Genetic Influence on Color Perception: Study on Retinal Photopigments & Vision Genes - Pro, Apuntes de Psicología

¡Descarga Genetic Influence on Color Perception: Study on Retinal Photopigments & Vision Genes - Pro y más Apuntes en PDF de Psicología solo en Docsity! Copyright 2001 Psychonomic Society, Inc. 244 Psychonomic Bulletin & Review 2001, 8 (2), 244-261 The recent growth of molecular genetics research has generated much interest in the relations between genetic potentialities and human behaviors. Although investigat- ing such relations are fraught with many complexities and important ethical considerations, there are some cases in which the linkage between genes and physical realization is relatively straightforward. One such case is that of the genetic basis of human color perception. This paper investigates the linkages between individ- ual’s genetic potential for possessing retinal photopig- ments (or the visual pigments responsible for color per- ception) and individual color perception differences. We begin by introducing some concepts and terminology that will be used throughout this paper. Next, we review recent key findings in molecular genetics research on ret- inal photopigment opsin genes. Related results from color perception psychophysics and color and cognition research are then described. Finally, we describe an analysis of color vision genes and present new results from an experiment that investigated the connection between photopigment opsin genes and color perception. We discuss the results of this new research as related to existing color perception theories, as well as the implications for psychological stud- ies of color processing. To begin, a brief review of the molecular genetics under- lying the biological bases of color vision, including de- fining terms and concepts, is in order. Concepts and terms are simply defined now and are discussed in detail later on. Interested readers can find additional material in Mol- lon (1995), Nathans (1997), M. Neitz and J. Neitz (1998), and Zegura (1997). In the present paper, we investigate photopigment opsin genes, which are simply defined as the genetic sequences responsible for the response properties of the photosen- sitive material (i.e., opsin tuned cis-retinal) in human retinas. Retinal photopigments occupy cone cells in the retina that respond maximally to specific portions of the visible electromagnetic spectrum. Three general classes of photopigments are known to exist: those most sensitive to the short-wavelength region of the spectrum (abbrevi- ated SWS cones), those most sensitive to the middle- wavelength region (MWS cones), and those most sensi- tive to the long-wavelength region (LWS cones). Due to the intricacies of gene expression mechanisms, people who possess the genetic code for the three classes of photosensitive retinal cones may or may not “express” Portions of this research were presented at the 1998 European Con- ference on Visual Perception, the 1998 meeting of the Optical Society of America, and the 1998 meeting of the Psychonomic Society. Partial support was provided by the National Science Foundation (Grant NSF- 9973903 to K.A.J.) and a UCSD Hellman Faculty Award (to K.A.J.). The authors acknowledge the many helpful suggestions made on earlier versions of this manuscript by G. Paramei, L. Hurvich, R. Mausfeld, S. Link, R. M. Boynton, D. I. A. MacLeod, V. Bonnardel, R. G. D’An- drade, N. Alvarado, L. G. Carrera, and K. Goldfarb. Correspondenc e should be addressed to K. A. Jameson, Department of Psychology, Uni- versity of California at San Diego, 9500 Gilman Dr., La Jolla, CA 92093- 0109 (e-mail: kjameson@ucsd.edu) . Richer color experience in observers with multiple photopigment opsin genes KIMBERLY A. JAMESON and SUSAN M. HIGHNOTE University of California at San Diego, La Jolla, California and LINDA M. WASSERMAN University of California at San Diego School of Medicine, La Jolla, California Traditional color vision theory posits that three types of retinal photopigments transduce light into a trivariate neural color code, thereby explaining color-matching behaviors. This principle of trichro- macy is in need of reexamination in view of molecular genetics results suggesting that a substantial percentage of women possess more than three classes of retinal photopigments. At issue is the ques- tion of whether four-photopigment retinas necessarily yield trichromatic color perception. In the pres- ent paper, we review results and theory underlying the accepted photoreceptor-based model of trichro- macy. A review of the psychological literature shows that gender-linked differences in color perception warrant further investigation of retinal photopigment classes and color perception relations. We use genetic analyses to examine an important position in the gene sequence, and we empirically assess and compare the color perception of individuals possessing more than three retinal photopigment genes with those possessing fewer retinal photopigment genes. Women with four-photopigment genotypes are found to perceive significantly more chromatic appearances in comparison with either male or fe- male trichromat controls. We provide a rationale for this previously undetected finding and discuss im- plications for theories of color perception and gender differences in color behavior. RICHER COLOR EXPERIENCE 245 all three classes in their retinas. For example, a person with gene sequences for SWS, MWS, and LWS cone types may express, or physically manifest, only two of those types retinally (e.g., SWS and MWS cones). Thus, each person has a genotype (i.e., the genetic potential, or genes, pre- sent in their DNA) and a phenotype (i.e., the realized manifestation, or expression, of genetic potentialities in their genetic code). Thus, the phenotype need not be a full representation of the genotype. The three classes of photopigments respond to visible light and are generally believed to transmit their signals into a postreceptoral neurally trivariant processing sys- tem (MacLeod, 1985). This defines the three-channel color signal processing called trichromacy , which gov- erns perceptual behaviors such as color matching. Tri- chromacy, then, is a theory of color perception based on three receptor classes that feed into three neural channels and that ultimately produce the rich system we experi- ence as color percepts (see Brindley, 1960, pp. 198–221, for further discussion of the three-receptor hypothesis). A fact that complicates research on this issue is that the only certain method for exactly determining a per- son’s expressed phenotype is to directly examine the retina through “invasive” methods such as microspectro- photometry or in vivo imaging (Roorda & Williams, 1999). Thus, although tests for color vision abnormali- ties exist, are widely used (e.g., anomaloscope color- matching and pseudoisochromatic-plate tests), and are fairly accurate in classifying some phenotypes that arise, they are not designed for determining what cell classes are actually expressed in an individual’s retinas. In any case, enough can be shown to prove that people exist who are trichromats (i.e., with trivariate color vision), who are dichromats (i.e., lacking one of the three standard signals of color experience, also called color-blindness ), and who are anomalous trichromats (with trivariate color vision but with an anomalous shift in the sensitivity in ei- ther the LWS or MWS systems). Shifts in sensitivity found in anomalous trichromats can occur as the result of naturally occurring amino acid mutations (or polymorphisms ) in specific locations (or codons) of the gene sequence of a given retinal photo- pigment class. In such cases, a person may, for example, have normal (or wildtype) SWS and MWS genes but may have a mutated LWS gene (or LWS-photopigment poly- morphism) that produces the shift in that cone class’s re- sponse sensitivity. Polymorphisms of this sort occur rather frequently on MWS and LWS gene sequences and thereby add to the variety of possibly expressed retinal pheno- types in a given population. It is also possible for a person to possess both a poly- morphic variant for a given cone class and the gene se- quence for the wildtype, or normal, variant of that same photopigment class. In effect, such people possess the genetic potential for two variants of the same photopig- ment class, with peak sensitivity at different (albeit very close) spectral frequencies. Such people when female are called heterozygotes , and, below, we assess and review some possible consequences of heterozygote perception. THE BIOLOGICAL BASIS OF TRICHROMACY We start with a review of recent findings relevant to the perception of people possessing more than three classes of retinal cones. The linkage between color perception phenomenology and the neurophysiological basis of color vision has generated an impressive record of psycholog- ical and