OPTICAL REVIEW Vol. 15, No. 6 (2008) 295–301
Interference of Verbal Labels in Color Categorical Perception Kenji Y OKOI, Tomoaki N ISHIMORI, and Shinya SAIDA Department of Applied Physics, National Defense Academy, 1-10-20 Hashirimizu, Yokosuka, Kanagawa 239-8686, Japan (Received July 14, 2008; Accepted September 5, 2008) Previous studies demonstrated that color categorical perception (CP; better cross-category than within-category discrimination) was reduced by verbal interference, suggesting that CP is mediated by verbal labeling. Here, we examined chromatic generality and experience-dependency of verbal interference in CP using the Stroop effect. We employed a simultaneous two-alternative forced choice discrimination task. Congruent or incongruent words were presented prior to discrimination. In experiment 1, incongruent color names reduced CP regardless of color boundary pairs. Next, we used noncolor words that seemed to be associated with color through experience. The results showed that the tested noncolor words did not modify CP (experiment 2). However, combined presentation of color and shape produced Stroop interference (experiment 3). Our finding suggests that familiarity or mastery of categorized information through experience may be evaluated by verbal interference. # 2008 The Optical Society of Japan Key words: color categorical perception, Stroop effect, verbal labeling, familiarity, color-vision deficiency
1.
the basis of this finding, they claimed that retention of a target color name must be needed for CP. Using the Stroop effect, Suegami and Michimata examined the contribution of color labels to CP.12) They performed a simultaneous 2AFC discrimination task. Verbal interference was presented prior to the task. The results showed that an incongruent color name eliminated CP while it survived in a congruent color name condition, suggesting that CP would be mediated by verbal labels at the encoding stage. These previous studies strongly suggest that verbal labeling is involved in CP. However, only a blue-green boundary was used throughout these experiments, therefore, the generality of verbal interference on color CP is unclear. Our first approach was to evaluate the generality. If CP relies on verbal labeling, Stroop interference should be observed at other categorical boundaries. Another intention of this experiment was to investigate the association between color category and verbal labeling. Although Pilling et al. argued that the target color name is critical for CP, color names themselves were not inherent but acquired through learning. Consequently, if a specific noncolor word is closely linked to a certain color, it may interfere with CP. Moreover, the magnitude of Stroop effect may provide an index of the familiarity or the mastery for linkage between labeling and visual information. To this end we examined Stroop interference by noncolor words that were common only for the subjects. Regarding association between CP and verbal labeling, we tentatively adopted anomalous trichromats as subjects. It has been suggested that they can name color categories in a similar manner to normal subjects in spite of the partial loss of color information on the retina.14) On the other hand, it was also found that categorical efficiency for anomalous trichromats in a color visual search was poorer than that for normal subjects.15) This discrepancy may indicate different processes for naming and detection, and thus may provide new insight into the organization of CP. Since Stroop interference originates from rivalry between labeling and
Introduction
Color is essentially continuous sensation that originates from the excitation of photoreceptors. However, we often classify colors into several discrete categories, such as ‘‘red’’ or ‘‘green’’. The primary feature of color categorical perception (CP) is better discrimination across category boundaries than within the same category.1) Although CP is a very common phenomenon in our daily lives, its underlying process is controversial. Many studies have supported the idea that CP is a truly perceptual process, as its name indicates.2–9) For instance, Matsuzawa reported that a chimpanzee could classify Munsell samples into several categories that were similar to those of humans.3) Franklin et al. compared CP between infants and adults6) and between English infants and Himba toddlers,7) and found that infant color categorization was comparable to adult color categorization and that color term knowledge did not modify CP. In contrast, some studies have claimed that CP must be mediated by verbal labels, not by perceptual mechanisms.10–13) Roberson and Davidoff examined color CP using a delayed two-alternative forced-choice (2AFC) recognition memory task.10) During the inter-stimulus interval (ISI), one of the three kinds of interference was inserted: no (blank), visual, or verbal interference. In the no interference condition and in the visual interference condition, the CP effect persisted. However, verbal interference removed the difference of performance between the crosscategory stimuli and the within-category stimuli. There was no difference whether the interfering words were color name or noncolor words. Thus they concluded that CP was produced by verbal labels. Pilling et al. investigated the predictability of interference.11) They found that when the interference type was fixed, as in Roberson and Davidoff, the CP effect was eliminated, while randomized (unpredictable) interference did not remove CP. Moreover, the target color name could be retained during the task even with verbal interference. On 295
296
OPTICAL REVIEW Vol. 15, No. 6 (2008)
K. Y OKOI et al.
color perception, it might be a useful tool to examine the association between them. 2.
