, 8: 41-101.

Bornstein, M.H. (1978) Chromatic vision in infancy, in Advances in Child Development and Behavior (H. W. Reese & L.P. Lipsitt, eds.), New York: Academic Press, pp. 117-182.

Bornstein, M.H. (1975) The Influence of Visual Perception on Culture, American Anthropologist, 77:774-798.

Bornstein, M.H. (1985) On the Development of Color Naming in Young Children, Brain and Language, 26: 72-93.

Bornstein, M.H. (1987) Perceptual categories in vision and audition, in S. Harnad (ed.), Categorical Perception, Cambridge: Cambridge University Press.

Bornstein, M.H., Kessen, W. & Weiskopf, S. (1976) Color vision and hue categorization in young human infants, Journal of Experimental Psychology, 2: 115-129.

Boster, J. (1986) Can Individuals Recapitulate the Evolutionary Development of Color Lexicons?, Ethnology 25: 61-74.

Bousfield, J. (1979) The World Seen as a Colour Chart, in Classifications in their Social Context (R.F. Ellen & D. Reason, eds.), New York: Academic Press.

Boynton, R.M. (1988) Color Vision, Annual Review of Psychology, 39: 69-100.

Boynton, R.M. & Gordon, J. (1965) Bezold-Brcke shift measured by color-naming technique, Journal of the Optical Society of America, 55: 78-86.

Bradley, A., Switkes, E. & De Valois, K. (1988) Orientation and spatial frequency selectivity of adaptation to colour and luminance gradings, Vision Research, 28: 841-856.

Branstetter, K.B. (1977) A Reconstruction of Proto-Polynesian Color Terminology, Anthropological Linguistics, 19: 1-25.

Brenner, A. (1982) Colour Terms in the Old Testament, Sheffield: JSOT Press.

Bright, W. (1952) Linguistic Innovations in Karok, International Journal of American Linguistics, 18: 53-62.

Brown, A.M. (1990) Development of visual sensitivity to light and color vision in human infants: A critical review, Vision Research, 30: 1159-1188.

Brown, R. & E. Lenneberg (1954) A Study in Language and Cognition, Journal of Abnormal and Social Psychology 49:454-62.

Brown, D.E. (1991) Human Universals, New York: McGraw-Hill.

Bulmer, R.H.N. (1968) Karam Colour Categories, Kivung, 1: 120-133.

Burns, B. & Shepp, B.E. (1988): Dimensional Interactions and The Structure Of Psychological Space: The Representation Of Hue, Saturation, and Brightness, Perception & Psychophysics, 43: 494-507.

Burns, S.A., Elsner, A.E., Pokorny, J. & Smith V.C. (1984) The Abney effect: Chromaticity coordinates of unique and other constant hues, Vision Research, 24: 479- 489.

Campbell, F.W. (1983): Prologue, In Colour Vision: Physiology and Psychophysics (J.D. Mollon & L.T. Sharpe, Eds.). Academic Press, London.

Centlivres Demont, M. & P. Centlivres (1978) Dnomination de Couleurs en Milieu Pluri- Ethnique: L'Exemple Nord-Afghan, in Tornay (1978) 259-284.

Chang, J.J. & J.D. Carroll (1980) Three are Not Enough: An INDSCAL Analysis Suggesting that Color Space Has Seven (1) Dimensions, Color Research and Application, 5: 193-206.

CIE (1987) International Lighting Vocabulary, Publication CIE No. 17.4, Geneva: Bureau Central de la Commission Electrotechnique Internationale.

Cole, G.R., Stromeyer III, C.F. & Kronauer, R.E. (1990): Visual Interactions with luminance and chromatic stimuli, Journal of the Optical Society Of America A, 7: 128- 139.

Collier, G.A. (1973) Review of Berlin & Kay (1969), Language, 49: 245-248.

Collier, G.A. et al (1976) Further Evidence for Universal Color Categories, Language, 52: 884-890.

Conklin, H.C. (1955) Hanuno Color Categories, Southwest Journal of Anthropology, 11: 339-344.

Conklin, H.C. (1973) Color Categorization (review Berlin & Kay 1969), American Anthropologist, 75: 931-942.

Connolly, R.M. (1897) Social Life in Fanti-Land, Journal of the Royal Anthropological Institute of Great Britain and Ireland, 26: 128-154.

Corbett, G. & G. Morgan (1988) Colour Terms in Russian: Reflections of Typological Constraints in a Single Language, Journal of Linguistics, 24:31-64.

Crane, H.D. & Piantanida, T.P. (1983): On seeing reddish green and yellowish blue, Science, 221: 1078-1080.

Curtis, E.S. (1916) The North American Indian, vol. 11, New York: Johnson Reprint.

D'Zmura, M. (1991) Color in Visual Search, Vision Research, 31: 951-966.

Darwin, C. (1877) Biographische Skizze eines kleinen Kindes, Kosmos, 1: 367-376; partly in English in Mind, 2: 285-294..

Davidoff, J. (1974) The Psychological Relationship Between Lightness and Saturation, Perception & Psychophysics, 16: 79-83.

Davidoff, J. (1991) Cognition Through Color, Cambridge MA: The MIT Press.

Davidoff, J. (1992) What is a colour space? Behavioral and Brain Sciences, 15: 34-35.

Davies, I.R.L., G.G. Corbett, G. Laws, H. McGurk, A.E.St.G. Moss & M.W. Smith (1991) Linguistic Basicness and Colour Information Processing, International Journal of Psychology, 26: 311-327.

Dawson, G.M. (1887) Notes and Observations on the Qwakiool People of the Northern Part of Vancouver Island and Adjacent Coasts, Made During the Summer of 1885, Fairfield WA: Ye Galleon Press (1973).

Dehouve, D. (1978) Transformation de la Dnomination des Couleurs Dans les Languages Domines: Un Cas Mexicain, in Tornay (1978) 285-304.

De Valois R.L. & G.H. Jacobs (1968) Primate Color Vision, Science, 162:533-540.

De Valois, R. L. & K. K. De Valois (1975) Neural Coding of Color, in Handbook of Perception (E.C. Carterette & M.P. Friedman, eds.), New York: Academic Press, pp. 117- 166.

De Valois, R.L. & K.K. De Valois (1993) A multi-stage color model, Vision Research, 33: 1053-1065.

De Valois, R.L., I. Abramov & G.H. Jacobs (1966) Analysis of Response Patterns of LGN Cells, Journal of the Optical Society of America, 56: 966-977.

Derefeldt, G. (1991) Colour appearance systems, in The Perception of Colour (P. Gouras, ed.), London: Macmillan, vol. 6, pp. 218-261.

