Running head: Color Perception
Correspondences should be sent to: Dr. Robert Goldstone
Bloomington, IN. 47405
Subjects were shown simple objects and were asked to reproduce the colors of the objects. Even though the objects remained on the screen while subjects reproduced the colors and the objects' shapes were irrelevant to the subjects' task, subjects' color perceptions were influenced by the shape category of the object. For example, objects that belonged to categories with redder objects were judged to be more red than identically colored objects belonging to another category. Further experiments showed that the object categories that subjects use, rather than being fixed, depend on the objects to which subjects are exposed.
The notion that experience and expectations can influence perception can be traced at least back to the Sapir-Whorf hypothesis (Whorf, 1941) and the "New Look" movement (Bruner & Postman, 1949). This original work and its revivals (Niedenthal & Kitayama, 1994; von Hippel, Hawkins, & Narayan, 1994) stress that high-level cognitive processes do not simply operate on fixed, perceptual inputs; high-level processes may also create lower-level percepts. One notion from this literature is that our concepts and categories influence perception. The current experiments are concerned with the influence of learned categories on color contrast and assimilation effects. A contrast effect occurs when the dimension value ascribed to a stimulus is distorted away from its true value in the direction opposite to other presented stimuli's dimension values. An assimilation effects occurs when the dimension value is distorted toward other presented stimuli.
Contrast and assimilation effects may depend not just on the other set of stimuli in the experimental context, but also on similarities and categories that are formed among these stimuli. There is evidence that conceptual similarity influences contrast effects. In the Ebbinghaus illusion, the size of a central object appears to be smaller when it is surrounded by large, rather than small, objects. Coren and Ennis (1993) have shown that this illusion is strongest when the central and surrounding objects belong to the same conceptual category (e.g. all of the objects are dogs) than when the surrounding objects belong to a different category than the central object.
In the current experiments, subjects judged the
color of objects that belonged to different shape or conceptual
categories. Even though the object categories were irrelevant
for the color judgment task, these categories may still exert
an influence on color perception. Because of the current experimental
controls, if such an influence is found, it would have to be
due to learned rather than pre-experimental categories or category-to-color
associations. Subjects assess the color of an object by modifying
a second identically-shaped object until it has the identical
color. Contrast or assimilation effects are revealed by systematic
misestimations of an object's color.
In order to show an influence of shape categories on perception of an object's hue, the ideal situation is to arrange two or more items from different categories to have objectively identical colors. The judged hues of these equated objects can then be compared. In the two-categories condition of Experiment 1, six colored objects belong to two categories: straight-edged letters, and curved numerals. One of the letters has objectively the same hue as one of the numerals. The subjects' task is simply to judge the hue of an object by adjusting an identically-shaped object so that it possesses the same hue.
Subjects. Eighty-five undergraduate students from Indiana University served as subjects in order to fulfill a course requirement. An additional forty-seven subjects were placed in the control condition.
Materials. Colored alpha-numeric symbols were used as materials. The symbols T, E, L, 8, 9, and 6 were rendered on Macintosh II SI screens in the Geneva type font. Each symbol was approximately 5.2 cm by 4.6 cm wide. The luminance of all of the shapes was 27.6 cd/m2. As represented in Figure 1, the six symbols were given five hues along a continuum from red to violet. Going from the most red to the most violet hue, the 1976 CIE (Commission Internationale de L'Eclairage) values for the five hues were: u'=.3291, v'=.4603; u'=.3017, v'=.4097; u'=.2788, v'=.3676; u'=.2631, v'=.3376; and u'=.2480, v'=.3134. These values were chosen so that subjective differences between adjacent hues were approximately equal (Goldstone, 1994), and so that all of the hues were noticeably different from each other. Using as a reference point a black body source that produces equal energy at all wavelengths (u'=.2009, v'=.4609), the five hues can be assigned the following wavelengths (in nms): 511, 527, 541, 554, and 565.
