Effects of Latitudinal Position on Color Perception: The Case of Brazil. Joshua B. Wortman. University of California, San Diego

Transcription

1 Running head: COLOR PERCEPTION AND COGNITION Effects of Latitudinal Position on Color Perception: The Case of Brazil Joshua B. Wortman University of California, San Diego Address ms. Correspondence to: Joshua Wortman Dept. of Psychology University of California, San Diego 9500 Gilman Dr. La Jolla, CA (U.S.A.) Tel: Lab:

2 Latitude and Color Perception 2 Abstract Human perception and discrimination of colors vary as correlated with distance from the equator; specifically populations with closer proximity to the equator have less ability to discriminate shades of blue (Bornstein, 1973). Evidence is based on data collected from various populations through color naming tasks or from performance on color discrimination tasks; however, comparing evidence coming from distinct cultural groups prevents assessment of how much performance variations are attributable to nonculturally related differences. This study investigates the hypothesis that distance from the equator is correlated with perception of short wave length colors even among people with common language and nationality in order to elucidate how non-cultural, environmental factors affect color perception and naming patterns. Experiments were conducted on Portuguese speaking Brazilians resident in one of three coastal cities: 1 o, 8 o, and 22 o South of the equator. A color naming task and a color discrimination task were used. Some correlations between latitudinal location and performance are found supporting the hypothesis that distance from the equator is related to the perception and cognition of color, independent of culture.

3 Latitude and Color Perception 3 Effects of Latitudinal Position on Color Perception: The Case of Brazil Color is used widely as a tool for studying perception and cognition. Color discrimination tasks are used to assess just noticeable differences (JNDs) at different wave lengths of visible light because humans do not discriminate all areas of the color space equally. The eye is more sensitive to certain wavelengths causing the JND sensitivity function for hue discrimination to vary with wavelength (Kay & Kempton, 1984). If linguistic data are considered, color discrimination tasks allow assessment of the degree that language affects perception and the JND function. Previous studies have shown color difference judgments are related to color terms in language (e.g. Davidoff, Davies, & Robertson, 1999b; Kay & Kempton, 1984). The extent to which language systems influence cognitive structure and affect perception is debated. Supporters of the view that language structure determines world view, such as Sapir and Whorf, assert linguistic categorization of colors allows improved discrimination of hue across color-term lexical boundaries and diminished hue discrimination within a color category group (Kay & Kempton, 1984). For example, there are several languages that do not lexically discriminate blue from green and some do not discriminate blue from black (Bornstein, 1973). These languages are referred to as grue languages due to their conjoining of green and blue. Studies have shown that grue languages often make semantic confusions mimicking tritanopic vision (Boynton & Gordon, 1965). Color naming studies are also used to identify the basic color terms used to describe color space. Each basic color term applies to a variety of hues. A psychological line dividing color space between any two color terms is called a lexical category boundary.

4 Latitude and Color Perception 4 Identifying these color term boundaries is necessary to assess how linguistic categorization affects discrimination. From color name studies it is shown that not all language cultures have the same variety of color terms (Berlin & Kay, 1969). Berlin and Kay compiled data from 98 languages to assess how basic color terms vary across language. Some languages they investigated did not have a robust color term nomenclature and some had up to 11 basic color terms. They explained these differences with an evolution-related hypothesis, stating that older languages have few color terms and adopt more as they develop technologically. It is evident by analyzing the data that languages with fewer color terms, particularly the grue languages that collapse green and blue into one linguistic category, originate in locations proximal to the equator (Bornstein, 1973). This patterning shows that neither the linguistic relativism nor the evolutionary perspective adequately explain crosscultural differences in color term use and perception. Other variables are involved as well. From examining the NASA global U.V. Erythemal exposure charts, it is seen that distribution of ultra violet sunlight is most intensely concentrated around the equator, decreasing with progression towards the poles (Total Ozone Mapping Spectrometer, 2004). The coincidence of grue languages with the equator suggests there may be some physiological factor related to ultra violet light exposure underlying perceptual and cognitive differences found among populations at different distances from the equator. Language patterns of populations near the equator correlate with patterns expected of people whose eyes have been exposed to high quantities of ultra violet radiation.