Derrington, A.M., Krauskopf, J. & Lennie, P. (1984) Chromatic mechanisms in lateral geniculate nucleus of macaque, Journal of Physiology, 357: 241-265.

Desimone, R., Schein, S.J., Moran, J. & Ungerleider, L.G. (1985) Contour, color, and shape analysis beyond the striate cortex, Vision Research, 25: 441-452.

DeYoe, E.A. & Van Essen, D.C. (1988) Concurrent processing streams in monkey visual cortex, Trends in Neuroscience, 11: 219-226.

Dimmick, F.L. & Hubbard, M.R. (1939) The spectral location of psychologically unique yellow, green, and blue, American Journal of Psychology, 52: 242-254.

Dobkins, K.R. & T.D. Albright (1993) What happens if it changes color when it moves?, Vision Research, 33: 1019-1036.

Dournes, J. (1978) Les Races de Couleurs, in Tornay (1978) 369-399.

Duck, M. (1987) Bezold-Brcke phenomenon and Goethe's rejection of Newton's Opticks, American Journal of Physics, 55: 793-796.

Durbin, M. (1972) Basic Terms - Off-Color?, Semiotica, 6: 257-278.

Ejima, Y & Takahashi, S. (1985) Interaction between short- and longer-wavelength cones in hue cancellation codes: Nonlinearities of hue cancellation as a function of stimulus intensity, Vision Research, 25: 1911-1922.

Ejima, Y. & Takahashi, S. (1984) Bezold-Brcke hue shift and nonlinearity in opponent- color processes, Vision Research, 24: 1897-1904.

Ejima, Y., Takahashi, S. & Akita, M. (1986) Achromatic sensation for trichromatic mixtures as a function of stimulus intensity, Vision Research, 26: 1065-1071.

Elsner, A.E., Burns, S.A. & Pokorny, J. (1987) Changes in constant-hue loci with spatial frequency, Color Research and Application, 12: 42-50.

Estvez, O. & Dijkhuis, T. (1983) Human pattern evoked potentials and colour coding, in Mollon & Sharpe (1983) 261-268.

Evans, R.M. (1974) The Perception of Color, New York: Wiley.

Evans-Pritchard, E.E. (1933-35) Imagery in Ngok Dinka Cattle Names, Bulletin of the School of Oriental Studies, 7: 623-628.

Evans-Pritchard, E.E. (1940) The Nuer: A Description of the Modes of Livelihood and Political Institutions of a Nilotic People, Oxford: Clarendon Press.

Filliozat, J. (1957) Classement des Couleurs et des Lumires en Sanskrit, in Meyerson (1957) 303-311.

Finkelstein, M.A. (1988) Spectral tuning of opponent channels is spatially dependent, Color Research and Application, 13: 106-112.

Finkelstein, M.A., Harrison, M. & Hood, D.C. (1990) Sites of sensitivity control within a long-wavelength cone pathway, Vision Research, 8: 1145-1158.

Finkelstein, M.A., & Hood, D.C. (1984), Detection and discrimination of small, brief lights: variable tuning of opponent channels, Vision Research, 24: 175-181.

Fischer, W. (1965) Farb- und Formbezeichnungen in der Sprache der Altarabischen Dichtung, Wiesbaden: Harrassowitz.

Flanagan, P., Cavanagh, P. & Favreau, O.E. (1990): Independent orientation-selective mechanisms for the cardinal directions of colour space, Vision Research, 30: 769-778.

Forge, A. (1970) Learning to See in New Guinea, in Socialization (P. Mayer, ed.), London: Tavistock.

Franzen, W. (1990) Die Sprachen und das Denken: Kleine Bestandsaufnahme zum linguistischen Relativismus (Sapir-Whorf-Hypothese), Conceptus, 24: 3-32.

Friedl, E. (1979) Colors and Culture Change in Southwest Iran, Language in Society, 8: 51-68.

Fukui, K. (1979) Cattle Colour Symbolism and Inter-Tribal Homicide among the Bodi, Senri Ethnological Studies, 3: 147-177.

Fuld, K., Werner, J.S. & Wooten, B.R. (1983) The possible elemental nature of brown, Vision Research, 23: 631-637.

Furrell, J.W. (1885) Light from the East on the colour question, The Nineteenth Century, 17: 321-330.

Gage, J. (1978) Colour in History: Relative and Absolute, Art History, 1 (1) 104-130.

Gage, J. (1993) Colour and Culture: Practice and Meaning from Antiquity to Abstraction, London: Thames and Hudson.

Gardner, H. (1987) The Mind's New Science, New York: Basic Books.

Garro, L.C. (1986) Language, Memory, and Focality: A Reexamination, American Anthropologist, 88: 128-136.

Garth, T.R. (1922) The color preferences of five hundred and fifty-nine full-blood Indians, Journal of Experimental Psychology, 5: 392-418.

Gtje, H. (1967) Zur Farbenlehre in der muslimischen Philosophie, Der Islam, 43: 280- 301.

Gatschett, A.S. (1879) Adjectives of Color in Indian Languages, American Naturalist, 13: 475-85.

Geddes, W.R. (1946) The Colour Sense of Fijian Natives, British Journal of Psychology, 37: 30-36.

Geiger, L. (1871) Zur Entwicklungsgeschichte der Menschheit, Stuttgart: Cotta; translated as Contributions to the History of the Development of the Human Race, London: Trbner (1880).

Geiger, L. (1872) Ursprung und Entwickelung der menschlichen Sprache und Vernunft, 2 vols., Stuttgart: Cotta.

Gernet, J. (1957) L'Expression de la Couleur en Chinois, in Meyerson (1958) 295-301.

Gibson, J.J. (1979) The Ecological Approach to Visual Perception, Hillsdale NJ: Lawrence Erlbaum.

Gladstone, W.E. (1858) Homer's Perceptions and Use of Colour, in his Studies on Homer and the Homeric Age, vol. 3. Oxford: Oxford University Press, pp. 457-499.

Gladstone, W.E. (1877) The Colour Sense, Nineteenth Century, 2: 366-88.

Goethe, J.W. von (1840) Theory of Colours, London: Murray; reprint 1970, Cambridge MA: MIT Press.

Gouras, P. (1984): Color Vision, In Progress In Retinal Research (N. Osborn & J. Chader, Eds.), London: Pergamon Press), 3: 227-260.

Gouras, P. (1991) Precortical physiology of colour vision and Cortical mechamisms of colour vision, in The Perception of Colour (P. Gouras, ed.), London: Macmillan, vol. 6, pp. 163-197.

Gouras, P. & H. Eggers (1984) Hering's Opponent Color Channels Do Not Exist in the Primate Retinogeniculate Pathway, Ophthalmic Research, 16: 31-35.