Procedure. On each trial, a colored symbol was displayed in the upper left corner of the computer screen. A second black token of the same symbol was displaced 14 cm from the first symbol, toward the lower right corner of the screen. A horizontal 27 cm black line was presented at the bottom of the screen. The left end of the line was labeled "Red" and the right end was labeled "Violet." By moving a cursor along this line with a mouse, subjects were able to immediately alter the appearance of the second symbol. Subjects were instructed to adjust the second symbol's color so as to match the first symbol's color. When the two tokens were judged to have identical colors, subjects registered their judgment by pressing a button on top of the mouse. In order to assure that judgments were based on the actual color of the symbol rather than absolute position on the hue scale, the red-to-violet scale was offset by a random physical distance between 2 and 8 cm.
Figure 1 shows only one of the stimuli counterbalancings used. For half of the subjects, "E," "L," and "T" were given red hues; for the other half of subjects, "6," "8", and "9" were given red hues. The assignment of particular hues to the three numerals and the three letters was randomized. In all cases, one of the letters and one of the numerals shared the central hue value in Figure 1.
Subjects saw each of the six symbols repeated 36 times each. After subjects finished estimating the hue of a symbol, the screen was erased for one second, and then the next trial commenced. The entire experiment took approximately 35 minutes to complete.
An additional control condition was run to verify that misestimations were caused by letter and digit categories rather than simply the relative hues of the six symbols. In this control condition, the same symbol, randomly selected from "E," "L," "T," "6," "8," and "9," was displayed on each trial, but on each trial was randomly given one of the five hues used in the two-categories condition. The middle hue was used twice as frequently as the other four hues, as was the case in the two-categories condition.
Results and Discussion
The results of central interest concern how subjects' estimates for hues depart from the symbols' actual hue. Not all of the systematic misestimations are necessarily due to the category membership (letters vs. numerals) of symbols. Several perceptual factors could create small systematic distortions in perceived hues (Abramov & Gordan, 1994), such as a bias to see a red-orange hue as more prototypically red than it actually is. However, if we restrict our attention to "L" and "8" in Figure 1, and if there is a different pattern of judgments between them, then the difference cannot be attributable to perceptual properties of particular hues -- "L" and "8" have the same hue.
As shown in Figure 2, the mean overestimations, in nanometers, for "T," "E," "L," "8," "9," and "6" were -2.24, 0.16, .56, 1.84, 3.2, and 2, respectively. Positive overestimations indicate that the hue was judged to be more violet than it actually was; underestimations (negative overestimations) indicate the hue was judged to be redder than it actually was. A planned T-test indicates that "8" was overestimated to a greater extent than was "L," t(84)=10.3, p<.01. The amount of misestimation for each of the symbols, with the exception of "E," significantly departs from 0.0, t(84)>2.8, p < .05. Overestimation differences of 0.42 are significant by Fisher's post hoc probabilistic least significant difference (PLSD), p<.05.
In general, the estimates for the more violet symbols ("8," "9," and "6" in Figure 2) were displaced to the violet end of the scale, relative to the redder symbols ("T," "E," and "L"). Estimations for "9," "6," "E," and "T" from the two-category condition can be compared to the estimates obtained from the one-category control condition. Relative to their controls, "T" was underestimated (t(130) = 8.7, p <.01), "E" was underestimated (t(130)=2.4, p < .05, "9" was overestimated (t(130) =2.7, p < .05), and "6" was insignificantly underestimated (t(130) = 1.4, p > .1). The pattern of results for "T," "E," and "9" was not compatible with a model that distorts an item's perception in the direction of the average dimension value for the category. Such a model would predict that "T" would be overestimated (toward the average hue for letters or symbols in general) and "E" and "9" would not be systematically misestimated. Thus, the influence of categorization seems to implicate a polarization or caricaturization (Goldstone, 1993) process that can be stated by: "If X belongs to a category that has a large (or small) value on a dimension relative to other categories, then distort X's perceived dimension value to make it appear larger (or smaller)."