Experiment 1: Generality of Verbal Interference
Experiment 1 used the same general method as Suegami and Michimata.12) The method was a simultaneous 2AFC discrimination task. The verbal labels were presented irrelevantly prior to the discrimination. If Stroop words affect CP, it would be reduced or eliminated under the incongruent condition. We selected the blue-purple boundary in addition to the blue-green boundary to clarify the generality of verbal interference. The characteristics of color discrimination of anomalous trichromats were also evaluated by locating the stimulus chromaticities near confusion lines. 2.1 Method 2.1.1 Stimuli We selected 16 stimuli from Munsell samples, which consisted of four Munsell hues (1.44B, 8.76BG, 6.06BG, 4.04BG) at Munsell chroma 7.4 combined with four Munsell values ð5:40; 5:80; 6:20; 6:60Þ for the blue-green boundary condition. These samples were identical to Pilling et al.11) and Suegami and Michimata.12) A preliminary experiment confirmed that these hues straddle the blue-green boundary at approximately 7.5BG (see also ref. 2). For the blue-purple boundary condition, four Munsell hues (9.00B, 1.02PB, 3.04PB, 5.06PB) at Munsell chroma 7.4 were combined with the four Munsell values so that half of the stimuli fell into the blue category and the other half into the purple category. These chromaticities approximately aligned with the protanope’s and the deuteranope’s confusion lines in the CIE 1931 chromaticity diagram.16) Verbal labels were Japanese basic color names of ‘‘blue’’ and ‘‘green’’ for the blue-green condition and those of ‘‘blue’’ and ‘‘purple’’ for the blue-purple condition. They were typed in Japanese Hiragana characters by 90-point MSMincho font and presented in the center of a gray rectangle frame subtended 5 10 deg. Test stimuli were filled rectangles (3 7 deg) presented under the target frame. All the stimuli in a trial had equal luminance. Munsell samples were simulated on a 21 in.-CRT (EIZO Flexscan T966-BK) in a dark room and checked using a Minolta CS-100 colorimeter. Chromaticities of the simulated samples were calculated by interpolating the CIE 1931 ðx; yÞ equivalents of Munsell samples under the CIE standard illuminant C.16) They are summarized in Appendix. The experiments were controlled by Psychtoolbox in MATLAB on an Apple Power Macintosh G3. 2.1.2 Procedure At the beginning of a session, the subject viewed a gray screen at a distance of approximately 60 cm for adaptation. An illustration of a trial sequence is shown in Fig. 1. First, a gray frame was presented on a black background for 2500 ms and fixation crosses were displayed in the center of the frame for 500 ms. After the fixation, a verbal label was shown for 250 ms, except under the control condition in
(a)
(b)
(c)
(d)
Fig. 1. An illustration of a trial sequence in experiment 1. (a) a gray frame for 2500 ms, (b) fixation crosses for 500 ms, (c) verbal interference for 250 ms, and (d) a target frame and two test patches were presented until a response was given.
which only the frame was presented. Following the disappearance of the verbal label, the gray frame was colored by a target color and two test patches were presented under the frame. The subjects were instructed to respond to which patch was identical to the target frame as fast as possible by pressing a key. No feedback was given. The blue-green boundary condition and the blue-purple boundary condition were performed in separate sessions. Under each boundary condition, all adjacent hue pairs in the 16 samples (four hues four values) were used, giving eight within-category pairs and four cross-category pairs. Verbal interference conditions consisted of congruent, incongruent, and control conditions. The verbal labels agreed with the target color in the congruent condition, while they were inconsistent in the incongruent condition. No label was presented in the control condition. The verbal interference conditions were randomized and counterbalanced in a session, so that the verbal labels provided no clue about the target color. The subjects were informed of the irrelevance between the verbal labels and the target colors. A total of 144 trials, which consisted of each combination of color pair (12), target color in a pair (2), target patch location (2), and verbal interference (3), were performed in a session. Three sessions were repeated for each boundary condition. 2.1.3 Subjects Six subjects participated in experiment 1. All were fourth grade cadets and a staff member at the National Defense Academy. Four of them had normal color vision. The others were protanomalous and deuteranomalous as confirmed by the 100-hue test.