Grubb, D. McC. (1977) A Practical Writing System and Short Dictionary of Kwakw'ala (Kwakiutl), Ethnology Service Paper No. 34, Ottawa: National Museums of Canada.

Haenny, P.E., Maunsell, J.H.R., & Schiller, P.H. (1988) State dependent activity in monkey visual cortex, Experimental Brain Research, 69: 245-259.

Hamayon, R. (1978) Des Fards, des Moeurs et des Couleurs, in Tornay (1978) 207-247.

Hamp, E.P. (1980) Notes on Proto-Polynesian Colours, Anthropological Linguistics, 22: 390-391.

Hardin, C.L. (1988): Color For Philosophers: Unweaving The Rainbow. Hackett, Indianapolis.

Harkness, S. (1973) Universal Aspects of Learning Color Codes: A Study in Two Cultures, Ethos 2: 175-200.

Heggelund, P. (1992) A bidimensional theory of achromatic color vision, Vision Research, 32: 2107-2119.

Heggelund, P. (1993) Simultaneous luminance contrast with chromatic colors, Vision Research, 33: 1709-1722.

Heider, K.G. (1970), The Dugum Dani, Chicago: Aldine Publishing Co.

Heider, K.G. (1979) Grand Valley Dani: Peaceful Warriors, New York: Holt, Rinehart and Winston.

Helmholtz, H. von (1911): Handbuch Der Physiologischen Optik. Band Ii. Voss, Hamburg. Quoted From Helmholtz' Treatise On Physiological Optics (J.P.C. Southall, Ed.). Dover, New York (1962).

Helson, H. & Lansford, T. (1970) The role of spectral energy of source and background color in the pleasantness of object colors, Applied Optics, 9: 1513-1562.

Hering, E. (1964): Outlines of a Theory of the Light Sense, transl. L.M. Hurvich & D. Jameson, Cambridge: Harvard University Press; first German edition 1878, translation of 1920 edition.

Herne, G. (1954) Die slavischen Farbenbenennungen: Eine semasiologisch- etymologische Untersuchung, Uppsala: Almqvist & Wiksell.

Hess, J.-J. (1920) Die Farbbezeichnungen bei Innerarabischen Beduinenstmmen, Der Islam, 10: 74-86.

Hess, R.F., Mullen, K.T. & Zrenner, E. (1989) Human photopic vision with only short wavelength cones: post-receptoral properties, Journal of Physiology, 417: 151-172.

Hickerson, N.P. (1953) Ethnolinguistic Notes from Lexicons of Lokono (Arawak), International Journal of American Linguistics, 19: 181-190.

Hickerson, N.P. (1971) Review of Berlin & Kay (1969), International Journal of American Linguistics, 37: 257-70.

Hickerson, N.P. (1983) Gladstone's Ethnolinguistics: the Language of Experience in the Nineteenth Century, Journal of Anthropological Research, 39: 26-41.

Hickerson, N.P. (1988) A Reconsideration of Lokono Color Terms, paper presented at the International Congress of Americanists, Amsterdam 1988.

Holenstein, E. (1985) Sprachliche Universalien: Eine Untersuchung zur Natur des menschlichen Geistes, Bochum: Brockmeyer.

Holmer, N.M. (1954/55) Amerindian color semantics, International Anthropological Linguistic Review, 2: 158-166.

Hood, D.C. & Finkelstein, M.A. (1983) A case for the revision of textbook models of color vision: the detection and appearance of small brief lights, in Mollon & Sharpe (1983) 385-398.

Hopkins, E.W. (1883) Words for Color in the Rig Veda, American Journal of Philology, 4: 166-91.

Hubel, D.H. & Livingstone, M.S. (1991) Comment on Ingling & Grigsby 1990, Vision Research, 31: 1655-1656.

Hubel, D.H. & M.S. Livingstone (1990) Color and contrast sensitivity in the lateral geniculate body and primary visual cortex of the Macaque monkey, Journal of Neuroscience, 10: 2223-2237.

Hunt, R.W.G. (1977) The specification of colour appearance, Color Research and Application, 2: 55-68.

Hurlbert, A. (1991) Deciphering the colour code, Nature, 349: 191-193.

Hurvich, L.M. (1981): Color Vision. Sunderland: Sinauer.

Hurvich, L.M. (1985) Opponent-colours theory, in Ottoson & Zeki (1985) 61-82.

Hurvich, L.M. and D. Jameson (1955) Some quantitative aspects of an opponent-colors theory. II. Brightness, saturation and hue in normal and dichromatic vision. Journal of the Optical Society of America, 45:602-16.

Ikeda, M. & Nakano, Y. (1986) Spectral luminous-efficiency functions obtained by direct heterochromatic brightness matching for 2 deg and 10 deg fields, Journal of the Optical Society of America A, 3: 2105-2108.

Indow, T. (1988): Multidimensional studies of Munsell color solid, Psychological Review, 95: 456-470.

Ingling, C.R. (1977) The spectral sensitivity of the opponent-color channels, Vision Research, 17: 1083-1089.

Ingling, C.R. & E. Martinez (1985) The spatiotemporal properties of the r-g X-cell channel, Vision Research, 25: 33-38.

Ingling, C.R. & Grigsby, S.S. (1990) Perceptual correlates of magnocellular and parvocellular channels: Seeing form and depth in afterimages, Vision Research, 30: 823- 828.

Ingling, C.R. & Grigsby, S.S. (1991) Perceptual correlates of magnocellular and parvocellular channels: Seeing form and depth in afterimages, Vision Research, 31: 1657- 1658.

Ingling, C.R., Russell, P.W., Rea, M.S. & Tsou, B.H.-P. (1978) Red-green opponent spectral sensitivity: disparity between cancellation and direct matching methods, Science, 201: 1221-1223.

Irwin, E. (1974) Colour Terms in Greek Poetry, Toronto: Hakkert.

Jacobs, G.H. (1992) Data and interpretation in comparative colour vision, Behavioral and Brain Sciences, 15: 40-41.

Jacobson-Widding, A. (1979) Red-White-Black as a Mode of Thought: A Study of Triadic Classification by Colours in the Ritual Symbolism and Cognitive Thought of the Peoples of the Lower Congo, Uppsala: Almquist and Wiksell.

Jameson, D. and L.M. Hurvich (1955) Some quantitative aspects of an opponent-colors theory. I. Chromatic responses and spectral saturation. Journal of the Optical Society of America, 45:546-52.

Javadna, A. & Ruddock, K.H. (1988) Simultaneous processing of spatial and chromatic components of patterned stimuli by the human visual system, Spatial Vision, 3: 115-127.

Jochelson, H. (1908) Material, Culture and Social Organization, Jesup North Pacific Expedition, vol. VI, part 2, p. 415.

Jordan, G., & Mollon, J.D. (1993) A study of women heterozygous for colour deficiencies, Vision Research, 33: 1495-1508.