biological research. The 19th century Young– Helmholtz three-component theory developed the idea that color vision is trichromatic due to the presence of three retinal visual pigments, or “photopigments” (see Brind- ley, 1960). Genetic research showed that color vision is a sex-linked trait, because the genes coding for long- wavelength-sensitive (LWS) and medium-wavelength- sensitive (MWS)1 photopigments are X chromosome in- herited,2 and the genetic sequences associated with these photopigments were isolated (Nathans, Piantanida, Eddy, Shows, & Hogness, 1986; Nathans, Thomas, & Hogness, 1986). Further work determined that genotypes involving more than three photopigment opsin variants are com- mon and that mechanisms governing the expression of such photopigment opsin genes allow for the possibility of an individual expressing more than three retinal pho- topigment types (Dartnall, Bowmaker, & Mollon, 1983; Merbs & Nathans, 1992a; Mollon, 1992, 1995; J. Neitz & Jacobs, 1986; J. Neitz, M. Neitz, & Jacobs, 1993; Winderickx, Battisti, Hibiya, Motulsky, & Deeb, 1993). Recent studies show that these commonly occurring genetic polymorphisms produce variations in spectral tun- ing of expressed photopigments. Such spectral shifts are attributable to amino acid substitutions at specific loca- tions in the opsin gene (Deeb & Motulsky, 1996). The X-linked inheritance feature, when coupled with the pos- sibility of opsin gene polymorphisms, allows for a con- siderable percentage of females to be heterozygous at cer- tain critical amino acid positions for MWS or LWS opsin genes. That is, females who possess two distinct genetic variants at certain codons (with one variant on each of two X chromosomes).3 Previous research proposed that such genetic heterozygosity may have perceptual conse- quences in individuals who actually express all four types of photopigment genes, because each gene type produces different retinal photopigment sensitivities, in effect yield- ing a four-cone-class retinal phenotype (Deeb & Motul- sky, 1996; DeVries, 1948; Jordan & Mollon, 1993; Krill & Beutler, 1965; Mollon, 1992; Nagy, MacLeod, Heyne- man, & Eiser, 1981; Schmidt, 1955).4 A few investigators have conjectured that some indi- viduals who possess four photopigments in their retinas might have a dimension of perceptual experience that is not experienced by trichromat individuals (Deeb & Mo- tulsky, 1996; Mollon, 1992, 1995).5 As discussed by Cohn, Emmerich, and Carlson (1989) heterozygous females fail to be detected by the use of an anomaloscope, although there are reported shifts in their anomaloscope color matches (Crone, 1959; Feig & Ropers, 1978; Krill & Beutler, 1964; Pickford, 1959; Schmidt, 1955), as well as shifts using flicker photometry (Crone, 1959; Ya- 248 JAMESON, HIGHNOTE, AND WASSERMAN pated for cash payment, or were volunteers for research participa- tion. DNA specimens from each subject were analyzed in a manner following existing research using standard procedures described in detail in Appendices A and B. On the basis of blind polymerase chain reaction (PCR) genetic assays, 64 subjects were partitioned into the six genotype groups seen in Table 1. Results Table 1 presents the results from PCR genetic analysis of 64 subjects. These results were further refined by re- sults from standard color vision screening methods using the Ishihara Pseudo-Isochromatic Color Plates and the Farnsworth–Munsell 100-Hue Test.11 Together, the ge- netic analysis and the color vision screening results of our subjects allow the determination of the subject par- titions given in Figure 1. Figure 1 schematically depicts the genotype–phenotype classifications of our sample of 64 subjects. Figure 1 relates gender partitions to identi- fied genotypes and relates genotype to the predicted phe- notype partitions (Figure 1, left column). All predicted phenotype partitions were defined genotypically and, when possible (i.e., for trichromat and dichromat parti- tions only), were also verified using standardized percep- tual tests. Existing color vision screening tests are not ap- propriate, given the heterozygote’s potential for deviation beyond trichromacy, for confirming the heterozygote partition.12 Genetic analyses (summarized in Table 1) and the subject classification scheme (represented in Figure 1) yield the identification of three subject partitions. These are (1) subjects who are likely expressors of four- photopigments in their retinas (i.e., 23 female heterozy- gotes), (2) those likely expressors of three-pigment retinas (i.e., 22 male and 15 female functioning trichromats), and (3) those who may possess the genes required for three- pigment retinas but who, due to a genetic expression event, function as if they possess only two retinal photopigments, as indicated by unequivocal failure in standard color vision screening (i.e., 4 functionally dichromat males). The subjects depicted in Table 1 and Figure 1 also par- ticipated in a color perception experiment. The experi- ment was designed to serve three aims: (1) to go beyond the limitations found in standardized color vision screen- ing, (2) to evade the assumptions of trichromacy inher- ent in methods that primarily isolate metameric equiva- lence mechanisms (e.g., Rayleigh matching), and (3) to make use of contextually rich viewing conditions that in- Table 1 Sixty-Four Subjects Classified by PCR Analyses of Red and Green Gene Codon 180 Amino Acid Sequences Amino Acids Identified Possible Configurations of Photopigment Genotypes Suggested by Retinal Photopigment in Codon 180 Phenotypes of Amino-Acids Amino Acids on Codon 180 of the Likely to Arise Given the PCR Tests on X1 2 X2 or X1 Red and Green Opsin Gene Arrays Codon 180 Results Serine and Alanine (Ser)X1 2 (Ala)X2 XSer: Nrml.R. gene and (a) Nrml.R.+ Ab.R. Female (n 5 23) (Ala)X1 2 (Ser)X2 XAla: Ab.R. gene (b) Nrml.R.+ Nrml.G. or : Nrml.G. gene (c) Nrml.R.+ Ab.R.+ Nrml.G. or : Ab.R. and Nrml.G. gene Alanine Female (Ala)X1 2 (Ala)X2 XAla: Ab.R. gene (d) Ab.R. (n 5 7) or : Nrml.G. gene (e) Nrml.G. or : Ab.R. and Nrml.G. gene (f ) Ab.R.+ Nrml.G. Serine Female (Ser)X1 2 (Ser)X2 XSer: Nrml.R. gene (g) Nrml.R. (n 5 8) Alanine Male (Ala)X1 XAla: Ab.R. gene (h) Ab.R. (n 5 14) or : Nrml.G. gene (i) Nrml.G. or : Ab.R. and Nrml.G. gene ( j) AB.R.+Nrml.G. Serine and Alanine (Ser+Ala)X1 XSer+Ala: Nrml.R. gene and Nrml.G.gene (k) Nrml.R.+ Nrml.G. Male (n 5 5) or : Hybrid Nrml.R. + Nrml.G. gene (l) Nrml.R. or : Hybrid Nrml.R. + AB.R. or : Hybrid Nrml.R. + AB.R. and Nrml.G. gene Serine Male (n 5 7) (Ser)X1 XSer: Nrml.R. gene (m) Nrml.R. Note—Abbreviations as follows: Nrml., normal; AB., abnormal; R., red; G., green. X1 denotes a single X chromosome, and X2 denotes a second X chromosome. Column 1 presents subject types grouped by PCR analysis results at codon 180 of the red and green photopigment opsin genes (described in Appendix B). Column 2 suggests the arrangement of amino acids (serine or alanine) on the available X chromosomes for the males and females tested. Column 3 details the photopigment opsin genes most likely arising from the amino-acid sequences identified on the available chromosomes. Column 4 presents possibly occurring photopigment phenotypes given the present genotype analysis. (Phenotypes in column 4 do not include information on the short-wavelength-, or blue-, sensitive pigment gene.) Appendix A provides details of the genotype classification procedure. Note that only one phenotype listed in column 4 involves the expression of two LWS photopigments plus one MWS photopigment [Case (c) of the serine + alanine female group]. Any of the other possible phenotypes listed either have fewer photopigment classes expressed or have the same number of photopigment classes [(f), ( j), and (k), which, compared with the probability of expressing (c) or (a) of the heterozygote geno- type, are estimated to occur as an expressed phenotype less frequently, given the phenotypes possible within those respective genotypes]. Table 1 presents the most general analysis regarding the codon 180 genotype–phenotype relation. Further complexities arising from gene number, X in- activation, and other expression mechanisms continue to be studied by molecular geneticists and await specification for the opsin genes. Although the subject group partitions characterized in Table 1 will likely become more complex with further analyses of genetic mechanisms (e.g., the spec- ification of detailed differences implied by MWS gene serine 180, the expression of a greater number of pigment genes from a single X chromo- some, or the unresolved expression consequences of chimeric, or hybrid, genes), our crucial assertion that serine + alanine females will exhibit a more diverse opsin genotype and will possess the genetic potential to phenotypically express more kinds of retinal photopigment classes relative to the other subject classes identified here will remain