The results from the control condition suggest that
"T"'s underestimation and "6"'s overestimation
were partially due to anchor contrast effects that moved endpoint
items in the direction opposite to the other presented items (King,
1980; Lane, 1965). However, the differences between the two-categories
condition and the one-category control indicate that category-insensitive
endpoint effects are not sufficient to explain the results. Overall,
Experiment 1 shows a category-level influence on symbols that
is compatible with a polarization process that makes objects that
belong to a category that contains relatively red (violet) objects
seem even more red (violet).
Experiment 2 further tested the influence of category-level information on hue perception by exploring the nature of the categories that influence perception. Because there is no explicit categorization task required of subjects and therefore no feedback about what categories are correct, it might be thought that there must be fixed constraints on what categories will influence perception. For example, if subjects sort objects into numerals and letters, one may argue that these categories must be automatic and context-independent ways of sorting the symbols.
Although no category feedback is provided, category-level information is still provided by the particular items shown to subjects (Clapper & Bower, 1994; Rumelhart & Zipser, 1985). If a set contains "8," "6," "E," and "T," then subjects might spontaneously create categories for numerals and letters, but if squares and triangles are also shown, then subjects might create categories of symbols and shapes. Consequently, Experiment 2 explored the possibility that, rather than fixed categories determining perceptual distortion, subjects spontaneously create context-dependent categories that influence hue perception.
As shown in Figure 3, within the low similarity set, there appears to be two categories of shapes: five-sided polygons and two-lined branches. If subjects are sensitive to these categories, then assimilation within these categories is expected to occur, as was found in Experiment 1. Accordingly, although Objects C and D are identically shaped and colored, D may be judged to be redder than C because D is assimilated toward the other polygon in the low similarity set. However, in the high similarity set, the identical polygons may not be judged as belonging to the same category. In this set, all of the shapes are five-sided polygons, and consequently this description lacks diagnosticity. Instead, the naturally constructed categories may be upward-pointing prongs and right-pointing prongs. If these categories are created, then C is no longer placed in the same category as any other polygon. In short, depending on the entire set of shapes in a context, the same two shapes may or may not be placed in the same category. The pattern of hue distortions can reveal the nature of the implicitly formed categories.
Subjects. Sixty-three undergraduate students from Indiana University were evenly assigned in a pseudorandom manner to the three presentation conditions.
Materials. Colored shapes, rendered on Macintosh IISI computers and shown in Figure 3, were used as materials. The hues of the shapes are shown in Table 1. The CIE coordinates for the 516 nm shapes were U'=.308 and V'=.422, for the 541 nm shapes were U'=.296 and V'=.399, and for the 565 nm shapes were U'=.259 and V'=.330.
The shapes were designed to fall into two shape categories: five-sided polygons and two conjoined thick lines. In addition, Shape A (B, C, and D are identical to A) was designed to be clearly distinguishable from the other polygons. A's prongs face to the right whereas the other polygons' prongs face upwards.
Procedure. The experiment's procedure was essentially the same as used in Experiment 1. While a colored shape was displayed, subjects modified a replica of the shape so that it possessed the same hue. The same scale and trial randomizations were used.
Three separate groups of subjects saw three different sets of shapes. The three shape sets are shown in Figure 3. Subjects saw 36 repetitions of each shape. On each presentation, the shape possessed the hue described in Table 1. The experiment took approximately 30 minutes for subjects to complete.