Juillerat, B. (1978) Vie et Mort dans le Symbolisme Iafar des Couleurs (Nouvelle- Guine), in Tornay (1978) 497-523.

Katz, D. (1935) The World of Colour, London: Kegan Paul.

Kay, P. (1975) Synchronic Variability and Diachronic Change in Basic Color Terms, Language in Society, 4: 257-270.

Kay, P. (1977) Language evolution and speech style, in B.G. Blount & M. Sanches (eds.), Variability and Change: Sociocultural Dimensions of Language Change, New York: Academic Press, pp. 21-33.

Kay, P. & C.K. McDaniel (1978) The Linguistic Significance of the Meanings of Basic Color Terms, Language, 54: 610-646.

Kay, P., B. Berlin, & W.R. Merrifield (1991) Biocultural Implications of Color Naming, Linguistic Anthropology, 1: 12-25.

Kidder, J.N. (1989) Comment on Duck 1987, American Journal of Physics, 57: 1042- 1043.

Kim, A.I. (1985) Korean Color Terms: An Aspect of Semantic Fields and Related Phenomena, Anthropological Linguistics, 27: 425-436.

Kinkade, M.D. (1988) Proto-Salishan Color, in W. Shipley (ed.), In Honor of Mary Haas, Berlin: Mouton de Gruyter, pp. 443-466.

Kirchhoff, A. (1879) Ueber Farbensinn und Farbenbezeichnung der Nubier, Zeitschrift fr Ethnologie, 11: 397-402.

Klee, P. (1961) The Thinking Eye, London: Lund Humphrey.

Kooi, F.J. & De Valois, K.K. (1992) The role of color in the motion system, Vision Research, 32: 657-668.

Krause, E. (1877) Die geschichtliche Entwicklung des Farbensinnes, Kosmos, 1: 264-272.

Krauskopf, J. & B. Farell (1990) Influence of colour on the perception of coherent motion, Nature, 348: 328-331.

Krauskopf, J., Williams, D.R. & Heeley, D.W. (1982) Cardinal directions in color space, Vision Research, 22: 1123-1131.

Krauskopf, J., Williams, D.R., Mandler, M.B. & Brown, A.M. (1986) Higher order color mechanisms, Vision Research, 26: 23-32.

Kristol, A.M. (1978) Color: Les Langues Romanes Devant le Phnomne de la Couleur, Bern: Francke.

Kristol, A.M. (1980) Color Systems in Southern Italy: A Case of Regression, Language, 56: 137-145.

Krger, J. & Fischer, B. (1983) Colour columns and colour areas, in Mollon & Sharpe (1983) 291-296.

Kuehni, R.G. (1983): Color: Essence and Logic. Van Nostrand, New York.

Kuschel, R. & T. Monberg (1974) `We don't talk much about colour here': A Study of Colour Semantics on Bellona Island, Man, 9:213-242.

Ladd-Franklin, C. (1901) Color-Introspection on the Part of the Eskimo, Psychological Review, 8: 396-402.

Lakoff, G. (1987) Women, Fire, and Dangerous Things, Chicago: The University of Chicago Press.

Land, E.H. (1986) Recent Advances in Retinex Theory, Vision Research, 26: 7-21.

Laude-Cirtautas, I. (1961) Der Gebrauch der Farbbezeichnungen in den Trkdialekten, Wiesbaden: Harrassowitz.

Lee, B.B., Martin, P.R. & Valberg, A. (1989): Nonlinear summation of M- and L-cone inputs to phasic retinal ganglion cells of the macaque, The Journal Of Neuroscience, 9: 1433-1442.

Lenneberg, E.H. (1953) Cognition in Ethnolinguistics, Language, 29: 463-471.

Lenneberg, E.H. & J. Roberts (1956) The Language of Experience: a Study in Methodology, Memoir 13, Indiana University Publications in Anthropology and Linguistics, Baltimore: Waverly Press.

Lennie, P. & D'Zmura, M. (1988): Mechanisms of color vision, Critical Reviews In Neurobiology, 3: 333-400.

Lennie, P., Krauskopf, J. & Sclar, G. (1990) Chromatic mechanisms in striate cortex of macaque, Journal of Neuroscience, 10: 649-699.

Lewitz, S. (1974) Recherches sur le vocabulaire Cambodgien, Journal Asiatique, 162: 151-170.

Lienhardt, G. (1970) Divinity and Experience: the Religion of the Dinka, Oxford: Clarendon Press.

Lincoln, N.J. & J.C. Rath (1980) North Wakashan Comparative Root List, Canadian Ethnology Service Paper No. 68, Ottawa: National Museums of Canada.

Livingstone, M.S. & D.H. Hubel (1984) Anatomy and Physiology of a Color System in the Primate Visual Cortex, Journal of Neuroscience, 4: 309-356.

Livingstone, M.S. & D.H. Hubel (1988) Segregation of Form, Color, Movement, and Depth: Anatomy, Physiology, and Perception, Science, 240: 740-749.

Lockhead, G.R. (1992) Psychophysical Scaling: Judgments of Attributes of Objects?, Brain and Behavioral Sciences, 15: 543-601.

Lucy, J. & R.A. Shweder (1979) Whorf and His Critics: Linguistic and Nonlinguistic Influences on Color Memory, American Anthropologist, 81: 581-615.

Lumsden, C.J. (1985) Color Categorization: A Possible Concordance Between Genes and Culture, Proceedings of the National Academy of Science of the U.S.A., 82: 5805-5808.

Lumsden, C.J. & E.O. Wilson (1981) Genes, Mind, and Culture. The Coevolutionary Process, Cambridge MA: Harvard University Press.

Luria, A.R. (1976) Cognitive Development: Its Cultural and Social Foundations, Cambridge MA: Harvard University Press; first Russian edition 1974.

Mach, E. (1919) Die Analyse der Empfindungen, Jena: Fischer.

MacLaury, R.E. (1986) Color in Mesoamerica, Vol. 1: A Theory of Composite Categorization, Doctoral Dissertation, University of California, Berkeley.

MacLaury, R.E. (1987) Color-Category Evolution and Shuswap Yellow-with-Green, American Anthropologist, 89:107-124.

MacLaury, R.E. (1991) Social and Cognitive Motivations of Change: Measuring Variability in Color Semantics, 67:34-62.

MacLaury, R.E. (1992) From Brightness to Hue: An Explanatory Model of Color- Category Evolution, Current Anthropology, 33: 137-186.

MacLaury, R.E. & B. Galloway (1988) Color Categories and Color Qualifiers in Halkomelem, Samish, Lushootseed, and Yakima, Working Papers of the 23rd International Conference on Salishan and Neighboring Languages, Eugene WA.