unchanged . RICHER COLOR EXPERIENCE 249 voke color discrimination mechanisms. This color per- ception experiment is now described. A COLOR PERCEPTION EXPERIMENT USING A DIFFRACTED SPECTRUM STIMULUS A spectrum is typically characterized as the compo- nent bands of light produced when sunlight is passed through a prism. In 1664, Sir Isaac Newton, followed by Hermann von Helmholtz in 1867, delineated component bands of colored light, or spectral chromatic bands, by drawing lines perpendicularly through spectra. Newton specified seven chromatic delineations in the spectrum (red, orange, yellow, green, blue, indigo, and violet; as referenced in Shapiro, 1984), whereas Helmholtz iden- tified only four color sections (red, green, blue, and pur- ple) in a prismatic spectrum (Campbell, 1986).13 To our knowledge, there are no existing theories that provide behavioral predictions for Newton’s task of de- lineating chromatic bands perceived in the spectrum. Nor is there much discussion in the color perception lit- erature of why a person with normal color perception would identify seven bands versus some other greater or lesser number of chromatic bands. When one views a spectral stimulus, one perceives a continuous gradation of color from one end of the spectrum to the other. There is nothing inherent either in the spectrum or in the human perception of it that would compel the identification of the seven chromatic divisions originally identified by Newton or into any other number of divisions found by any subsequent researcher. The behavioral task used in the present experiment re- sembles methods used in the early work of Newton and Helmholtz. The stimulus used in the present experiment is a modif ication of that used in more recent work (Smeulders, Campbell, & Andrews, 1994) which repre- sents a modern-day consideration of Newton’s study of spectral chromatic appearances. Smeulders et al. used a diffracted spectral stimulus in a study of 4 “normal” color vision adults and 1 “protanomalous adult (so-called red- blind)” (p. 928). Their task required subjects to identify and delineate bands (i.e., indicate beginning and end points) of colored light in a projected spectrum. In par- ticular, they examined whether progressively more bands were perceived with increasing delineations, and inves- tigated the limits in the number of bands perceived in the subdivided spectrum. Smeulders et al. provided a mod- ern test of the validity of the spectral delineation task in- troduced by Newton. They did not investigate the link- age between photopigment genotypes and chromatic delineations in a spectrum, which is the focus of the pres- ent study. Method Subjects. Sixty-four University of California, San Diego, subjects previously described in the genetic study of photopigment opsin genes participated in the perceptual experiment now described. During the perceptual assessment portion of this study, both sub- jects and experimenters were naive to the genotyping and pheno- typing analysis presented earlier in Table 1 and Figure 1. Materials . The subjects’ task was to provide spectral delineation s for a diffracted spectral stimulus subjectively experienced as a self- luminous gradient ordering of chromatic components. Figure 2 pre- sents a schematic of the experimental apparatus used consisting of 10 simple optical components . Apparatus. Figure 2 schematically depicts the configuration of the experimental apparatus. The light source employed was a 500- W halogen illuminant with a broad spectral power distribution ex- tending to the 380 to 780-nm spectral range, with a component well into the long-wavelength end of the spectrum. All optical equip- ment was obscured by blackout material, which also eliminated all scattered source light. Measured luminance of the display was 36 lumens/m 2 (very roughly approximating half the brightness of a normal desktop computer display viewed in a well-lit ambient) or 0.4 lumens for the dimensions of our projected stimulus. The dif- fracted spectrum stimulus was back-projected onto a clear lucite panel mounted in a black 6 3 5 foot rigid display. The subject sat in front of the fixed panel and, for each trial, made responses by drawing on the front of the display where translucent tracing paper was mounted. The projected stimulus viewed by the subjects sub- tended approximately 44º horizonta l 3 21º vertical visual angle. Experimental procedure. The experiment was self-paced, av- eraging approximately 1.5 h in duration. Prior to participation, all subjects were fully dark-adapted for a minimum of 15 min. Al- though the experiment was conducted in dark ambient conditions , the stimulus luminance was well above the rod intrusion threshold for all subjects. The projected diffracted spectrum stimulus was viewed binocularly and was experienced by the subjects as a lumi- nous “rainbow” spanning a horizontal gradient with violet on one extreme end and red on the other extreme end. A variety of task instructions were used in the experiment to as- sess different forms of information regarding the subject’s percept. Figure 1. Subject partitions used to analyze the perceptual data. The right column shows the gender distribution of our sam- ple. The middle column shows the distribution of six subject groups (asterisks denote subjects found to be functionally classi- fied as red-abnormal by the Farnsworth–Munsell 100-Hue Test and Ishihara Pseudo-Isochromatic Plates, 11th ed.). The left col- umn shows the predicted phenotypes based on genotype and color vision assessment. Analysis of the behavioral data uses the clas- sification scheme shown in the left column. Genotypes presented in the middle column were determined by PCR genetic assay de- scribed in Appendix B. 250 JAMESON, HIGHNOTE, AND WASSERMAN For example, concerning the width of the subject’s percept (or per- ceived “gamut”), two separate tasks instructed the subject to “Mark the Right [or Left] extreme edge of the rainbow.” Regarding the num- ber and location of distinct chromatic bands (a.k.a. “delineations ”) the subjects responded to eight different task instructions, including “Mark all the edges for all the bands of color in the rainbow,” and “Starting from the Left [or Right] side of the display carefully count to the 1st [or 2nd, . . . , or 7th] band of color and mark both edges of the 1st [or 2nd, . . . , or 7th] band of color.” Finally, seven different instructions regarding to the location of best-exemplar appearance s were used: “Indicate by a single mark the position of the Best Ex- ample of Violet [or Blue, Green, Yellow, Orange, Red, or Purple] in the rainbow.” Data from three types of judgments are presented here: (1) the task in which the subjects were instructed to demark “all the edges for all the bands of color in the rainbow” (hereafter called “delin- eation data”), (2) the task in which the subjects indicated the loca- tion of the best-exemplar appearance for several color categories (called “best-exemplar data”), and (3) the task in which the subjects were instructed to “mark the Right [or Left] extreme edge of the rainbow” (called “perceived gamut” data). Each subject’s delinea- tion data were based on six repetitions of this task. Each subject’s best-exemplar locations were based on three judgments per each color category. Each subject’s perceived gamut data were based on three judgments per edge. All judgments were randomly presented within the experiment’s 80 total trials. All subjects were given a different random order of stimuli. Each trial sequence was terminated by actuating a photographic flash, which created an energy mask designed to eliminate stimulus after- images from trial to trial. Between trials, the stimulus was shielded from view by a closed aperture, and the room ambient remained dark while the experimenter (1) set up the response sheet for the next trial, (2) established a calibration mark for the trial, and (3) ex- plained the next task to the subject. Each trial required an estimated ~10–30 sec to complete, with an estimated intertrial interval of ~20–60 sec. The physical attributes of the projected display were constant from trial to trial; however, in debriefing, some subjects re- ported uncertainty regarding whether stimulus manipulations were made across trials. Following the experimental session, subjects were screened for color vision deficiencies using the Farnsworth– Munsell 100–Hue Test and Ishihara Pseudo-Isochromatic Color Plates (11th ed.). We assume that the ability to perceive and delineate bands of chromatic difference along the spectrum is a function of the detec- tion and discrimination of noticeable differences in spectral wave- lengths (cf. Boynton, Schafer, & Neun, 1964; Smeulders et al., 1994). Thus, the relation between photopigment genotype and performance in the spectral delineation task is hypothesized to be that increases in classes of expressed photopigments impact detection and