Results and Discussion
The two control conditions produced highly similar patterns of misestimation, and consequently, the results from Objects A and B were combined. The cleanest comparisons are between shapes that have identical hues. The experiment was explicitly designed to manipulate perceptions of the centrally hued shape (A, B, C, and D in Figure 3). The distribution of hue overestimations for these shapes is given in Figure 4. For these shapes, the set context had a significant influence on hue overestimation, F(2,40)=16.3, p<.01, and each of the average overestimations was significantly different from the others, t(20)>2.0, p<.05. As Figure 4 shows, relative to B from the second control condition, when the second polygon was added in the low similarity set, D was underestimated. However, C was overestimated relative to A when the same polygon was added to the context of other upward-pointing polygons. These results are consistent with the two polygons belonging to the same category in low similarity set, but the same shapes belonging to different categories in the high similarity set when the other items in the set were more similar to the upward-pointing polygon. The pattern of results in Figure 4 indicates both a repelling force of the upward-pointing polygon on C (relative to A's perceived hue) when they belonged to different categories, and an attracting force of the upward-pointing polygon on D (relative to B) when they were likely to be categorized together.
Experiments 1 and 2 demonstrate that subjects' hue
perceptions are distorted by the judged objects' shape categories.
The results indicate an automatic process of shape categorization
that exerts a perceptual influence on hue. The categorization
is automatic in the sense that shape categorization is not required
in order to accomplish the task, and yet still is naturally performed
in the course of a color perception task. Task counterbalancings
and same-hue controls eliminate explanations for perceptual distortions
that do not take into account the shape category of judged objects.
Experiment 1 shows that shape categories can influence hue perception
despite their lack of pre-experimental association to hue, and
Experiment 2 further shows that the shape categories that guide
hue perception are defined by the experimental context rather
than being preset.
Fixed and Created Categories
Previous research has shown that the color appearance of an object is influenced by the context to which an object belong (Bruner, Postman, & Rodrigues, 1951; Delk & Fillenbaum, 1965; Harper, 1953; White & Montgomery, 1956). Among other results, these experiments show that the color of an object appears stronger when the color is combined with a semantically appropriate object (e.g. a yellow hue is combined with a banana-shaped object).
Other research, including the current experiments, show that new contexts can influence perception of a dimension even though they have no pre-experimental associations to the dimension (Manis, Nelson, & Shedler, 1988; Marks, 1992; Marks & Warner, 1991; Wedell, in press). Although many of these experiments have shown contrast effects in the opposite direction to the effects currently obtained, they do argue for the creation of sub-contexts within the larger experimental context. For example, Marks shows that under conditions in which a 500-Hz tone was generally loud and a 2,500-Hz tone was generally soft, when two equally loud tones were played, the 2,500-Hz tone was judged to be much louder than the 500-Hz tone.
The current experiments further indicate that contexts cannot simply be based on single dimension values or preset categories. In Experiment 2, the results suggest that the two objects were placed in the same category when other objects were highly dissimilar to the two objects, but the same objects were placed in different categories when other objects were similar. Some researchers have found evidence for contexts that are defined by critical boundaries on physical dimensions (Marks & Warner, 1991) -- events that differ by more than X units are placed in different contexts. The current experiments indicate that, in some circumstances, the critical boundary X must be flexibly tuned to the stimulus set rather than being a single fixed value.
The context-defining categories seem not to be preset,
but to be learned during the experiment. Category learning can
affect the perceptual discrimination of color attributes, such
that attributes that are diagnostic for categorization become
perceptually sensitized (Goldstone, 1994). However, the current
experiments show that category learning can also proceed without
explicit categorization feedback (cf. Clapper & Bower, 1994).
The technique offers a potential tool for revealing people's
implicit categories that is relatively free of task demands and
the "problem solving stance" typical of many category
Contrast and/or Assimilation
In Experiment 1, objects with the same hues were perceptually distorted in the direction of the other objects in their context. This effect is in the opposite direction of Wedell (in press) and Marks (1992). Other researchers, however, have found results in the same direction as the current results (Manis, Biernat, & Nelson, 1991; Tafjel & Wilkes, 1963). The discrepancy between the results may be due to factors such as object extremity, object ambiguity, sequence effects, or resource demands that have been shown to mediate whether contrast or assimilation effects are found (Foti & Hauenstein, 1993; Herr, Sherman, & Fazio, 1983; Lockhead & King, 1983; Manis et al, 1988; Ward, 1990; Wedell, 1990). In particular, one possibility is suggested by Biernat, Manis, and Nelson's (1991) result that objective scales are more likely to produce assimilation effects than are subjective rating scales. In Experiments 1 and 2, precautions were taken to assure that subjects were making objective hue judgments rather than using an arbitrary scale.