MacNichol, E.F. Jr, Levine, J.S., Mansfield, R.J.W., Lipez, L.E. & Collins, B.A. (1983) Microspectrophotometry of visual pigments in primate receptors, in Mollon & Sharpe (1983) 13-38.

Magnus, H. (1877) Die geschichtliche Entwickelung des Farbensinnes, Leipzig: Veit.

Magnus, H. (1880) Untersuchungen ber den Farbensinn der Naturvlker, Jena: Fischer.

Man, T.M. & D.L. MacAdam (1989) Three-dimensional scaling of the uniform color scales of the Optical Society of America, Journal of the Optical Society of America A, 6: 128-140.

Martin, K.A.C. (1988) From enzymes to visual perception: a bridge too far?, Trends in Neuroscience, 9: 380-384.

Martindale, C. & Moore, K. (1988): Priming, prototypicality, and preferences, Journal of Experimental Psychology, 14: 661-670.

Mathiot, M. (1979) Folk-Definitions as a Tool for the Analysis of Lexical Meaning, in Ethnolinguistics: Boas, Sapir and Whorf Revisited (M. Mathiot, ed.), The Hague: Mouton.

Maxwell-Stuart, P.G. (1981) Studies in Greek Colour Terminology, Vols. I and II, Leiden: Brill.

Merrigan, W.H. (1989) Chromatic and achromatic vision of Macaques: Role of the P pathway, Journal of Neuroscience, 9: 776-783.

Mervis, C.B. & E.M. Roth (1980), The Internal Structure of Basic and Non-basic Color Categories, Language, 57: 384-405.

Meyerson, I. (1957) Problmes de la Couleur, Exposs et discussions du Colloque du Centre de Recherches de Psychologie comparative tenu Paris les 18, 19, 20 mai 1954, Paris: S.E.V.P.E.N.

Michael, C.R. (1983) Color processing in primate striate cortex, in Mollon & Sharpe (1983) 277-290.

Michael, C.R. (1985) Non-oriented double opponent colour cells are concentrated in two subdivisions of cortical layer IV, in Ottoson & Zeki (1985) 199-210.

Michael, C.R. (1988) Retinal afferent arborization patterns, dendritic field orientations, and the segregation of function in the lateral geniculate nucleus of the monkey, Proceedings of the National Academy of Science USA, 85: 4914-4918.

Miller, G.A. & P.N. Johnson-Laird (1976) Language and Perception, Cambridge MA: Harvard University Press.

Mollon, J.D. (1987) On the nature of models of colour vision, Die Farbe, 34: 29-46.

Mollon, J.D. (1989): "Tho She Kneeld In That Place Where They Grew ...": The uses and origins of primate colour vision, Journal Of Experimental Biology, 146: 21-38.

Mollon, J.D. (1990) Neurobiology: the Club-Sandwich Mystery, Nature, 343: 16-17.

Mollon, J.D. (1992) Review of Davidoff 1991, Trends in Neuroscience, 15: 194.

Mollon, J.D., & L.T. Sharpe (eds.) (1983) Colour Vision: Physiology and Psychophysics, London: Academic Press.

Monberg, T. (1971) Tikopia Color Classification, Ethnology, 10: 349-358.

Montag, E.D. & Boynton, R.M. (1987) Rod influence in dichromatic surface color perception, Vision Research, 27: 2153-2162.

Morgan, G. & G. Corbett (1989) Russian Colour Term Salience, Russian Linguistics, 13: 125-141.

Moss, A., I. Davies, G. Corbett & G. Laws (1990) Mapping Russian Basic Colour Terms Using Behavioural Measures, Lingua, 82: 313-332.

Moss, A.E. (1989a) Basic Colour Terms, Lingua, 78: 313-320.

Moss, A.E. (1989b) Does Russian Have a Basic Term for Purple?, Lingua, 27: 145-155.

Neitz, J. & Jacobs, G.H. (1990) Polymorphism in normal human color vision and its mechanism, Vision Research, 30: 621-636.

Neitz, J., M. Neitz & G.H. Jacobs (1993) More than three different cone pigments among people with normal colour vision, Vision Research, 33: 117-122.

Nepote, J. & H.D. Khing (1978) L'Organisation du Champ de la Couleur en Cambodgien, in Tornay (1978) 83-107.

Newcomer, P. & J. Faris (1971) Review of Berlin & Kay (1969), International Journal of American Linguistics, 37: 270-275.

Newton, I. (1952): Opticks, Dover Publications.

Nickerson, D. (1981) OSA Uniform Color Scale Samples: A Unique Set, Color Research and Application, 6: 7-9.

Nickerson, D. & S.M. Newhall (1943) A Psychological Color Solid, Journal of the Optical Society of America, 33: 419-422.

Ohmura, H. (1988) Disparities of shape and color in determining the type of apparent movement, Japanese Psychological Research, 30: 1-10.

Ottoson, T. & S. Zeki (eds.) (1985) Central and Peripheral Mechanisms of Colour Vision, New York: Macmillan.

Panoff-Eliet, F. (1971) Review of Berlin & Kay (1969), L'Homme, 11 (4) 100-3.

Paritsis, N.C. & Stewart, D.J. (1983): A Cybernetic Approach To Colour Perception. Gordon & Beach, New York.

Paulus, W. & Krger-Paulus, A. (1983) A new concept of retinal colour coding, Vision Research, 23: 529-540.

Pokorny, J. (1959) Indogermanisches Etymologisches Wrterbuch, Bern: Francke.

Pokorny, J., S.K. Shevell & V.C. Smith (1991) Colour appearance and colour constancy, in The Perception of Colour (P. Gouras, ed.), London: Macmillan, vol. 6, pp. 43-6

Pointer, M.R. (1980) The concept of colourfulness and its use for deriving grids for assessing colour appearance, Color Research and Application , 5:99-107.

Preuss, R.J. (1981) Colour Naming in Young Children, M.A. Thesis, Vancouver: The University of British Columbia.

Pridmore, R.W. (1990) Model of saturation and brightness: Relations with luminance, Color Research and Application, 15: 344-357.

Pritsak, O. (1954) Orientierung und Farbsymbolik: Zu den Farbenbezeichnungen in den Altaischen Vlkernamen, Saeculum, 5: 376-383.

Proulx, P. (1988) Algic Color Terms, Anthropological Linguistics, 30: 135-149.

Purdy, D. Mcl. (1931): Spectral hue as a function of intensity, Journal of Psychology, 43: 541-559.

Purdy, D. Mcl. (1937): The Bezold-Brcke phenomenon and contours for constant hue, American Journal of Psychology, 49: 313-315.

Quinn, P.C., Rosano, J.L. & Wooten, B.R. (1988): Evidence that brown is not an elemental color, Perception & Psychophysics, 43: 156-164.