discrim- ination in such a way as to produce increases in the number of chro- matic bands perceived and delineated in the diffracted spectrum. Regarding a subject’s placement of best-exemplar locations in the spectrum, we made the conservative prediction that the data would be comparable to the spectral locations of unitary hues investigate d in previous studies (Boynton et al., 1964; Dimmick & Hubbard, 1939; Purdy, 1931; Westphal, 1910, cited in Boring, 1942). The ra- tionale for this was the following: First, because unitary-hue (or best-exemplar) locations in anomalous trichromats and “normal” trichromats were previously found to be similar for the percepts of blue, green, and yellow, we expected our subjects’ data to display a similar agreement. Second, we expected to observe substantial in- dividual variation in best-exemplar locations within all of our sub- ject partitions. Third, we expected that individual variation in some cases would exceed the differences observed between our groups’ average locations. Thus, we predicted reasonable correspondence s between all our subjects’ best-exemplar locations and the spectral locations of unitary hues found in existing research. Data analysis. As mentioned above, the present analyses exam- ine for each subject (1) the total number of bands delineated in the diffracted spectrum display (six repeated observations per subject) , (2) the locations of the best example of the individual colors red, or- ange, yellow, green, blue, violet,14 and purple (three repeated ob- servations per color for each subject), and (3) the locations of each extreme edge perceived in the stimulus (three separate observation s per edge, left and right extreme, per subject). All subjects’ responses were processed blind and verified for measurement and recording accuracy by two independent experimenters. Unless otherwise stated, all tests of significance reported use two-tailed Student’s t tests for unequal samples. All reported tests of significance on mean mea- sures were also verified by appropriate tests on medians. Results Delineation of spectral chromatic bands. Our first hypothesis was that the subjects expressing more than three photopigments would have a different phenome- nological color experience. This could be demonstrated in the ways they segment the chromatic components of the diffracted spectrum. Thus, people with four photo- pigments are expected to experience more chromatic “bands” in the rainbow than are trichromat or dichromat individuals. Similarly, trichromat individuals should ex- perience more bands than should dichromats. With respect to behavioral measures, the general prediction is that the genetic potential to express more than three photopig- Figure 2. Experiment 1 apparatus and materials. The light source employed was a 500-W halogen lamp (1) with a broad spectrum energy component extending beyond the 400 to 700-nm range and with a substantial component in the long-wavelength end of the spectrum. A rectangular aperture (2) was used as an image plane for collimating the source, and an iris (3) was used to minimize scattered light. Collimated light from lens 4 impinged on a blazed diffraction grating (6) and was controlled by an in- tervening switch activated aperture (5). Lenses (7 and 8) were used to form a magnified image of the diffracted spectral image reflected off the grating (6). This image was back-projected onto the Lucite panel (9) and viewed binocularly by subjects. Inter- trial energy masking was achieved using a photography flash (10). The entire apparatus depicted in this figure was configured within a 3 3 3 foot area (drawing is not to scale). RICHER COLOR EXPERIENCE 253 experimental stimulus. Diffraction (as opposed to re- fraction) of the spectrum permits the approximation of wavelength by a regression line of measured wavelength against 25 metric distances equally spaced across the stimulus range. Comparing the density and location of symbols across the top and bottom panels illustrates the manner with which the heterozygotes differ from the fe- male trichromats in their chromatic banding behavior: In general, heterozygotes perceived more delineations in the spectrum and exhibited finer grained discrimination differences in the interval between approximately 580 and 780 nm. Figure 3 also compares the responses of the subjects with Newton’s observation of seven perceptible chromatic delineations in the spectrum. Among the heterozygotes, 21 (91%) identified a median number of delineations of seven or more bands, whereas 8 (53%) trichromat females identified a median number of delineations of seven or more bands. Regarding Newton’s hypothesized seven chromatic bands, 4 heterozygotes and 2 trichromat females exhibited a median number of bands equal to seven. A univariate analysis of variance for between-subjects effects of the width (as the dependent variable) of each subject’s spectral range showed that the tested subject groups did not differ significantly regarding the individ- ually perceived width of the diffracted spectrum [F(3,64) 5 0.465, p 5 .79]. This finding excludes the possibility that increases in banding are simply attributable to an in- crease in perceived width of the spectrum by individual heterozygotes. Spectral location of best-exemplar appearances. Our second hypothesis was that the subjects putatively expressing more than three photopigments would not dif- fer significantly from the “normal” trichromat subjects re- garding the placement of best-exemplar locations for red, orange, yellow, green, blue, violet, and purple appearances in the diffracted spectrum. Thus, we expected that our heterozygous females would locate the above-mentioned appearances in spectral locations similar to the locations given by the trichromat females and males and that said locations would agree with existing data on the location of unitary hue experiences in the spectrum (Boynton et al., 1964; Dimmick & Hubbard, 1939; Purdy, 1931; West- phal, 1910, cited in Boring, 1942). Figure 4 presents the data for the heterozygous females and the trichromat females regarding the placement of best-exemplar appearances for the tested hue categories (red, orange, yellow, green, blue, violet, and purple).15 In Figure 4, each horizontal series of symbols represents a single subject’s data for the placement of violet, purple, blue, green, yellow, orange, and red bands. Data for the female trichromats (n 5 15) are displayed in the bottom panel, and data for the female heterozygotes (n 5 23) are displayed in the top panel. The data from both groups Figure 4. Best-exemplar locations for trichromat and heterozygote females. Similar to Figure 3’s representa- tion, individual subjects from each group are represented horizontally as a series of position locations and are sep- arately grouped as trichromats and heterozygotes within the two presented panels. Note that a comparable de- gree of individual variability in the placement of best exemplars is seen across groups. Individual best-exemplar variability is conveyed by the horizontal spread of points seen in the data columns that define best-exemplar re- gions for violet, purple, blue, green, yellow, orange, and red (from left to right). Groupwise averages for the place- ment of best-exemplar locations (represented by the red symbols) are largely in agreement across the two groups (error bars for the groupwise averages represent 1 standard deviation). 254 JAMESON, HIGHNOTE, AND WASSERMAN lend support to existing normative data on the spectral location of hue–wavelength association (presumably based on purely trichromatic samples) in that the consensus on the locations of blue, green, and yellow were, respec- tively, approximations to 481, 539, and 590 nm.16 In Fig- ure 4, there is also agreement between both partitions of female subjects for the average location of the best ex- emplars of all the tested appearances. That is, both groups’ best-exemplar locations were within one standard devi- ation of other groups’ average best-exemplar locations.17 In addition, concerning the location of best-exemplar positions, we found fairly substantial and equivalent amounts of individual variation within each of the sub- ject groups represented. This individual variation was similar for all colors tested, and it occurred to a similar degree across all groups. Further implications of this finding for the general determination of best-exemplar hues and unitary-hue spectral locations will be discussed in a forthcoming paper. The main result of the best-exemplar location analysis is that, consistent with our prediction, on average the two groups of females we assessed similarly located the po- sitions of hue best exemplars in the diffracted spectrum. This finding indirectly lends confidence to the chromatic banding result presented above and provides another possible clue as to why previous investigations of het- erozygote