Experiment 1 may reflect either a contrast or assimilation effect. For example, "L" in Figure 1 is underestimated relative to "8." This may be due to "L" being assimilated to the other letters, or to "L" being contrasted away from the numerals. Experiment 2 teases apart these two possible contributions to perceptual distortions, and suggests an influence of both processes -- assimilation within one's immediate category, and contrast from dissimilar categories. Assimilation within one's immediate category is suggested by comparing the low similarity set to the controls in Figure 3. When an object is introduced that is likely to be placed in the same category as D, then estimates of D move toward that object. Contrast from competing categories is suggested by comparing the low similarity set to the controls. Here, the introduction of the object that hypothetically belongs to a different category moves C's estimates away from the object.
One line of research (King, 1988; Lockhead, 1988)
argues that assimilation effects are found when two events are
perceived as a single whole, and contrast effects are found when
two separable events are perceived. The current results are consistent
with this claim. Objects that "go together" assimilate
to each other, and objects that are psychologically separated
are contrasted. Objects may go together because they are close
in time (Lockhead & King, 1983; Ward, 1990), space (Coren
& Girgus, 1980), or conceptual similarity (Coren & Enns,
1993). The current manipulation of experimentally presented categories
may simply be another way of altering the grouping of objects.
Subjects seem to create categories that will make a coarse split
between all of the presented objects. Within this coarse division,
objects seem to be assimilated. Objects seem to be contrasted
away from objects that fall on the opposite side of the coarse
The Locus of Categorization's Influence
The nature of subjects' color judgment task eliminates several potential explanations that do not posit a perceptual effect of object categories. Influences of memory encoding should be minimal. Although there is evidence that expectations can bias people's memories for colors (Belli, 1988), in the current experiments, only very short term memory is required. Both the displayed object and the manipulated object are viewed at the same time, and it take very little time to scan from one object to the other. Furthermore, because the subjects' task is to match two objects' colors, there are few biases associated with numeric rating scales.
More generally, the results may be compatible with
either perceptual encoding or response selection accounts of categorization's
influence on judgments. Other research has suggested that contrast
and assimilation effects occur early in the encoding stage of
processing (Wedell, 1994; Srull & Wyer, 1980). While the
current results concur with these findings to the extent that
the hue judgments required of subjects are direct and simple,
the results also indicate that categorization processes operate
before contrast and assimilation processes. Categorization, or
the creation of separate shape-based contexts within the overall
experimental context, is a necessary precursor for the obtained
perceptual distortions to occur.
One type of model for the type of perceptual distortions of "8" and "L" found in Experiment 1 maintains that judgments are combinations of particular item information and more general category-level information (Hellstrom, 1985). For example, Huttenlocher, Hedges, and Duncan (1991) model spatial judgment tasks by combining evidence from the particular object locations and from the general spatial category of the objects' location.
In Experiment 1, the end-point objects were contrasted away from their category means. Although incompatible with a model that simply averages specific and category-level means, this result is consistent with a model that treats category-level information as relational. For example, in Experiment 1, the letter category may have been represented, in part, as "red, relative to the numerals" (Goldstone, 1993). This relational, rather than absolute, category-level information could distort even the reddest letter so that it would appear redder than it was. The result is also consistent with Krueger and Rothbart's (1990) evidence that people distort item representations in order to make the items' categories highly differentiated and separated from each other.