Rabin, J. & Adams, A.J. (1992) Lightness induction in the S-cone pathway, Vision Research, 9: 1771-1774.

Radloff, W. (1871) Die Hausthiere der Kirgisen, Zeitschrift fr Ethnologie, 3: 285-313.

Ratliff, F. (1976) On the Psychophysiological Bases of Universal Color Terms, Proceedings of the American Philosophical Society, 120: 311-330.

Ratner, C. (1989) A Sociohistorical Critique of Naturalistic Theories of Color Perception, Journal of Mind & Behavior, 10: 361-372.

Ray, V.F. (1953) Human Color Perception and Behavioral Response, Transactions of the New York Academy of Sciences, [2] 16: 98-105.

Riley, C.L. (1963) Color-direction symbolism: An example of Mexican-South-Western contacts, Amrica Indgena, 23:49-60.

Rivers, W.H.R. (1901) Colour vision, in Reports of the Cambridge Anthropological Expedition to Torres Straits, Volume II: Physiology and Psychology, Cambridge: At the University Press, pp. 1-63; see also Popular Science Monthly, 59: 44-58.

Rivers, W.H.R. (1902) The Colour Vision of the Eskimos, Proceedings of the Royal Anthropological Institute of Great Britain and Ireland, pp. 143-149.

Rivers, W.H.R. (1903) Observations on the Vision of the Uralis and the Sholagas, Bulletin of the Madras Government Museum, 5: 3-18.

Rivers, W.H.R. (1905) Observations on the Senses of the Todas, British Journal of Psychology, 1: 321-396.

Rosch Heider, E. (1971) "Focal" Color Areas and the Development of Names, Development Psychology, 4: 447-455.

Rosch Heider, E. (1972a) Universals in Color Naming and Memory, Journal of Experimental Psychology, 93: 10-20.

Rosch Heider, E. (1972b) Probabilities, Sampling, and Ethnographic Method: The Case of Dani Colour Names, Man 7:448-466.

Rosch Heider, E. (1973a) Natural Categories, Cognitive Psychology, 4: 328-350.

Rosch Heider, E. (1973b) On the Internal Structures of Perceptual and Semantic Categories, in Cognitive Development and the Acquisition of Language (I.E. Moore, ed.), New York: Academic Press, pp. 111-144.

Rosch Heider, E. & D.C. Olivier (1972) The Structure of the Color Space in Naming and Memory for Two Languages, Cognitive Psychology, 3: 337-354.

Rosch, E. (1974), Linguistic Relativity, in Human Communication: Theoretical Explorations (A. Silverstein, ed.), New York: Halsted Press, pp. 95-121.

Rosch, E. (1975) Cognitive Reference Points, Cognitive Psychology, 7: 532-547.

Rosch, E. (1978) Principles of Categorisation, in Cognition and Categorisation (E. Rosch & B.B. Lloyd, eds.), Hillsdale NJ: Erlbaum, pp. 28-49.

Rowe, C. (1974) Conceptions of colour and colour symbolism in the ancient world, in The Realms of Colour (A. Portmann & R. Ritsema, eds.), Eranos, 41 (1972) 327-363.

Sahlins, M. (1976) Colors and Cultures, Semiotica, 16: 1-22.

Saunders, B.A.C. (1992) The Invention of Basic Colour Terms, Utrecht: ISOR

Saunders, B.A.C. (1993) On the Concept of 'Basic' in Berlin & Kay's Basic Color Terms, Antropologische Verkenningen, 12: 35-50.

Saunders, B.A.C. (1995) Disinterring Basic Color Terms: A Study in the Mystique of Cognitivism, History of the Human Sciences 8 (4):19-38.

Saunders, B.A.C. (1995a) Four Theses on Colour. Presented at the Workshop at the Zentrum fr interdisziplinre Forschung (ZiF), Universitt Bielefeld, Germany, 28 November - 1 December.

Saunders, B.A.C. (1996) A Model of Colour and its Fallacies, Proceedings of the Einstein meets Magritte Conference, Brussels, Belgium May-June 1995 (forthc.)

Saunders, B.A.C. & J. van Brakel (1988) Re-evaluating Basic Colour Terms, Cultural Dynamics, 1: 359-378.

Saunders, B.A.C. en J. van Brakel (1995) Translating the World Color Survey, Post- Modernism and Anthropology ed. (K. Geuijen, D. Raven, & J. de Wolf eds.), Assen: van Gorcum, pp.161-78.

Saunders, B.A.C. & J. van Brakel (1996) The Phantom Objectivity of Colour, Proceedings of the Conference: Linguistic Foundations of Translation , Liverpool, U.K., Sept. 1995 (forthc.).

Schefrin, B.E. & J.S. Werner (1990) Loci of Spectral Unique Hues Throughout the Life Span, Journal of the Optical Society of America A, 7: 305-311.

Schein, S.J., Marocco, R.T. & DeMonasterio, F.M. (1982) Is there a high concentration of color-selective cells in area V4 of monkey visual cortex?, Journal of Neurophysiology, 47: 193-213.

Schiller, P.H., Logothetis, N.K. & Charles, E.R. (1990): Functions of the colour-opponent and broad-band channels of the visual system, Nature, 343: 68-70.

Schulze, W. (1910) Etymologisches, Sitzungsberichte der kniglichen Preussischen Akademie der Wissensenschaften, pp. 787-808.

Schwentner, E. (1915) Eine sprachgeschichtliche Untersuchung ber den Gebrauch und die Bedeutung der altgermanischen Farbenbezeichnungen, Doctoral Dissertation, Gttingen.

Senft, G. (1987) Kilivila Color Terms, Studies in Language, 11: 313-346.

Sepper, D.L. (1988) Goethe Contra Newton, Cambridge: Cambridge University Press.

Shapley, R. (1990): Visual sensitivity and parallel retinocortical channels, Annual Review Of Psychology, 41: 635-658.

Shepard, R. N. (1991) The Perceptual Organization of Colors: An Adaptation to Regularities of the Terrestrial World? In J.H. Barkow et al. (eds) The Adapted Mind , Oxford and New York: OUP, pp. 495-532.

Sherman, P.D. (1981) Color Vision in the Nineteenth Century, Bristol: Hilger.

Shevell, S.K. (1992) Redness from short-wavelength-sensitive cones does not induce greenness, Vision Research, 32: 1551-1556.

Shevell, S.K. & Handte, J.P. (1983) Postreceptoral adaptation in suprathreshold color perception, in Mollon & Sharpe (1983) 399-407.

Shevell, S.K. & Humanski, R.A. (1988) Color perception under chromatic adaptation: Red/green equilibria with adapted short-wavelength cones, Vision Research, 28: 1345- 1356.