perception did not find color perception dif- ferences similar to those we found here in comparisons of female heteorzygotes with female trichromat controls. That is, the heterozygotes’ richer color experience may have gone undetected in previous normative research, be- cause (at least for the case of best-exemplar data), aggre- gating heterozygote data can yield results that resemble that found for trichromat subjects. If the subjects’ geno- type information were not readily available, one may be un- able to differentiate these two groups strictly on the basis of their spectral positions of unitary-hue percepts. An issue of further interest is what such heterozygous color perception differences would contribute to existing nor- mative trichromatic standards (e.g., Vl). We now discuss our results in the context of color perception theories. DISCUSSION The results presented above indicate that, in the pres- ent study, female subjects who most likely express four photopigments in their retinas experience a different per- cept, in comparison with females who most likely ex- press three photopigments. We believe that this kind of in- creased color-differentiation behavior in heterozygote females has gone undetected because of the empirical methods used in previous research. In contrast to methods previously used to assess the putative four-photopigment vision of heterozygotes, our experiment used binocular viewing of a contextually complex stimulus consisting of a chromatic gradient of heterogeneous luminance. Al- though the stimulus configuration used here, due to the need for empirical control and manipulation, by no means achieved the viewing complexity present in a “natural” scene, it was clearly a substantial step closer to real-world viewing complexity when compared with a Rayleigh match stimulus configuration.18 In the present psycho- physical task, the subjects were asked to judge a rela- tively complex percept. In essence, visual processing in the present study most likely required the use of addi- tional perceptual mechanisms beyond those required by a classical color-matching task. Although added stimu- lus complexity increases the number of potential expla- nations for the variation observed, only explanations ul- timately based on the serine–180–alanine substitution in photopigment opsin genes (discussed in Appendix A) can explain why, in these data, increased spectral delin- eations should occur only for heterozygous females. The photopigment sensitivity curves produced by the serine and alanine opsin variants are believed to be iden- tically gaussian-shaped with peak response sensitivities differing by 4–7 nm (Asenjo, Rim, & Oprian, 1994; Mol- lon, 1995; M. Neitz, J. Neitz, & Jacobs, 1995). Although 4 nm may seem a minor difference, in many instances, small perturbations in a physical or biological system re- sults in substantial consequences: For example, peak re- sponses of MWS and LWS photopigments differ by an estimated 20–30 nm, but their associated percepts are the distinctly different sensations of green and red (MacLeod & von der Twer, 2000). Thus, a reasonable as- sertion is that the serine/alanine 4- to 7-nm difference in photopigment sensitivity could produce a phenomeno- logical effect of two perceptually distinguishable reddish appearances (Mollon, 1992). Jameson and Hurvich (1956) originally developed the now well-established color-opponency theory, which states that small fixed shifts in photoreceptor spectral sen- sitivity lead directly to systematic alterations in spectral response functions for the paired chromatic and achro- matic opponent response systems. This model has given predictions that agree with data on the color perceptions and discriminations of anomalous trichromats and dichro- mat observers (Jameson & Hurvich, 1956). We believe an extension of the theory may similarly serve to describe the perception and discrimination mechanisms govern- ing the color phenomenology of four-pigment observers. There has been considerable study of the relationship between perceived color and spectral wavelength, and, although there are many variables that effect the appear- ance of spectral wavelength, the general notion is that “the relation between the wavelength of spectral radiant energy and perceived hue is so well known that it is common-place to talk about [spectral] light as if it were colored” (Boynton et al., 1964). Even so, there are no ex- isting theories in the color perception literature that de- scribe why specific chromatic bands are necessarily per- ceived by an observer viewing a spectrum. Nor is there an explanation as to why some colors that are associated with wavelength are found to be distinguishable at one location of the spectral continuum but not present at an- other nearby location. RICHER COLOR EXPERIENCE 255 The classical three-receptor model of color vision ex- plains that there are three retinal pigments that maxi- mally respond at three different spectral frequencies, and these peak sensitivities are associated with spectral re- gions subjectively described as reddish, greenish, and bluish in hue.19 Additionally incorporated in the estab- lished view is a zone theory of color vision (i.e., G. E. Mueller’s as discussed in Wyszecki & Stiles, 1982, pp. 634– 639) that provides the mechanistic basis for the location of hues aqua, green, yellow, and orange in the spectrum (presented by Thomson, 1954). Taken together, such a three-receptor-based stage theory (or some variant of it) is generally shared and accepted by color vision scien- tists as the model underlying color vision processing. However, beyond this, we need further explanatory the- ory to understand other hue–wavelength relations widely seen in spectra by trichromats (e.g., violet) and hetero- zygotes (e.g., violet, magenta, burgundy, and salmon). Existing color vision models do not provide an adequate account of the mechanisms underlying perceptual salience for these additional percepts, nor do they for- mally predict their spectral locations. We rule out possible color-semantic explanations for these findings and emphasize again that our results are a refutation of the notion that the normative seven chro- matic spectral bands (e.g., red, orange, yellow, green, blue, indigo, and violet) are simply culturally acquired, or perhaps are socialized conventions arising from a “basic” color nomenclature. This is seen in that we found that the two groups of subjects who are most likely to be similarly socialized (i.e., females of two different, so- cially undetectable, genetically based phenotypes) showed significantly different behaviors in our experimental task, whereas two groups of subjects who are most likely to have different socialization experiences (i.e., males and females of trichromat phenotype who receive gender- specific socializations) actually exhibited behaviors that were not significantly different in our task. The question specifically evaluated by this study is what is at the basis of the observed differences in how in- dividuals perceive hues associated with spectral wave- length, and, in particular, what model will help explain the results found in the present study for the different genotype–phenotype groups we have discussed. Because the accepted explanations of the red, green, and blue spectral percepts have historically been receptor- based, it makes sense to first extend the receptor-based stage model explanation to account for other perceived spectral bands. To begin with, we could adopt the idea that chromatic appearances are based on the separation or proximity of spectral response-function peaks (De- Marco, Pokorny, & Smith, 1992; J. Neitz & M. Neitz, 1994; Piantanida, 1976; Pokorny & Smith, 1977, 1982; Regan, Reffin, & Mollon, 1994). This idea states that pigments separated by only 2–3 nm in effect increase the range of color matching relative to greater pigment sep- aration (Regan et al., 1994). This construct, when com- bined with the additional input of a fourth photopigment and an added mechanism that accounts for the amount of the total overlap contributed by all available photorecep- tor systems at a given wavelength location, would begin to capture the observed increases in discrimination ability as the number of cone classes increases. Such a model could explain why heterozygotes with four overlapping classes of photosensitive cells (some of which are sepa- rated by 4–7 nm) might experience a finer color discrim- ination ability in certain regions relative to female tri- chromats. Such an “area-overlap” mechanism additionally incorporated as a stage in existing zone theories would also account for the pattern of results shown here between all the groups assessed—that is, data from our heterozy- gotes, “normal” trichromat females and males, and di- chromat males. In the present paper, we refrain from elaborating on the details of this photopigment area- overlap