5 Latitude and Color Perception 5 It is known that ultra violet radiation causes the eyes to age and either a decrease in spectral sensitivity to short wave length light or an increase in macular pigment that filters the ultra violet light along with short wave length visible light. Long wavelength visible light sensitivity is less affected. The hypothesis investigated here considers that color vision sensitivity is susceptible to degradation from exposure to ultra violet light. This being the case, if the localities of a population subtend substantial latitudinal distance, then cognitive and perceptual nonuniformities should be evident in the experience of short wave length light, even when the population represents one linguistic culture. Thus, the intention is to find differences in performance on color tasks by varying degrees of latitude holding the culture as constant. It is expected that an interaction will occur having performance for long wavelength (red) hues unaffected by distance from the equator and performance with short wavelength (blue) hue being dependent on distance from the equator. For this study, Brazil was investigated because it is the longest country in the world North to South, allowing the greatest degree of latitude to be traversed within one language population. Data were collected from urban populations in Belem (S01 o, W48 o ), Recife (S08 o, W35 o ), and Rio de Janeiro (S22 o, W43 o ). To find support for the hypothesis, it is expected that as distance from the equator increases, performance on a color discrimination task will improve for short wave length hues and the color nomenclature will become more robust for describing short wave length hues, whereas performance regarding long wave length hues should be unaffected.

6 Latitude and Color Perception 6 Method Participants Participants were recruited from hotels and malls in three different urban coastal Brazilian cities (n B = 16, n R = 21, n RJ = 20). They had mean age of 29.4 years (SD = 9.7). All were tested for color blindness using six Ishihara color plates presented as images on a computer monitor. Participants were not allowed to have lived more than five years outside of their area of residence to be considered. Due to problems some individuals faced with filling out the response sheet correctly or for having spent more than five years in another city, not all individuals were considered in all analyses. Design and Stimuli A preliminary data sheet was filled taking into account literacy, age, education, profession, eye color, and where they had lived during their life. Experiments were conducted on a laptop computer with 14.1 LCD XGA screen. Subjects sat 5 feet away from the computer with their line of sight perpendicular to the monitor. The subjects were assessed for color vision normality by completing a preliminary color vision test composed of six Ishihara color plates presented as slides on the monitor. Subjects that demonstrated normal color vision continued. The first experiment had subjects name 37 color tiles appearing one at a time on uniform grey (50% luminance) background. Presented colors varied in hue and were either 50% or 100% saturated in HSL coordinates. Subjects responded freely with no time constraints. Results were analyzed for dominant color term per hue and variety of terms within a color category. The second experiment was a discrimination task following a two-alternative forced choice model.

7 Latitude and Color Perception 7 Subjects completed 36 trials by discriminating difference in color between three equal sized squares (placed in a row) with equal illumination and similar but varying hue (Figure 1). These were presented on the same uniform 50% luminance grey background. The task was to choose which extreme square was most different in color from the center square. Error rates were measured for hue by collapsing trials into three hue groups (red, green, and blue). Data were analyzed using a 3 x 3 repeated measures analysis of variance design (error rate in three hue groups among participants from one of three cities). Procedure Experiments were conducted in Portuguese. Participants heard an explanation of the procedures and were asked to read and sign a consent form summarizing the procedure and informing them of their right to withdraw from the experiment at any time. Due to the literacy constraints of some participants, the preliminary data sheet and test response sheets were filled out with the assistance of the experimenter. Results The color naming task did not uniformly poll all areas of color space. Several basic color terms were used by respondents to describe the hues presented. The four most predominant were vermelho, rosa, verde, and azul. These names were often modified with an additional descriptor to make more specific color naming judgments. Results revealed that distance from the equator had a significant effect on the number of modifiers for some basic color terms, but not for others. Figure 2 shows the number of variations for some color terms that were used during the color naming task (i.e. light