OPTICAL REVIEW Vol. 15, No. 6 (2008)
2.3 Discussion For both boundary conditions, better discrimination in cross-category pairs than in within-category pairs was observed under the control condition and the congruent condition. However, the CP effect was reduced under the incongruent condition, suggesting that it was disturbed by incongruency between verbal labeling and color category. Thus CP involved verbal labeling. Correspondence between both boundary conditions suggested that verbal interference in CP was restricted not only to the blue-green boundary but also to other boundaries. This tendency was commonly observed in most of the subjects. The performance of anomalous trichromats was not inferior to that of normal subjects for the blue-green boundary condition. However, discrimination for the bluepurple boundary condition seemed to be difficult for them in this experiment. This result may be reasonable since the chromaticities of the stimuli were located along confusion lines. However, they can name blue and purple as well as blue and green in daily activities. Recent studies suggest that they rely on subtle visual cues acquired through learning to distinguish confusing colors such as blue and purple.14) Thus, they may utilize qualitatively different cues to discriminate blue-green and blue-purple. 3.
Experiment 2: Effect of Noncolor Words Associated with Color
Experiment 1 demonstrated that Stroop interference between color category and color name removed CP both at the
Proportion Correct
(a)
297
1
0.9
0.8 within cross 0.7
Proportion Correct
(b)
control
congruent
incongruent
control
congruent
incongruent
control
congruent
incongruent
1
0.9
0.8
0.7
(c)
1
0.9 Proportion Correct
2.2 Results Figure 2 represents the mean proportion of correct response. For the blue-green boundary condition, the averaged performance of the anomalous trichromats (¼ 0:914) was quite comparable to that of the normal subjects (¼ 0:932). Therefore, the performances were averaged for all of them in Fig. 2(a). A two-way analysis of variance (ANOVA) was performed using category type and verbal interference as factors. The results showed that targets were better recognized in cross-category pairs than in within-category pairs (F1;5 ¼ 17:32, p < 0:01). Post-hoc Tukey’s honestly significant difference (HSD) test revealed that CP was significant in the congruent and in the control condition (p < 0:01), while there was no significant difference between category types in the incongruent condition (p ¼ 0:110). For the blue-purple boundary condition, the correct rates between the normal subjects (¼ 0:882) and the anomalous trichromats (¼ 0:638) were distinctive, so that they were plotted separately in Figs. 2(b) and 2(c), respectively. A two-way ANOVA was performed only for the results of the normal subjects, because the number of anomalous trichromats was not enough to analyze the performance statistically. The results showed a significant main effect of category type (F1;3 ¼ 24:31, p < 0:05). Post-hoc test revealed that CP was significant in the congruent and in the control condition (p < 0:05), while not significant in the incongruent condition (p ¼ 0:351).
K. Y OKOI et al.
0.8 0.7 0.6 0.5
Fig. 2. Mean proportion of correct responses for different types of category pairs with no (blank), congruent, or incongruent verbal labels in experiment 1. (a) the blue-green boundary condition (n ¼ 6), (b) the blue-purple boundary condition of the normal subjects (n ¼ 4), and (c) the anomalous trichromats (n ¼ 2). Error bars represent the standard error of the mean.
blue-green boundary and at the blue-purple boundary. This Stroop effect has been thought to reflect the difficulty in ignoring irrelevant semantic information. However, color
OPTICAL REVIEW Vol. 15, No. 6 (2008)
3.1 Method 3.1.1 Stimuli and procedure The same stimuli and the same procedure were used as in experiment 1 except for the verbal labels. Kouku (‘‘air’’ related with blue) and rikujo (‘‘ground’’ related with green) were used as the verbal labels and displayed in Japanese Kanji characters in the blue-green boundary condition. Kouku and kaijo (‘‘maritime’’ related with purple) were used in the blue-purple boundary condition. These conditions were applied in separate sessions.