model. Future empirical testing should allow a more detailed description of this idea in a forthcoming report. One further modeling improvement supposes that the additional photopigment serves two different functions. First, the fourth pigment may possibly be serving as a normal pigment variation that is integrated into a standard three-dimensional structure of metameric classes (thus, heterozygotes with four-photopigment classes expressed are not discernible on the basis of color-matching mea- sures). Second, the fourth pigment may also feed some higher order (probably cortical) mechanisms that take ad- vantage of the signal from the fourth pigment for color dis- crimination in a way that differs from judging metameric class equivalences. This second possibility is not unlikely since, in other species, wavelength information can be used for various perceptual tasks quite independently from the structure of metameric color codes. That is, color vision and wavelength-dependent behavior can indepen- dently coexist in the same animal (e.g., bees exhibit tri- chromatic color vision exclusively in feeding and in rec- ognition of the hive, while, at the same time, using spectrally narrow-band receptor channels for a variety of other tasks, such as celestial orientation and navigation; see Menzel, 1985, for a discussion). Thus, one might ven- ture the speculation that the present results indicate that the fourth pigment—though it seems to merge with trichromatic equivalence classes of primary codes—is ex- ploited by independent mechanisms that deal with the dis- crimination of border discontinuity and the like. Such ar- chitectural implications very much resemble those from studies on cortical color blindness, or achromatopsia (Stoerig, 1998; Troscianko et al., 1996). Our results demonstrate that, under experimental con- ditions using contextually rich viewing circumstances, the heterozygote phenomenological color experience differs significantly from that of the trichromat. What this suggests at the level of theory is that perhaps a re- formulation of the model underlying color perception is warranted. Ideally, the reformulation should aim to inte- grate two apparently different kinds of color perception into a richer more complex model of visual mechanisms. Such a theory might account for the present findings by supposing two different modes of color vision: one em- 258 JAMESON, HIGHNOTE, AND WASSERMAN pigments in the human retina: Correlations between deduced protein sequences and psychophysically measured spectral sensitivities. Jour- nal of Neuroscience, 18, 10053-10069 . Sharpe, L. T., Stockman, A., Knau, H., & Jaegle, H. (1998). Macular pigment densities derived from central and peripheral spectral sensi- tivity differences. Vision Research, 38, 3233-3239 . Sjoberg, S. A., Neitz, M., Balding, S. D., & Neitz, J. (1998). L-cone pigment genes expressed in normal colour vision. Vision Research, 38, 3213-3219 . Smeulders, N., Campbell, F. W., & Andrews, P. R. (1994). The role of delineation and spatial frequency in the perception of the colors of the spectrum. Vision Research, 34, 927-936. Stockman, A., & Sharpe, L. T. (1998). Human cone spectral sensitivi- ties: A progress report. Vision Research, 38, 3193-3206 . Stoerig, P. (1998). Wavelength information processing versus color perception: Evidence from blindsight and color-blind sight. In W. G. K. Backhaus, R. Kliegl, & J. S. Werner (Eds.), Color vision: Perspec- tives from different disciplines (pp. 131-147). New York: Walter de Gruyter. Swaringen, S., Layman, S., & Wilson, A. (1978). Sex differences in color naming. Perceptual & Motor Skills, 47, 440-442. Thomas, L. L., Curtis, A. J., & Bolton, R. (1978). Sex differences in elicited color lexicon size. Perceptual & Motor Skills, 47, 77-78. Thomson, L. C. (1954). Sensations aroused by monochromatic stimuli and their prediction. Optical Acta, 1, 93-102. Troscianko, T., Davidoff, J., Humphreys, G., Landis, T., Fahle, M., Greenlee, M., Brugger, P., & Phillips, W. (1996). Human colour discrimination based on a non-parvocellular pathway. Current Biol- ogy, 6, 200-210. Wilder, D. G. (1970). The photopic spectral sensitivity of color normal, protanopic and deuteronopic observers. Unpublished doctoral dis- sertation, University of California, Los Angeles. Winderickx, J., Battisti, L., Hibiya, Y., Motulsky, A. G., & Deeb, S. S. (1993). Haplotype diversity in the human red and green opsin genes: Evidence for frequent sequence exchange in exon 3. Human Molecular Genetics Polymorphism, 2, 1413-1421 . Winderickx, J., Lindsey, D. T., Sanocki, E., Teller, D. Y., Motul- sky, A. G., & Deeb, S. S. (1992). Polymorphism in red photopigment underlies variation in color matching. Nature, 356, 431-433. Wyszecki, G., & Stiles, W. S. (1982). Color science: Concepts and methods, quantitative data and formulae (2nd ed.). New York: Wiley. Yasuma, T., Tokuda, H., & Ichikawa, H. (1984). Abnormalities of cone photopigments in genetic carriers of protanomaly. Archives of Ophthalmology , 102, 897-900. Zegura, S. L. (1997). Genes, opsins, neurons, and color categories: Clos- ing the gaps. In C. L. Hardin & L. Maffi (Eds.), Color categories in thought and language (pp. 283-292). Cambridge: Cambridge Uni- versity Press. NOTES 1. Here, we use the accepted terminology of color vision science: long-wavelength-sensitive (or LWS) and medium-wavelength-sensitive (or MWS) photoreceptors to denote retinal cones maximally sensitive to the long-wavelength portion of the visible spectrum and medium- wavelength portion of spectrum, respectively. By convention, the terms used in genetics research for photopigment opsin genes are red and green for LWS and MWS, respectively. 2. The gene corresponding to the short-wavelength-sensitive, or bluish-sensitive, photoreceptor is located on chromosome 7 and is ac- quired through autosomal inheritance. Anomalies of this gene are rare and will not be discussed in this paper. 3. The percentage of females in the general population with hetero- zygous photopigment opsin genotypes is currently estimated between 15% (Jordan & Mollon, 1993) and 50% (M. Neitz & J. Neitz, 1998). Among these, 15% –20% of heterozygous females possess dimor- phisms in locations of the gene array which are known to impact ob- servers’ spectral sensitivities. It is yet unclear whether the cumulative impact of those polymorphisms occuring in less crucial locations of the gene array has a significant impact on the spectral sensitivity of the het- erozygous females who possess them. 4. Although phenotypes expressing more than four classes of photo- pigments are possible, we will limit our discussion to the simplest case involving people expressing four photopigments. The phrase four- photopigment retinas is used throughout this paper to imply four dif- ferent classes of photopigments potentially expressed in individuals’ retinal cone cells. (The input of the scotopic, or rod, system is not at issue in the discussion presented here.) Genetic mechanisms contribut- ing to phenotypic expression of more than three retinal photopigment s is discussed elsewhere (e.g., M. Neitz & J. Neitz, 1998). 5. Note that trichromacy is a composite system that, when maxi- mized, consists of the output of three retinal photopigment classes fed into three separate neurophysiological color channels. Individuals with four-photopigment retinas would function as trichromats if, during the process of producing sensations, the visual system reduced the input from the four-photopigment classes into a three-channel system. This four-photopigmen t/three-channel model has generally been the accepted theory among vision researchers as the most plausible theory for four- photopigment retinal processing (MacLeod, 1985; Miyahara, Pokorny, Smith, Baron, & Baron, 1998; Nagy et al., 1981; M. Neitz & J. Neitz, 1998). The term tetrachromat implies an extension of the three-channel system and is appropriate only for four-photopigment individuals if it were demonstrated that their retinas with four-photopigment classes were feeding into a greater-than-three-channel system and producing a higher dimensional perceptual experience, relative to the trichromat percept. Until such a higher dimensional percept is demonstrated, only the term four-photopigment individuals is warranted for individuals with a four-photopigment class retinal phenotype . 6. Another term for heterozygote females is “female carriers of color abnormality,” or “carriers of color-blindness ” (Jordan & Mollon, 1993; Mollon, 1995; Nagy et al., 1981). Typically, female carriers of color ab- normality are genetically heterozygous for red–green deficiency. That is, when genetically tested, they usually carry genes for both normal and ab- normal photopigments (for one photopigment class) as seen in geneti- cally identified heterozygotes. (Prior to 1985, the manner of identifying carriers of color abnormality was identified indirectly through their color deficient relatives, whereas heterozygotes are identified directly via ge- netic test.) Female carriers of color abnormality also most likely express three normal pigments in addition to an abnormal shifted retinal pigment. 