8 Latitude and Color Perception 8 blue, sky blue, etc. are considered varieties of blue). In accordance with the hypothesis, the number of azul variations utilized to name the stimuli varies significantly with distance from the equator, F (2,42) = 3.34, p<.05. The group farthest away from the equator (22 o south) utilizes a larger variety of color names to describe blue hues compared to the group 8 o south of the equator, t=2.80, p<.001, but not compared to the group nearest the equator, t=0.91, p<.19. The number of variations for the other basic color terms (vermelho, rosa, and verde) do not vary significantly with distance from the equator. Data supports the hypothesis that people farther from the equator have more robust naming schemes to describe blue regions of color space, but for longer wave length colors such as red and pink, distance from the equator does not affect cognition or nomenclature. The color naming data were next examined for basic color term use frequency, different than the previous measurement, now multiple instances of each color term s use are counted. Figure 3a-e shows frequency that residents in each city describe a given hue presentation with the given color term (i.e. azul). The degree of congruency between cities can be seen. The osculating frequency seen for use of the term azul at the short wave length end of the color spectrum, visibly larger among near equator residents, indicates some hues in this region of the color spectrum are described with names other than azul. Comparing this data trend in Figure 3d to Figure 3e, it is seen that the troughs of the oscillation for use of azul correspond to the presentation of 50% illuminant stimuli and to an increase in response frequency of terms branco and cinza (white and grey in English). Instead of comparing the frequency that azul is used by respondents in each city we compare the frequency that branco and cinza are used. Looking only at the last 15

9 Latitude and Color Perception 9 blue color presentations, branco and cinza are used significantly more often in Recife to describe these blue hues than in Rio de Janeiro, t (14) =2.26, p<.02. Consistent with the hypothesis, oscillations in the blue region may indicate that respondents in Recife had more difficulty seeing the azul quality of these hues, they were more likely to see them as achromatic. Data from each participant in the color discrimination task were divided into three color-grouped bins containing 12 trials each. Judgment errors were summed for each bin. Thus each participant provided three measures. Average error among participants was calculated for each of the three cities investigated. Results were analyzed with 3 x 3 repeated measures analysis of variance. Figure 4 illustrates the results of the color discrimination task for participants in the three Brazilian cities (n=46). Cities are shown in order of increasing distance from the equator. The mean number of incorrect responses for three hue types (red, green, and blue) is displayed for each city. Thus, a smaller value implies better performance. Results of the repeated measures analysis reveal a significant effect of hue, F(2,86)=6.15, p<.003. Some colors were missed more than others irrespective of latitude. The analysis also shows some evidence that discrimination performance for blue hues improved with increasing distance from the equator while discrimination performance of red hues remained constant, suggesting an emerging interaction though not statistically significant, F(4,86)=1.73, p<.15.

10 Latitude and Color Perception 10 Discussion From the color naming data we see both non-uniformities in the variety of names used and in the frequency that some basic color terms are used related to what city the data was collected from and to the area of color space being considered. People nearer to the equator, specifically Recife, differed significantly in their color naming patterns when compared to people from Rio de Janeiro. The data provide linguistic evidence that people farther from the equator make more cognitive distinctions (better discriminations) in blue color space and are more likely to see the subtle blue qualities of a pale hue than people nearer the equator. Results from the discrimination task revealed that hue differences in certain regions of the color spectrum are seen more accurately than others. This is congruent to previous findings (e.g. Kay & Kempton, 1984). The data did not show a significant interaction between discrimination of hues and latitude as the hypothesis predicted. It is possible that the interaction exists but remained undetected due to the resource and methodological limits of the experiment. Generally, the expectation remains that more significant results ought be found in both experiments, for example, when comparing data from people in Belem and Rio de Janeiro. In particular, the equator city investigated (Belem) had a much smaller group size in the analysis than the other cities (n B =9 compared to n R =21 and n RJ =17). 16 persons were investigated in Belem but seven from these were removed because they had spent too much time living in other cities. As well, data was collected in each city without control for age and age related effects, or for education. This was an unfortunate necessity due to time constraints.