(a)
Proportion Correct
names are not connected with color appearance by nature, but associated through massed practice and learning. Thus, noncolor words associated with color may affect CP, and the magnitude of Stroop effect may provide a reliable index of the familiarity or the mastery of objects and information associated with color through experience. In the Self-Defense Forces (SDF) of Japan, officials of the Air SDF wear bluish uniforms and those of the Ground SDF put on greenish uniforms. The color of uniforms are probably not familiar for ordinary people. However, subjects of the National Defense Academy may have acquired an implicit association between color category and specific noncolor words which are closely linked to these familiar objects. To clarify the Stroop effect between them, we used kouku and rikujo (‘‘air’’ and ‘‘ground’’ in Japanese, respectively) as the verbal labels for the blue-green boundary condition. For the blue-purple boundary condition, we selected kouku and kaijo (‘‘air’’ and ‘‘maritime’’ in Japanese) as the verbal labels. In the National Defense Academy, cadets always put name tags on their chests. Besides their names, the year grade and the assigned services of the SDF are indicated by color labels (blue, purple, and brown for the air, maritime, and ground services, respectively). Since the military training differs according to the services, these color categories and noncolor words may be conceptually related in the visual system.
K. Y OKOI et al.
1
0.9
0.8 within cross 0.7
(b)
Proportion Correct
298
control
congruent
incongruent
control
congruent
incongruent
1
0.9
0.8
0.7 Fig. 3. Mean proportion of correct responses in experiment 2. (a) the blue-green boundary condition (n ¼ 5), (b) the blue-purple boundary condition (n ¼ 4).
3.1.2 Subjects Five subjects who had taken part in experiment 1 participated in experiment 2. Four of them had normal color vision and the fifth individual was deuteranomalous.
since the average performance of the deuteranomalous subject (¼ 0:569) was lower than those of the normal subjects (¼ 0:910). The main effect of CP was significant (F1;3 ¼ 75:21, p < 0:01) and post hoc test showed that difference between cross- and within-category pairs was significant under all the verbal conditions (p < 0:05 for the congruent condition, p < 0:01 for the other conditions).
3.2 Results For the blue-green boundary condition, the averaged performance of the deuteranomalous subject (¼ 0:882) was equivalent to that of the normal subjects (¼ 0:911), so that the performances were averaged for all the subjects and plotted in Fig. 3(a). The same analysis as that in experiment 1 was performed and the significant main effect of CP was found (F1;4 ¼ 10:23, p < 0:05). Post hoc test showed that performances for cross-category pairs were significantly better than those for within-category pairs under all the verbal interference conditions (p < 0:05). For the blue-purple boundary condition, the results were averaged for the normal subjects and are shown in Fig. 3(b),
3.3 Discussion In experiment 2, CP persisted under all conditions and Stroop interference was not found. It might be argued that noncolor words did not interfere with color perception as suggested by Pilling et al.11) However, the association between selected words and color might not be bound tightly in our subjects, or our stimuli might be too abstract to remember actual objects. We tested this possibility in experiment 3. Although the selected noncolor words could not modify CP under all conditions, this result did not agree with Roberson and Davidoff.10) In their experiment, verbal interference was observed regardless of whether or not the
OPTICAL REVIEW Vol. 15, No. 6 (2008)
K. Y OKOI et al.
(a)
(b)
Fig. 4. An illustration of the stimulus in experiment 3. (a) a silhouette shape of the uniform in the blue-green boundary condition, (b) a simulated name tag in the blue-purple boundary condition.