In order to make hue distortions contingent on categories that are learned during the experiment rather than preset, it may be useful to borrow techniques from unsupervised connectionist networks (Rumelhart & Zipser, 1985). Such networks can create categories on the basis of the statistical properties of the presented objects even when no explicit categorization feedback is provided. Their model is able to create coarse categories when stimuli vary widely, and finer categories when all of the stimuli are highly similar. A model of the obtained hue distortions should be sensitive to object categories, and should allow these categories to depend on the similarity relations between the presented objects.
Abramov, I., & Gordon, J. (1994). Color appearance: On seeing red - or yellow, or green, or blue. Annual Review of Psychology, 45, 451-485.
Biernat, M., Manis, M., & Nelson, T. E. (1991). Stereotypes and standards of judgment. Journal of Personality and Social Psychology, 60, 485-499.
Bruner, J. A., & Postman, L. (1949). Perception, conception, and behavior. Journal of Personality, 18, 14-31.
Bruner, J. S., Postman, L., & Rodrigues, J. (1951). Expectations and the perception of color. American Journal of Psychology, 64, 216-227.
Clapper, J. P, & Bower, G. H. (1994). Category invention in unsupervised learning. Journal of Experimental Psychology: Learning, Memory, and Cognition, 20, 443-460.
Coren, S., & Enns, J. T. (1993). Size contrast as a function of conceptual similarity between test and inducers. Perception & Psychophysics, 54, 579-588.
Coren, S., & Girgus, J. S. (1974). Transfer of illusion decrement as a function of perceived similarity. Journal of Experimental Psychology, 102, 881-887.
Coren, S., & Girgus, J. S. (1980). Principles of perceptual organization and spatial distortion: The Gestalt illusions. Journal of Experimental Psychology: Human Perception & Performance, 6, 404-412.
Delk, J. L., & Fillenbaum, S. (1965). Differences in perceived color as a function of characteristic color. American Journal of Psychology, 78, 290-293.
Foti, R. J., & Hauenstein, N. M. (1993). Processing demands and the effects of prior impressions on subsequent judgments: Clarifying the assimilation/contrast debate. Organizational Behavior & Human Decision Processes, 56, 167-189.
Goldstone, R. L. (1994). influences of categorization on perceptual discrimination. Journal of Experimental Psychology: General, 123, 178-200.
Goldstone, R. L. (1993). Evidence for interrelated and isolated concepts from prototype and caricature classifications. Proceedings of the Fifteenth Annual Conference of the Cognitive Science Society. (pp. 498-503). Hillsdale, New Jersey: Lawrence Erlbaum Associates.
Harper, R. S. (1953). The perceptual modification of coloured figures. American Journal of Psychology, 66, 86-89.
Huttenlocher, J., Hedges, L. V., & Duncan, S. (1991). Categories and particulars: Prototype effects in estimating spatial location. Psychological Review, 98, 352-376.
Hellstrom, A. (1985). The time-order error and its relatives: Mirrors of cognitive processes in comparing. Psychological Bulletin, 97, 35-61.
Herr. P. M., Sherman, S. J., & Fazio, R. H. (1983). On the consequences of priming: Assimilation and contrast effects. Journal of Experimental Social Psychology, 19, 323-340.
King, M. C. (1981). Sequential effects in the variability of magnitude estimations of loudness. Dissertation Abstracts International, 41, 3609B. (University Microfilms No. DEN 81-05068.
King, D. L. (1988). Assimilation is due to one perceived whole and contrast is due to two perceived wholes. New Ideas in Psychology, 6, 277-288.
Kreuger, J., & Rothbart, M. (1990). Contrast and accentuation effects in category learning. Journal of Personality and Social Psychology, 59, 651-663.
Lane, H. (1965). The motor theory of speech perception: A critical review. Psychological Review, 72, 275-309.
Lockhead, G. R. (1988). Assimilation and contrast: Two processes or one? Response to King. New Ideas in Psychology, 6, 293-299.
Lockhead, G. R., & King, M. C. (1983). A memory model of sequential effects in scaling tasks. Journal of Experimental Psychology: Human Perception and Performance, 3, 461-473.