7. In effect, providing metameric, or phenomenologically matched, equivalences. 8. Interestingly, in Maloney’s (1992) formulation, accounting for empirical surface reflectances requires (at least) three degrees of free- dom in the surface reflectance functions, which requires four types of photoreceptor parameters in the algorithm. A similar conclusion fol- lows from that discussed by Boker (1997). 9. The amount and range of individual variation are nicely illus- trated by Judd (1945, Figure 2). 10. All behavioral data were collected using experimenters and sub- jects who were naive to the subjects’ underlying photopigment opsin genotypes. To simplify the present discussion, however, we present the genetic assay analyses first, followed by the methods and results of the empirical test of subjects’ color perception. 11. Note that standard tests available for assessing color vision ab- normalities (e.g., Ishihara Pseudo-Isochromatic Plates and Munsell– Farnsworth 100-hue test) are specifically designed to detect deviations from normal color vision that arise due to either (1) the lack of a photo- pigment opsin class of receptors or (2) the shift of one or more of the three photoreceptor classes presumed to exist (or some combination of both 1 and 2). These standard tests were not designed with the assess- ment of four-photopigment individuals in mind, and, for this reason, they are most likely inconclusive measures of the color perception abil- ities of four-photopigment retina individuals (an assertion actually sup- ported in the assessment of our subjects). Along these lines, Buckalew and Buckalew (1989), in a sample of 57 women, found an unexpectedly high incidence of female color deficiency (3% vs. the expected 0.2%) using Ishihara plates. It is possible that some of these are expressing four retinal photopigments . Moreover, Cohn et al. (1989) reported that, although heterozygotes are not generally detected by pseudo-isochromati c plates, they observed that when such plates are used under conditions that make the task more difficult (i.e., modification of the spectral profile of the illuminant), the performance of the heterozygote is impaired to a greater RICHER COLOR EXPERIENCE 259 degree than that of normal controls (p. 256). These findings are consis- tent with our suggestion, and accord with our data, that such color vision screening tests may erroneously identify four-pigment females as color deficient due to the theory used to construct the screening stimuli. 12. It is essential to note that, although the group of four-pigment het- erozygotes in Figure 1 is strictly a genotype classification, no individ- ual in the trichromat or dichromat partition possess the heterozygotes ’ genetic potential for expressing four photopigments. The overwhelming implication from existing genetics data on nonhuman primates and human retinas (e.g., Kraft, J. Neitz, & M. Neitz, 1998) is that a geno- typic female heterozygote will very likely be an expressor of a four- photopigment retina. However, this cannot be verified in our subjects be- cause physical access to their retinas was not possible; nor is it verifiable through any standard color vision screening test. By comparison, the trichromats and dichromats groups in Figure 1 have both (1) been shown by genetic analysis to have three photopigment genotypes (including anomalous genotypes that are known to produce, with a high probabil- ity, functionally color-blind males) and (2) been shown by standard color vision screening tests to be functionally trichromats and dichromats, re- spectively. Thus, the analysis in Figure 1 makes use of all available and appropriate measures for phenotype determination and necessarily im- poses the mixed genotype–phenotype nomenclature four-photoigment heterozygotes, trichromats, and dichromats for our three subject groups. 13. A prismatic spectrum refers to the component rays of light pro- duced by refraction when a broadband source, such as sunlight, is passed through a prism. A more precise way of creating a spectrum of compo- nent rays of light is by using diffraction rather than refraction. The two methods give rise to essentially the same end-product stimulus (i.e., a spectral stimulus appearing as a self-luminous rainbow-like gradient with colors ordered red, orange, yellow, green, blue, etc.), with the ex- ception that, relative to prismatic refraction, the diffracted spectrum has the advantages of being linearly related to wavelength by a relationship precisely specified by the diffraction grating parameters. 14. Violet was tested rather than Newton’s indigo because it was pre- sumed that the latter color category might not be as widely known among college undergraduates . 15. Figure 4 uses the same metric scale distances computed and rep- resented in Figure 3 described earlier. 16. Spectral locations of the additional chromatic appearances tested here (i.e., orange, violet, and purple) have not been widely studied in color vision research (see Dimmick & Hubbard, 1939, p. 245). 17. Although the data from our male subject groups are not repre- sented in Figure 4, it is worth mentioning that the trichromat males in general located best exemplars in similar positions as to those located by the two female groups presented and also supported existing data on the spectral locations of unitary blue, green, and yellow. Our dichromat subjects’ best-exemplar location data differed from those of the other tested subjects in ways one would expect given the color vision deficien- cies experienced by the subjects in the dichromat group. 18. That is, the stimulus used here is sufficiently rich to activate in- ternal mechanisms that are relatively closer to the mechanisms involved in viewing real-world scenes. By comparison, these mechanisms are si- lent in the Rayleigh bipartite field viewing condition. An aspect that distinguishes our stimuli from the Rayleigh match situation is that our visual scene fulfills an important requirement for the activation of mech- anisms that underlie the perception of real-world scenes—that is, a suf- ficiently high variation with respect to chromatic and luminance codes (as discussed by Mausfeld, 1998, pp. 223–224). There is ample evi- dence that core mechanisms of color perception are activated only if there is sufficient articulation that requires (at least) three degrees of freedom in surface reflectance functions implemented by Maloney (1992) using an algorithm, which, interestingly, requires four types of receptor pa- rameters in the “surface complexity condition.” 19. Allowing for the fact that each triple of distinct hue–wavelength sensations experienced by a given trichromat individual will vary some- what from that of any other trichromat individual. APPENDIX A Appendix A first describes the rationale for the genetic analy- sis component of this study and then describes the genotype variants and possible phenotypes arising when amino acid sub- stitutions occur on codon 180 of the MWS (green) and LWS (red) gene sequences. Genetic Structure Underlying Visual Pigment Variation Because the molecular structures of the LWS and MWS vi- sual pigment gene bases are 98% identical to each other, there are relatively few locations in the amino acid sequences where the LWS and MWS genes can be differentiated by a genetic test (Asenjo et al., 1994; Nathans, Piantanida, et al., 1986; Nathans, Thomas, & Hogness, 1986). For the LWS and MWS pigment genes, these include only 18 variable amino acid positions (see Jacobs, 1998, p. 2208). Substitutions of amino acids caused by differences in the genetic sequence can eventually produce dif- ferent spectral absorption properties in retinal cones. These “di- morphic” or “polymorphic” sites are locations in the gene se- quence where two different amino acids can be alternatively present. Among these 18 sites of variation, only 7 (at codons 116, 180, 230, 233, 277, 285, and 309) involve amino acid substitu- tions that produce shifts in spectral absorption of the X-linked photopigment opsins (Deeb & Motulsky, 1996; Jacobs, 1998).A1 One position in particular, position 180 on exon 3 of the LWS and MWS genes, is one of three positions in the gene sequence (along with codons 277 and 285 at which LWS and MWS genes can be uniquely distinguished) where a single amino acid sub- stitution determines a major portion of the spectral shift between LWS and MWS color vision pigments (Asenjo et al., 1994). This is exemplified by the fact that amino acid substitutions at codons 180, 277, and 285 can eventually transform a red-sensitive retinal photoreceptor into a green-sensitive photoreceptor, and vice versa. In the present study, we