Proportion Correct
(a)
299
1
0.9
0.8 within cross 0.7
4.
congruent
incongruent
control
congruent
incongruent
(b)
0.9
Proportion Correct
verbal labels were related to color. Pilling et al. argued that disappearance of CP by noncolor words should depend on the strategic bias of whether the subjects chose to retain the target color name. They suggested that CP was reduced when the target color name was not retained or was disrupted, whereas CP could survive verbal interference if the subjects could retain the target color name during the task. In our experiment, however, target color and test patches were presented simultaneously, so that no memory process was required. Our results supported the idea that verbal labels would interfere not at the memory stage, but at the encoding stage.12)
control
0.8
0.7
Experiment 3: Combination of Color and Shape
In experiment 2 we sought to learn whether noncolor words could induce the Stroop effect. However, selected verbal labels seemed to inadequate to interfere with color perception semantically. Regarding object recognition, it is suggested that color information is stored with respect to shape in long-term memory.17,18) Hence, conjoint presentation of color and shape may enhance association between verbal labels and color perception. In experiment 3, we examined whether the combination of color and shape could induce the Stroop effect. 4.1 Method 4.1.1 Stimuli and procedure All the stimuli including the verbal labels and the procedure were identical to those in experiment 2 except for the shape of the target frame and the test patch. For the blue-green boundary condition, we selected the verbal labels according to the color tones of the SDF uniforms. To help the association between the label and the uniform, shape of the target frame and the test patches were changed to a silhouette of the upper part of the uniform as shown in Fig. 4(a). The size of the stimulus was approximately 6 6 deg. The fixation cross and the verbal label were presented in the center of the uniform stimulus. For the blue-purple boundary condition, a National Defense Academy cadet name tag was simulated as shown in Fig. 4(b). A fictitious name was typed in the right part of
0.6 Fig. 5. Mean proportion of correct responses in experiment 3. (a) the blue-green boundary condition (n ¼ 5), (b) the blue-purple boundary condition (n ¼ 4).
the tag on a pale yellow background like an actual tag. The upper left part was colored red which represented the same grade as that of the participating cadets. The target color and the test patch were located in the lower left part of the tag. Each part was separated by narrow black lines. The whole tag and the lower left part subtended about 3 5:5 deg and 1:5 1:5 deg, respectively. Size of the simulated tag corresponded to the actual tag at the viewing distance. The fixation cross and the verbal label were shown under the target cue part. 4.1.2 Subjects Five subjects of experiment 2 participated in experiment 3. One of them was deuteranomalous. 4.2 Results For the blue-green boundary condition, the averaged performance of the deuteranomalous subject (¼ 0:844) was comparable to that of the normal subjects (¼ 0:946), so that the performances were averaged for all the subjects in Fig. 5(a). A two-way ANOVA revealed the main effect of
300
OPTICAL REVIEW Vol. 15, No. 6 (2008)
CP (F1;4 ¼ 31:81, p < 0:01). Post hoc test showed that the difference between cross- and within-category pairs was significant under all the verbal conditions (p < 0:01 for the incongruent condition, p < 0:05 for the other). For the blue-purple boundary condition, the performance of the deuteranomalous subject (¼ 0:509) was at a chance level, while that of the normal subjects (¼ 0:766) was fine. The results were averaged for the normal subjects and are plotted in Fig. 5(b). A two-way ANOVA showed that the main effect of CP was significant (F1;3 ¼ 22:83, p < 0:05). Post hoc test revealed that the CP was significant in the congruent condition (p < 0:01) and nearly significant in the control condition (p ¼ 0:063). However, there was no significant difference in the incongruent condition (p ¼ 0:327). 4.3 Discussion For the blue-purple boundary condition, we found Stroop interference in the incongruent condition, while CP was preserved in the other conditions. We cannot simply compare this result with that of experiment 2, because the stimulus configuration was not identical. However, it is suggested that the conjunctional presentation of color and shape enhanced semantic linkage between noncolor words and color category, so that the incongruent label interfered with CP. The fact that noncolor words, which are essentially irrelevant to color appearance and connected through experience, affected CP is very informative in revealing the organization of CP. It may also be applicable to estimate the underlying familiarity or mastery of object recognition in each subject. For the blue-green boundary condition, the result replicated that of the previous experiment 2. Stroop interference was not observed under any conditions. One possible interpretation is that simulation of the SDF uniform was not yet realistic. While the selected chromaticities in the blue-purple boundary condition were similar to those of the actual name tag [c.f. CIE 1931 ðx; yÞ chromaticities of the color labels for air and maritime services measured under the standard illuminant C were ð0:234; 0:292Þ and ð0:297; 0:239Þ, respectively], the color tones of the SDF uniforms were more saturated and deeper than the Munsell samples used. Though the tones were consistent with the presented stimuli categorically, discrepancy of the detailed color appearance may have hindered the association between them. This possibility may be tested by using photographic stimuli. Another possibility is the difference in familiarity of the simulated objects. Cadets at the National Defense Academy always put on their name tags and must pay attention to the tags to identify who is senior in any situation. This familiarity may strengthen verbal interference for the bluepurple boundary condition. In contrast, uniforms of the cadets differ from those of the SDF officials including the color tones, and color of the uniform of the cadets is standardized regardless of their assigned services. Therefore, association between blue-green categories and the color tones of the SDF uniforms may have been less recognized by the participating subjects.