Shevell, S.K., I. Holliday & P. Whittle (1992) Two separate neural mechanisms of brightness induction, Vision Research, 32: 23331-2340.

Shweder, R.A. & Bourne, E.J. (1984) Does the concept of the person vary cross- culturally?, in Culture Theory: Essays on Mind, Self, and Emotion (R.A. Shweder & R.A. LeVine, eds.), Cambridge: Cambridge University Press.

Simon, K. (1951) Colour Vision of Buganda Africans, East African Medical Journal, 28: 59-75.

Skard, S. (1946) The Use of Color in Literature: A Survey of Research, Proceedings of the American Philosophical Society, 90: 163-249.

Smith, L.B. & M.D. Sera (1992) A developmental analysis of the polar structure of dimensions Cognitive Psychology, 24: 99-142.

Snow, D.L. (1971) Samoan Color Terminology: A Note on the Universality and Evolutionary Ordering of Color Terms, Anthropological Linguistics, 13: 385-390.

Spencer, B. & F.J. Gillen (1927) The Arunta, vol. 2, London: Macmillan.

Steriade, M. & Deschenes, M. (1985) The thalamus as a neuronal oscillator, Brain Research Reviews, 18: 165-170.

Sternheim, C.E. & Boynton, R.M. (1966) Uniqueness of perceived hues investigated with a continuous judgmental technique, Journal of Experimental Psychology, 72: 770-776.

Stockman, A., MacLeod, D.I.A. & DePriest, D.D. (1991) The temporal properties of the human short-wave photoreceptors and their associated pathways, Vision Research, 31: 189-208.

Stromeyer, C.F., Eskew, R.T., Kronauer, R.E. & L. Spillmann (1991) Temporal phase response of the short-wave cone signal for color and luminance, Vision Research, 31: 787-803.

Suppes, P., Krantz, D.M., Luce, R.D. & Tversky, A. (1990) Foundations of Measurement, New York: Academic Press, volume II, chapter 15.

Tanaka, M., Lee, B.B. & Creutzfeldt, O.D. (1983) Spectral tuning and contour representation in area 17 of the awake monkey, in Mollon & Sharpe (1983) 269-276.

Tansley, B.W., Robertson, A.W. & Maughan, K.E. (1983) Chromatic and achromatic border perception, in Mollon & Sharpe (1983) 445-454.

Teller, D.Y. (1984) Linking propositions, Vision Research, 24: 1233-1246.

Teller, D.Y. (1991) Simpler Arguments Might Work Better, Philosophical Psychology, 4: 51-60.

Teller, D.Y. & Bornstein, M.H. (1987): Infant color vision and color perception, In Handbook Of Infant Perception (P. Salapatek & L. Cohen, Eds.), Academic Press, Vol. 1: 185-236.

Titchener, E.B. (1916) On Ethnological Tests of Sensation and Perception with Special Reference to Tests of Color Vision and Tactile Discrimination Described in the Reports of the Cambridge Anthropological Expedition to Torres Straits, Proceedings of the American Philosophical Society, 55: 204-236.

Thompson, E., A. Palacios, & F.J. Varela (1992) Ways of Coloring, Behavioral and Brain Sciences 15:1-74.

Topper, D. (1990) Newton on the number of colours in the spectrum, Studies in the History and Philosophy of Science, 21: 269-279.

Tornay, S. (1973) Langage et Perception: La dnomination des couleurs chez les Nyangatom du Sud-Ouest thiopien, L'Homme, 13 (4) 66-94.

Tornay, S. (1978a), Introduction, in Tornay (1978) IX-LI.

Tornay, S. (1978b), L'Etude de la Dnomination des Couleurs, in Tornay (1978): 639- 668.

Tornay, S. (1978c), Perception des Couleurs et Pense Symbolique, in Tornay (1978) 609-637.

Tornay, S. (ed.) (1978) Voir et Nommer les couleurs, Nanterre: Laboratoire d'Ethnologie et de Sociologie Comparative; cf. Hickerson 1980.

Ts'o, D.Y. & D. Gilbert (1988) The organization of chromatic and spatial interactions in the primate striate cortex, Journal of Neuroscience, 8: 1712-1727.

Turner, V. (1967) Color Classification in Ndembu Ritual: A Problem in Primitive Classification, in his The Forest of Symbols: Aspects of Ndembu Ritual, Ithaca: Cornell University Press, pp. 59-92.

Turton, D. (1978) La catgorisation de la couleur en Mursi, pp. 347-367 in Tornay (1978).

Turton, D. (1980) There's no Such Beast: Cattle and Colour Naming Among the Mursi, Man, 15:320-338.

Uchikawa, K., Uchikawa, H. & Kaiser, P.K. (1984) Luminance and saturation of equally bright colors, Color Research and Application, 9: 5-14.

van Brakel, J. (1992) The ethnocentricity of colour, Behavioral and Brain Sciences, 15 (1992) 53.

van Brakel, J. (1993) The plasticity of categories: The case of colour, British Journal for the Philosophy of Science, 44: 103-135.

van Brakel, J. (1994) The ignis fatuus of semantic universalia: the case of colour, British Journal for the Philosophy of Science, 45: 770-783.

van der Stap, P.A.H. (1966) Outline of Dani Morphology, The Hague: Nijhoff.

van Wijk, H.A.C.W. (1959) A Cross-Cultural Theory of Colour and Brightness Nomenclature, Bijdragen Land-, Taal-, en Volkenkunde, 115: 113-37.

Vasyutin, V.V. & A.A. Tishchenko (1989) Space Coloristics, Scientific American, July: 84-90.

Virchow, R. (1878) In Berlin anwesenden Nubier, Zeitschrift fr Ethnologie, 10: 333- 356.

Volbrecht, V.J., Werner, J.S. & Cicerone, C.M. (1990): Additivity of spatially induced blackness, Journal of the Optical Society Of America A, 7: 106-111.

von Kllay, U. (1939) Die zweierlei Farbenortungen einiger Indianerstmme Nordamerikas, Mitteilungen der Anthropologischen Gesellschaft, 49: 11-23.

Vos, J.J. (1986): Are unique and invariant hues coupled?, Vision Research, 26: 337-342.

Webster, M.A. & J.D. Mollon (1991) Changes in colour appearance following post- receptoral adaptation, Nature, 349: 235-238.

Weise, O. (1878) Die Farbenbezeichnungen der Indogermanen, Beitrge zur Kunde der indogermanischen Sprachen, 2: 273-290.

Werner, J.S., & Wooten, B.R. (1979) Opponent chromatic mechanisms: relation to photopigments and hue naming, Journal Of The Optical Society Of America, 69: 422- 434.