Manis, M., Biernat, M., & Nelson, T. F. (1991). Comparison and expectancy processes in human judgment. Journal of Personality and Social Psychology, 61, 203-211.
Manis, M., Nelson, T. E., & Shedler, J. (1988). Stereotypes and social judgment: Extremity, assimilation, and contrast. Journal of Personality and Social Psychology, 55, 28-36.
Marks, L. E. (1992). The contingency of perceptual processing: Context modifies equal-loudness relations. Psychological Science, 3, 285-291.
Marks, L. E., & Warner, E. (1991). Slippery context effect and critical bands. Journal of Experimental Psychology: Human Perception and Performance, 17, 986-996.
Niedenthal, P. M., & Kitayama, S. (1994). The heart's eye. San Diego: Academic Press.
Rumelhart, D. E., & Zipser, D. (1985). Feature discovery by competitive learning. Cognitive Science, 9, 75-112.
Srull, T. K., & Wyer, R. S., Jr. (1980). Category accessibility and social perception: Some implications for the study of person memory and interpersonal judgments. Journal of Personality and Social Psychology, 38, 841-856.
Tafjel, H., & Wilkes, A. L. (1963). Classification and quantiative judgment. British Journal of Social Psychology, 54, 101-114.
Ward, L. M. (1990). Critical bands and mixed-frequency scaling: Sequential dependencies, equal-loudness contours, and power function exponents. Perception & Psychophysics, 47, 551-562.
Wedell, D.H. (1990). Methods for determining the locus of context effects in judgment. In J. P. Caverni, J. M. Fabre & M. Gonzales (Eds.), Cognitive Biases (pp. 285-302). New York: Elsevier Science Publishers.
Wedell, D. H. (1994). Contextual contrast in evaluative judgments: A test of pre- versus post-integration models of contrast. Journal of Personality and Social Psychology, 66, 1007-1019.
Wedell, D. H. (in press). Contrast effects on paired comparisons: Evidence for both stimulus and response based processes. Journal of Experimental Psychology: Human Perception and Performance.
White, C. W., & Montgomery, D. A. (1976). Memory colors in afterimages: A bicentennial demonstration. Perception & Psychophysics, 19, 371-374.
Whorf, B. L. (1941). Languages and logic. in J. B. Carroll (ed.) Language, Thought, and Reality: Selected papers of Benjamin Lee Whorf. MIT Press (1956), Cambridge, Mass. (pp. 233-245).
von Hippel, W., Hawkins, C., & Narayan, S. (1994).
Personality and perceptual expertise: Individual differences
in perceptual identification. Psychological Science, 5,
I would like to thank Gregory Lockhead, Doug Medin, Paula Niedenthal, Robert Nosofsky, Richard Shiffrin, Jim Sherman, and Doug Wedell for many useful comments. The research was supported by National Science Foundation grant SBR-9409232.
Correspondences should be addressed to Robert Goldstone, Psychology Department, Indiana University, Bloomington, IN. 47405. The author's email address is firstname.lastname@example.org.
Figure 1. Stimuli used in Experiment 1. The shading of a symbol indicates its proportion of red, relative to violet, hue.
Figure 2. Results from Experiment 1. Overestimations indicate that a hue was judged to be more violet than it actually was. Underestimations occur when a hue was judged be redder than it actually was. Error bars show the standard deviations associated with each overestimation.
Figure 3. Four conditions used in Experiment 3. The shading of a shape indicates its proportion of red, relative to violet, hue. Shapes A, B, C, and D are identical. The high similarity set is the identical to Control 1, with addition of one upward-pointing polygon. The low similarity set is identical to Control 2, with the addition of the same upward-pointing polygon.
Figure 4. Results from Experiment 2. The set context influences the hue perception of objects that have identical hues and shapes. Although A, B, C, and D are identically shaped and hued, as shown in Figure 3, they are presented in different contexts, and are consequently given different hue estimates.