specifically examined this important codon (serine–180–alanine) of the gene sequence because this position has the following properties: (1) it is one of the seven sites where an amino acid substitution or “polymorphism” oc- curs,A2 (2) it is believed to produce some of the most substantial shifts in spectral sensitivity (Asenjo et al., 1994; Merbs & Nathans, 1993); (3) in general, shifts are readily apparent in codon 180 of the L-pigments but are either smaller or not de- tected in the M-pigments as evidenced by the measured spectral peaks (Asenjo et al., 1994; Merbs & Nathans, 1992b, 1993); and (4) substitutions occurring at position 180 are likely to yield the largest L-pigment peak absorption shift (M. Neitz & J. Neitz, 1998). Compared with relatively smaller spectral shifts produced by changes that occur at other locations in the gene array, the amino acid substitutions at position 180 play an essential role in producing individual differences in normal color vision and also play a role in modulating the severity of color vision de- fects (M. Neitz & J. Neitz, 1998). Thus, codon 180 is a crucial site to consider when investigating the impact of shifted-peak spectral sensitivity on perception. We examine the presence or absence of amino acid substitu- tions observed at codon 180 for both red and green opsin genes simultaneously. What can one infer regarding possible retinal phenotypes from genetics tests on specific codon 180 amino 260 JAMESON, HIGHNOTE, AND WASSERMAN APPENDIX A (Continued) acids in either the red (LWS pigment) or the green (MWS pig- ment) opsin gene? Although larger sample sizes are needed to accurately assess the population frequency of the serine–180 shift versus the alanine–180 shift, some investigations suggest that the general Caucasian population shows a ~56% occurrence for the “normal” serine 180 and a ~44% occurrence of a poly- morphic mutation alanine 180 for the red pigment gene. By comparison, the frequency of occurrence for the green pigment gene is ~96% “normal” alanine 180 and ~4% polymorphic ser- ine 180 (Winderickx et al., 1993). Asenjo et al. (1994), M. Neitz and J. Neitz (1998), and Sharpe, Stockman, Jaegle, et al. (1998), show that the serine amino acid at codon 180 is primarily linked to the normal red gene, whereas the presence of alanine at codon 180 can indicate the presence of a polymorphic (shifted) red photopigment gene, the presence of a normal green gene, or the presence of both a shifted red and a normal green gene. Winderickx et al. (1992) suggest that MWS- pigment gene 180 substitutions of serine for the normal alanine occurs with far less frequency than in the LWS-pigment gene. When MWS 180 substitutions do occur, they produce non- significant shifts in the absorption spectra of M-cone retinal photopigments (Asenjo et al., 1994).A3 At present, the consensus is that, although the resulting spec- tral sensitivity shifts produced by codon 180 substitutions are significant for LWS-pigments, they are far less common for MWS-pigments, and, when present, they yield MWS-pigment spectral absorption peaks essentially identical to the “normal” MWS-pigment (M. Neitz & J. Neitz, 1998). For the above rea- sons, we analyzed subjects’ DNA to specify the amino acids pres- ent at codon 180 in the red and green genes. We now summa- rize the resulting genotype variants arising from the presence or absence of particular codon 180 amino acids and the majority of possibly occurring phenotypes. Classifying Photopigment Opsin Genotypes On the basis of existing findings (Asenjo et al. 1994; M. Neitz & J. Neitz, 1998; Sharpe, Stockman, Jaegle, et al., 1998; Wind- erickx et al., 1993; Winderickx et al. 1992), presence of codon 180 amino acid variants for either the red or the green gene al- lows several possible phenotypes to be expressed. These are sum- marized as follows: (1) The presence of serine at codon 180 implies the detection of a normal red gene (because serine is primarily linked to the normal red gene at position 180);A4 and (2) the presence of ala- nine at codon 180 can indicate the presence of a polymorphic L-pigment gene (with a substantially shifted peak spectral sen- sitivity from the normal L-pigment gene), the presence of a nor- mal M-pigment gene, or the presence of both a shifted red gene and a normal green gene. What these scenarios imply for possible phenotype expres- sion is relatively straightforward when dealing with males who have only a single maternally inherited X chromosome. In this case, we know when we observe a male who tests positive for the presence of both serine and alanine and who possesses a single X chromosome with at least one normal red gene and with at least one normal green gene or a mutated red gene (we say “at least one” because we cannot rule out red and green gene copies occurring at other sites in the red and green gene sequences since we do not consider other sites in this analysis). By comparison, when we observe a male who tests positively for serine 180, we can infer that this individual has a normal red gene at codon 180, but not a normal green gene or a polymorphic red gene at position 180. Likewise, we would infer that a male with alanine 180 would have either a normal green gene or a polymorphic red gene at codon 180, but not a normal red gene at 180. The same codon 180 analysis for females is complicated by the fact that, relative to males, females have two X chromosomes capable of carrying red and green genes at codon 180. In females, one X chromosome is transmitted via maternal inheritance, whereas a second X chromosome is paternally inherited. If we consider a female who tests positive only for the pres- ence of serine at position 180, the following possibilities are al- lowed: this female has two X chromosomes, each with one nor- mal red gene. One copy occurs on codon 180 of the f irst X chromosome (X1), while a second identical copy might be found on codon 180 of the second chromosome (X2). The less probable scenario is that only one of the X chromosomes has opsin genes at position 180, and the second chromosome has none of the described red opsin genes at 180. Regardless of the sce- nario, this female still has at least one copy of the normal red gene. Alternatively, consider a female who tests positive for the pres- ence of the alanine polymorphism at position 180: This female has two X chromosomes, each with a normal green gene or a polymorphic red gene at codon 180, but not a normal red gene at 180 (although we cannot rule out a normal gene occurring elsewhere on another amino acid position in the red gene array). Thus, for this female, we observe the presence of a normal green gene and a mutated red gene on X1 and copies (or a subset of copies) of that normal green gene and a mutated red gene on X2. Depending on the order of these different genes in the red and green sequences of X1 and X2, and depending on X inactivation and other expression mechanisms, this female has the potential for codon 180 to produce at most a normal green photopigment opsin and a shifted red photopigment opsin. Even more varied is the female who tests positive for the presence of both serine and alanine amino acids at codon 180. In this heterozygote female, we observe an individual with two X chromosomes, each with at least one normal red gene and with at least one normal green gene or a mutated red gene. It is pos- sible that, for this female, the first X chromosome (X1) could have a different subset of genes from the second chromosome (X2). Thus, the this female might possess a normal red gene, a normal green gene, and a mutated red gene on X1 (or some sub- set of these genes); and on X 2 also a normal green gene and a mutated red gene. Depending on the order of these different genes in the red and green sequences of X1 and X2 and de- pending on X inactivation and other expression mechanisms, this serine-180–alanine female has more codon 180 gene com- binations possible on her two chromosomes than any of those males or females described above. Appendix A Notes A1. The other 11 positions are sites where the amino acid sequence can vary but where such variation does not produce shifts in spectral ab- sorption at the level of the retinal cones. A2. In a study of 45 color abnormal males, Sharpe, Stockman, Jae- gle, et al. (1998) report “except for position 180 the highly polymorphic nature of the amino acids within exon 3 makes it impossible to discern whether any particular exon 3 is derived from a red or a green pigment gene” (p. 10059). While the Sharpe, Stockman, Jaegle, et al. study does not focus on “normal” color vision, male and female, subjects, it does provide results which lend insight and may be generalized in the present study until a more extensive study of color normals is completed. A3. Moreover, according to Sharpe, Stockman, Jaegle, et al. (1998), “among red-green hybrid genes that carry serine at position 180, exon