Color naming and the effect of language on perception

A classic nature-versus-nurture debate in cognitive science concerns the relation between language and perception. The universalist view holds that language is shaped by universals of perception, while the opposing relativist view holds instead that language shapes perception, in a manner that varies with little constraint across languages. Over the years, consensus has oscillated between these two poles. In this talk, I argue that neither position is fully supported. I argue moreover that the universalist/relativist opposition itself should be resisted as a conceptual framework, since it paints with too broad a brush, and obscures interesting realities. I argue this general point using two case studies in the naming and perception of color. Color naming is near-optimal An influential universalist view of color naming holds that color categories across languages are organized around the universal focal colors black, white, red, green, yellow, and blue [1]. A recent relativist challenge holds in contrast that there are no such universal foci, and that color categories are instead defined at their boundaries by largely arbitrary linguistic convention [2]. Both of these views are partly supported by @ and partly challenged by @ existing data, which show universal tendencies in color naming [3], coupled with interesting cross-language variation in Dust where category boundaries are drawn [4]. In an attempt to resolve this tension, I argue for a third view, based on a proposal by Gameson and DIAndrade [J]K that color naming across languages reflects optimal or near-optimal partitions of an irregularly shaped perceptual color space [L]. My colleagues and I have formalized this proposal in terms of a wellformedness measure that captures the extent to which a given categorical partition of color space maximizes perceptual similarity within color categories and minimizes it across categories [N]. We propose that the color naming systems of the worldIs languages correspond to maxima or near-maxima in wellformedness @ i.e. to theoretically optimal color naming systems. We used simulations to create such theoretically optimal color naming systems, with nP3,4,J,L categories. We initialized each simulation by randomly assigning each of a discrete set of points in perceptual color space to one of the n categoriesQ we then adDusted these category labels through steepest ascent in wellformedness, until a maximum was reached. The results are displayed in Figure 1, together with selected languages from the World Color Survey (WCS) database [X]. Each color naming system is displayed on a standard color grid in which columns represent hues, and rows lightness. More broadly, we found that across the 110 languages of the WCS, color naming tended to be shaped in part by well-formedness. At the same time, our model also suggests where linguistic convention may get some wiggle roomK there are often several similar but different partitions that are roughly equally well-formed. Thus, the model suggests a specific middle ground between \nature] and \nurture] in color naming across languages @ one that corresponds to neither the standard universalist nor the standard relativist account. Figure 1. Model prediction (left) compared with selected WCS languages (right), for n=3,4,5,6 categories. Whorf hypothesis in the right visual field The Whorf [9] hypothesis, a classic relativist proposal, holds that language affects perceptionK when languages categorize reality differently, speakers of those languages should correspondingly perceive reality differently. There are a number of studies that support this idea, and a number that do not. I shall argue not for one side or the other of this existing debate, but instead for a novel proposalK that language affects perception primarily in the right visual field (RVF), and much less if at all in the left visual field (LVF). This idea, which is suggested by left hemisphere language dominance, is unanticipated by the traditional univeralist/relativist framing of the debate. Gilbert, Regier, Kay, and Ivry [10] tested this idea in the context of color perception. We reasoned that if language affects perception primarily in the RVF, one would predict three things. First, discrimination of stimuli with different names (e.g. a hue of \blue] and one of \green]) should be faster in the RVF than in the LVF, since the difference in names will heighten perceptual differences in the RVF. Second, the discrimination of stimuli with the same name (e.g. two different hues of \green]) should be slower in the RVF than in the LVF, since the sameness of the name will impede perceptual discrimination in the RVF. And third, this overall pattern should be disrupted by concurrent tasks that interfere with verbal processing, but not by concurrent tasks of comparable difficulty that only interfere with non-verbal processing. We found support for all three predictions, in experiments with English-speaking subDects. Drivonikou et al. [11] have replicated these findings. 15th Color Imaging Conference Final Program and Proceedings 1 Conclusions Neither of these findings @ that color naming is near-optimal, and that language affects color perception primarily in the RVF @ is anticipated by the traditional universalist-versus-relativist framing of the debate over language and perception, and neither sits particularly comfortably with it. Instead, these findings suggest that the world is a more interestingly differentiated place than that framing might lead one to expect. Acknowledgments I would like to acknowledge and thank my collaborators on this workK Aubrey Gilbert, Richard Ivry, Paul Kay, and Naveen Khetarpal. This work was supported by National Science Foundation Grants 041X2X3 (to TR) and 041X404 (to PK).