K. Y OKOI et al.
5.
General Discussion
The results from experiment 1 suggested that verbal interference was not a specific feature but a general property of CP. Berlin and Kay conducted a linguistic study and suggested that well-developed languages have 11 basic color terms.19) Many psychological studies have also supported this idea that basic color categories were used more consistently among subjects than nonbasic colors.20,21) However, little is known of whether the number of basic categories is optimal or not.22) Although the previous studies employed naming and memory paradigms to distinguish basic and nonbasic colors, verbal interference may also be an efficient means of characterizing them. Experiments 2 and 3 demonstrated that noncolor words could also affect CP when the words were closely linked with color. Relation between color name and color perception is generally linked through long-term learning. However, knowledge of the combination of noncolor words and color categories in our experiments was peculiar to the subjects. Freedman and Assad trained monkeys to classify visual motion directions into two discrete categories and found that abstract representations of visual motion were encoded in the lateral intraparietal area while fine selectivity of directions was retained in the middle temporal (MT) area, suggesting that new representation of visual categories could be acquired through experience.23) Our finding may provide insight into the acquisition of experience-dependent representation in CP. We adopted anomalous trichromats to explore the association between verbal labeling and color perception. For the blue-green boundary condition, their performance of discrimination was comparable to that of normal trichromats throughout our experiments. CP effects and Stroop interference were also observed, suggesting that they process the categorical boundary between blue and green in a similar manner to trichromats. In contrast, it seemed difficult for them to distinguish the blue-purple boundary in our experiments. However, they can recognize blue and purple, including the color labels on the name tag, in their activities. What caused this discrepancy? It has been suggested that they might utilize additional information to categorize confusing colors, not to discriminate them.24) This possibility may be clarified by categorical tasks like our experiments while adjusting discriminability. Characteristics of categorization by anomalous trichromats may be useful not only to clarify their amazing naming abilities, but also to examine acquisition of new categorization through learning. This issue needs further investigation. Categorization is one of the fundamental functions of our visual system. Many studies have suggested that CP is mediated by perceptual process.2–9) Especially, results from infants and animals, in which the acquisition of verbal labeling is difficult, strongly support perceptual superiority in CP. In contrast, verbal interference in CP demonstrates that CP involves verbal coding.10–13) Although this inconsistency is indeterminable at this point, Matsuno et al. investigated the effect of a color naming experience on color
OPTICAL REVIEW Vol. 15, No. 6 (2008)
categorization by chimpanzees.25) A chimpanzee was trained to learn symbolic color naming, while the other received less training. Distribution of color categories in a color space was found not to depend on the color naming experience, but categorization stability was influenced by the experience. Thus, CP may be formed by a perceptual process and polished by verbal labeling. Further investigation is needed to clarify the contribution of perceptual and verbal processes on categorization.
K. Y OKOI et al.
23) D. J. Freedman and J. A. Assad: Nature 443 (2006) 85. 24) K. Uchikawa: Ext. Abstr. Asia–Pacific Conf. Vision, 2008, p. 26. 25) T. Matsuno, N. Kawai, and T. Matsuzawa: Behav. Brain Res. 148 (2004) 157.