11 Latitude and Color Perception 11 Another limitation in the data is the latitude subtended between cities. The most eccentric city (Rio) was only 22 o south of the equator, still being inside the Tropic of Cancer line (S23.5 o ), is relatively equatorial. Though results in data from the discrimination task were not in accord with the hypothesis, visual inspection of the data, for example in Figure 4 suggests that there is no difference between individuals of the three cities in their ability to discriminate hues in the red part of the color spectrum but that there is a trend of improved performance in ability to discriminate hues in the blue and green parts of the spectrum as distance from the equator increases. The hypothesis that differences in color discrimination between populations can be partially explained by environmental phenomena such as ultra violet induced visual degradation rather than cultural or language differences was not adequately supported by the data. It is highly recommended that a future study improves on the methodological shortcomings of the present investigation before this issue can be resolved conclusively.

12 Latitude and Color Perception 12 References Bornstein, M. H. (1973). Color vision and color naming: a psychophysiological hypothesis of cultural difference. Psychological Bulletin, 80, Boynton, R. M., & Gordon, J. (1965). Bezold-Brucke hue shift measured by a color naming technique. Journal of the Optical Society of America, 55, Davidoff, J., Davies, I., & Robertson, D. (1999a). Color categories in a stone-age tribe [Addendum]. Nature, 402, 604. Davidoff, J., Davies, I., & Robertson, D. (1999b). Color categories in a stone-age tribe [Letter]. Nature, 398, Total Ozone Mapping Spectrometer, Urythemal U.V. Data (n.d.). Retrieved October 1, 2004, from Kay, P., & Kempton, W. (1984). What is the Sapir-Whorf hypothesis? American Anthropologist, 86, Lindsey, D. T., & Brown, A. M. (2002). Color naming and the phototoxic effects of sunlight on the eye. Psychological Science, 13,

13 Latitude and Color Perception 13 Author Note J. Wortman, Department of Psychology, University of California, San Diego. I d like to thank UCSD Undergraduate Research Scholarships for their support of this research. Without their assistance I would never have had the phenomenal experience of completing this research. To my mentor Donald MacLeod, for taking me under his wing. Also, to my research accomplice and life long friend Joana Simões de Melo Costa, for helping me every step of the way. Correspondence should be addressed to: J. Wortman, Dept. of Psychology, University of California, San Diego, 9500 Gilman Dr., La Jolla, CA (USA).

14 Latitude and Color Perception 14 Figure Caption Figure 1. Example of a trial used for the discrimination task. The task was twoalternative forced choice. Subjects were asked to choose which extreme square is more different in hue (color) from the center square. Subjects responded to 36 trials and were not constrained for time. Figure 2. Data from the color naming task revealed the number of variations on basic color terms that subjects used to describe the presented stimuli. The variety of azul hues applied varied significantly between Recife (n=21) and Rio (n=16), but not between Belem (n=9) and Rio. For verde, rosa, and vermelho, the city investigated did not have a significant effect on the variety of color names used. Figure 3. Color naming task results showing the proportion that members in each city used the specific color term to describe each hue presented. The hues presented are indicated by the colors on the abscissa axis. Troughs in oscillations indicate hues that are being described by other terms. For short wave length hues presented with HSL 50% illumination, branco and cinza emerge as chosen color terms. Figure 4. Mean errors when discriminating hues in different areas of color space. Data is shows performance for three coastal cities that differ in distances from the equator (1 o, 8 o, and 22 o South) (n=46).

15 Latitude and Color Perception 15 Figure 1

16 Latitude and Color Perception 16 Figure 2 VARIATIONS OF COLOR NAME Variety of Color Terms Used Belem (1 S) Recife (8 S) Rio (22 S) CITY (DISTANCE FROM EQUATOR) Azul Verde Rosa Vermelho

17 Latitude and Color Perception 17 Figure 3 RELATIVE USE a. b % ROSA 75.0% 50.0% 25.0% RELATIVE USE 100.0% 75.0% 50.0% 25.0% VERDE Belem Recife Rio 0.0% 0.0% c. HUE PRESENTED d. HUE PRESENTED 100.0% VERMELHO 100.0% AZUL RELATIVE USE 75.0% 50.0% 25.0% RELATIVE USE 0.0% 0.0% RELATIVE USE e % 75.0% 50.0% 25.0% HUE PRESENTED BRANCO + CINZA HUE PRESENTED 0.0% HUE PRESENTED

18 Latitude and Color Perception 18 Figure 4 ERRORS Mean Errors for Group in Discimination Task Red Green Blue Belem Recife Rio CITY