Wescott, R.W. (1970) Bini Color Terms, Anthropological Linguistics, 12: 349-360.

Whitely, W.H. (1973) Colour-words and Colour-values: The Evidence from Gusii, in Modes of Thought: Essays on Thinking in Western and Non-Western Societies (R. Horton & R. Finnegan, eds.), London: Faber and Faber.

Whittle, P. (1994) The psychophysics of contrast-brightness, in Gilchrist, A.L. (ed.) Lightness, Brightness and Transparency. Hillsdale NJ: Lawrence Erlbaum.

Wiegersma, S., & van Loon, A. (1989) Some variables in the blue (red) phenomenon, The Journal of General Psychology, 116: 259-269.

Wierzbicka, A. (1990) The Meaning of Color Terms: Semantics, Culture and Cognition, Cognitive Linguistics, 1: 99-150.

Wood, F.A. (1902) Color-Names and Their Congeners, Halle: Niemeyer.

Woodworth, R.S. (1905) Color sense in different races of mankind, Proceedings of the Society for Experimental Biology and Medicine, (3) 24-26.

Woodworth, R.S. (1910) The Puzzle of Color Vocabularies, The Psychological Bulletin, 7: 325-334.

Wyszecki, G. & W.S. Stiles (1982) Color Science, New York: Wiley, 2nd ed.

Yaguchi, H. & Ikeda, M. (1982) Nonlinear nature of the opponent-color channels, Color Research and Application, 7: 187-190.

Yaguchi, H. & Ikeda, M. (1983) Contribution of opponent-colour channels to brightness, in Mollon & Sharpe (1983) 353-360.

Zahan, D. (1974) White, Red and Black: Colour Symbolism in Black Africa, in The Realms of Colour (A. Portmann & R. Ritsema, eds.), Eranos, 41 (1972) 365-395.

Zaidi, Q. & D. Halevy (1993) Visual mechanisms that signal the direction of color changes, Vision Research, 33: 1037-1051.

Zeki, S. (1984) The construction of colours by the cerebral cortex, Proceedings of the Royal Institution of Great Britain, pp. 231-257.

Zeki, S. (1985) Colour pathways and hierarchies in the cerebral cortex, in Ottoson & Zeki (1985) 19-44.

Zeki, S. & Shipp, S. (1988) The functional logic of cortic connections, Nature, 335: 311- 317.

Zimmer, H. (1982) What Really Is Turquoise? A Note on the Evolution of Color Terms, Psychological Research, 44: 213-130.

Zld, B., Tth, T. & Tolna, J. (1986): Colour preference: A new approach, Perceptual and Motor Skills, 62: 739-752.

Zollinger, H. (1972) Das Farbensehen des Menschen als Interdisziplinres Forschungsproblem, Palette, 40: 25-31.

Zollinger, H. (1976) A Linguistic Approach to the Cognition of Color Vision in Man, Folia Linguistica, 9:265-293.

Zollinger, H. (1984) Why Just Turquoise? Remarks on the Evolution of Color Terms, Psychological Research, 46: 403-409.

Zrenner, E. (1983) Neurophysiological Aspects of Color Vision in Primates, Berlin: Springer; see also pp. 195-210 in Mollon & Sarpe (1983).

Zrenner, E. (1985) The zero signal detector, in Ottoson & Zeki (1985) 165-182.


* This target article was originally submitted in 1993. Literature appearing since then has not been included systematically. We are grateful to the referees for their thoughtful and constructive criticisms. A special debt is owed to P. Whittle whose intervention prevented a number of misunderstandings.

1. See Saunders 1993 for an account of the history of the concept of 'basic' in Basic Colour Term.

2. For summaries of the various criticisms, evaluation of the Berlin and Kay theory, and further references, see Saunders 1992, 1993, 1995, 1996; Saunders & van Brakel 1988, 1995, 1996; van Brakel 1993.

3. The reviews of Conklin 1973 and Sahlins 1976, although critical, were more sympathetic.

4. Summarising the Rosch experiments 1971, p. 453; 1972b, pp. 15-9, Saunders 1992, pp. 76-7 concludes that it is difficult to see what Rosch means by there being some evidence that the first four Berlin and Kay colours receive some statistical support.

5. Kay (1975) ignores Rosch's assertions and claims that mola and mili are WARM/COOL categories.

6. These experiments also raise the issue of the propriety of transposing laboratory methods into the field and what the status of the Subjects is. For example Rosch 1972a, p. 16 says "the testing was so arranged that only E's hands were visible to the S during testing."

7. Compare Collier 1973, Lucy and Shweder 1979, Garro 1986 on the relation between foci and saturation. Collier's 1976 claims need to be reevaluated.

8. Though it could be said that there are as many psychophysical models as there are colour scientists, the account we present tends to be the one taken over by adjacent disciplines.

9. For three colour cosmologies in African languages, see Jacobson-Widding 1979, Turner 1967, Whitely 1973, Zahan 1974; for four colour cosmologies common in South India, see Beck 1969. For five colour cosmologies see Baxter 1983, Beffa 1978, Gernet 1957 for Chinese, Pritsak 1954 for Altaic and Riley 1963, von Kllay 1939 for Amerindian languages; for the eight colour cosmology of Khmer, see Lewitz 1974, Nepote & Khing 1978.

10. Some of the languages on the Northwest Pacific Coast mentioned by Kinkade 1988 could be interpreted thus (for example Bella Coola, N. Lushootseed, N. Straits Samish, Songish). In particular his reconstructed Proto-Salish would support the interpretation. It dovetails with Boas 1891 who collected 'light blue' terms in the area. See Saunders and van Brakel 1996. Other possible examples of 2 BCTs for blue include Russian (Corbett and Morgan 1988, Morgan & Corbett 1989, Moss et al. 1990), Nepali (Bolton et al. 1980), some Indian languages (Furrell 1885, Rivers 1905), Urban Thai (Wierzbicka 1990), and ancient Greek (Maxwell-Stewart 1981, vol. 2, Irwin 1974, pp. 79-110).

11. Older sources referring to possible yellow-with-green or yellow-with-green-with-blue categories include Connolly 1897, p. 138 on Fanti; Spencer & Gillen 1927, p. 552 on Arunta; Bright 1952 on Karok (Karuk); Almquist 1883 on Chukchi. In recent literature, Forge 1970, p. 283 mentions a pale green powder used by the Abelam of New Guinea which is called yellow or blue; Senft 1987, pp. 329-30 mentions the Kilivila term digadegila yellow, green, blue which he claims shows a "...pattern of confusion." See also references in MacLaury 1991a, Berlin et al. 1991, van Brakel 1994.

12. Compare also Monberg 1971 on Tikopia, Snow 1971 on Samoa, and further references on Polynesian languages in Branstetter 1977.