Appendix Table A-1. CIE 1931 ðx; yÞ chromaticity coordinates for the simulated Munsell samples in the blue-green boundary condition. Hue
References 1) S. Harnad: in Categorical Perception: The Groundwork of Cognition, ed. S. Harnad (Cambridge University Press, Cambridge, 1987) p. 535. 2) M. H. Bornstein and N. O. Korda: Psychol. Res. 46 (1984) 207. 3) T. Matsuzawa: J. Human Evol. 14 (1985) 283. 4) R. M. Boynton, L. Fargo, C. X. Olson, and H. S. Smallman: Color Res. Appl. 14 (1989) 229. 5) K. Uchikawa and H. Shinoda: Color Res. Appl. 21 (1996) 430. 6) A. Franklin, M. Pilling, and I. Davies: J. Exp. Child Psychol. 91 (2005) 227. 7) A. Franklin, A. Clifford, E. Williamson, and I. Davies: J. Exp. Child Psychol. 90 (2005) 114. 8) K. Yokoi and K. Uchikawa: J. Opt. Soc. Am. A 22 (2005) 2309. 9) K. Koida and H. Komatsu: Nat. Neurosci. 10 (2007) 108. 10) D. Roberson and J. Davidoff: Mem. Cognit. 28 (2000) 977. 11) M. Pilling, A. Wiggett, E. Ozgen, and I. R. Davies: Mem. Cognit. 31 (2003) 538. 12) T. Suegami and C. Michimata: Tech. Rep. Atten. Cogn. (2005) 19. 13) A. L. Gilbert, T. Regier, P. Kay, and R. B. Ivry: Proc. Natl. Acad. Sci. U.S.A. 103 (2006) 489. 14) V. Bonnardel: Vis. Neurosci. 23 (2006) 637. 15) K. Yokoi, K. Watanabe, and S. Saida: submitted to Percept. Mot. Skills. 16) G. Wyszecki and W. S. Stiles: Color Science: Concepts and Methods, Quantitative Data and Formulae (Wiley-Interscience, New York, 2000) 2nd ed. 17) K. G. Nicholson and G. K. Humphrey: Perception 33 (2004) 339. 18) A. M. Proverbio, F. Burco, M. del Zotto, and A. Zani: Brain Res. Cogn. Brain Res. 18 (2004) 288. 19) B. Berlin and P. Kay: Basic Color Terms: Their Universality and Evolution (University of Chicago Press, Chicago, 1969). 20) K. Uchikawa and R. M. Boynton: Vision Res. 27 (1987) 1825. 21) R. M. Boynton and C. X. Olson: Vision Res. 30 (1990) 1311. 22) D. Roberson, I. Davies, and J. Davidoff: J. Exp. Psychol. Gen. 129 (2000) 369.
301
1.44B 8.76BG 6.06BG 4.04BG 1.44B 8.76BG 6.06BG 4.04BG 1.44B 8.76BG 6.06BG 4.04BG 1.44B 8.76BG 6.06BG 4.04BG
Value
6.60
6.20
5.80
5.40
L (cd/m2 )
x
y
18.5
0.224 0.226 0.229 0.234
0.281 0.294 0.309 0.322
16.0
0.220 0.222 0.225 0.230
0.279 0.293 0.309 0.322
13.8
0.216 0.217 0.221 0.226
0.276 0.291 0.308 0.322
11.7
0.212 0.213 0.216 0.221
0.274 0.289 0.307 0.321
Table A-2. CIE 1931 ðx; yÞ chromaticity coordinates for the simulated Munsell samples in the blue-purple boundary condition. Hue 9.00B 1.02PB 3.04PB 5.06PB 9.00B 1.02PB 3.04PB 5.06PB 9.00B 1.02PB 3.04PB 5.06PB 9.00B 1.02PB 3.04PB 5.06PB
Value
6.60
6.20
5.80
5.40
L (cd/m2 )
x
y
18.5
0.231 0.235 0.240 0.245
0.253 0.248 0.245 0.242
16.0
0.227 0.232 0.237 0.243
0.250 0.245 0.241 0.239
13.8
0.224 0.228 0.234 0.240
0.246 0.241 0.238 0.235
11.7
0.220 0.225 0.230 0.237
0.243 0.237 0.234 0.231