The paper proposes a number of models to examine through what mech-anisms a population of autonomous agents could arrive at a repertoire of perceptually grounded categories that is sufficiently shared to allow successful communication. The models are inspired by the main approaches to human categorisation being discussed in the literature: nativism, empiricism, and culturalism. Colour is taken as a case study. Although the paper takes no stance on which position is to be accepted as final truth with respect to hu-man categorisation and naming, it points to theoretical constraints that make each position more or less likely and contains clear suggestions on what the best engineering solution would be. Specifically, it argues that the collective choice of a shared repertoire must integrate multiple constraints, including constraints coming from communication.

A Bayesian computational model is described, which is able to account for the acquisition of the meanings of basic colour terms by children learning their first language. Examples of colours named by particular colour terms are stored in a conceptual colour space, and Bayesian inference is used to determine the extent of the extension of each colour term based upon these examples. This method can be extended to create a fuzzy set based denotation for each colour term by calculating the probability that each point in the colour space comes within the extension of each colour term. The learned categories show the prototype structure characteristic of colour terms, with there being a single best example of the category, marginal members of the category, and with intermediate colours being members of the category to a greater or lesser extent. This approach has the advantage over previous approaches to colour term semantics of being both flexible enough to account for the full range of colour term systems seen in the world's languages, while at the same time providing a precise and explicit account of how children may accomplish the task of learning colour terms. Further experiments reveal that learning is successful even when as many as fifty percent of the colour term examples presented to the model are erroneous, demonstrating that the theory is robust, and can account for the acquisition of colour terms in realistic as well as idealised situations.

ABSTRACT While recognizing the theoretical importance of context, current research has treated naming as though semantic meaning were invariant and the same mapping of category exemplars and names should exist across experimental contexts. An assumed symmetry or bidirectionality in naming behavior has been implicit in the interchangeable use of tasks that ask subjects to match names to stimuli and tasks that ask subjects to match stimuli to names. Examples from the literature are discussed together with several studies of color naming and basic emotion naming in which no such symmetry was found. A more complete model of naming is proposed to account for flexible mapping of names to items. Principles of naming are suggested to describe effects of stimulus sampling, differing access to terms, task demands, and other impacts on naming behavior. 1.

An expression-induction model was used to simulate the evolution of basic color terms in order to test Berlin and Kay's (1969) hypothesis that the typological patterns observed in basic color term systems are produced by a process of cultural evolution under the influence of universal aspects of human neurophysiology. Ten agents were simulated, each of which could learn color term denotations by generalizing from examples using Bayesian inference. Conversations between

Color naming and sunlight

Categorization of colour has been widely studied as a window into human language and cognition, and quite separately has been used pragmatically in image-database retrieval systems. This suggests the hypothesis that the best category system for pragmatic purposes coincides with human categories (i.e. the basic colours). We have tested this hypothesis by assessing the performance of different category systems in a machine-vision task. The task was the identification of the odd-one-out from triples of images obtained using a web-based image-search service. In each triple, two of the images had been retrieved using the same search term, the other a different term. The terms were simple concrete nouns. The results were as follows. (1) The odd-one-out task can be performed better than chance using colour alone. (2) Basic colour categorization performs better than random systems of categories. (3) A category system that performs better than the basic colours could not be found. (4) It is not just the general layout of the basic colours that is important, but also the detail. We conclude that (i) the results support the plausibility of an explanation for the basic colours as a result of a pressure-to-optimality, and (ii) the basic colours are good categories for machine vision image-retrieval systems.

Recent research has questioned the universal basis of color categorization and has instead emphasized cross-linguistic variation in boundaries of color categories. We propose that these cross-linguistically varying boundaries can be predicted from near-universal focal colors within the categories. In support of this proposal, we show that: (1) best example choices for color terms in 110 unwritten languages cluster near the prototypes for English white, black, red, green, yellow, and blue – we take these 6 points in color space to approximate universal foci; (2) best example choices cluster more tightly across languages than do category centroids; and (3) a computational model can predict color term boundaries from labelings of best examples reasonably well, for several languages, including one that has been taken to counterexemplify universal tendencies in color naming. Overview It has long been held that there are universal tendencies in color naming, in that linguistic color categories are organized around universally-shared focal points, or prototypes, in color space. Berlin and Kay (1969; B&K for short) showed that the best examples of color terms across a sample of 20 languages seemed to cluster in color space. That study and subsequent work (Kay & McDaniel, 1978; Kay & Maffi, 1999) showed that the most reliable and widespread of these clusters correspond to the six Hering primaries: white, black, red, green, yellow, and blue – suggesting that these points in color space may constitute a universal foundation for color naming. These foci in color space have also appeared to be cognitively privileged, in non-linguistic tasks with speakers of languages that have

This thesis investigates language acquisition and evolution, using the methodologies of Bayesian inference and expression-induction modelling, making specific reference to colour term typology, and syntactic acquisition. In order to test Berlin and Kay's (1969) hypothesis that the typological patterns observed in basic colour term systems are produced by a process of cultural evolution under the influence of universal aspects of human neurophysiology, an expression-induction model was created. Ten artificial people were simulated, each of which was a computational agent. These people could learn colour term denotations by generalizing from examples using Bayesian inference, and the resulting denotations had the prototype properties characteristic of basic colour terms.

Abstract not found

In this paper we examine the judged similarity among the eight basic focal colors, and their names, among female and male Chinese (n D 68) and English (n D 52) speaking respondents. The major � ndings are: (1) all respondents share approximately sixty percent of their knowledge of the judged similarity structures of both semantic and perceptual tasks, (2) there are genuine individual differences among respondents that account for about fourteen percent of their knowledge on average, (3) there are small but statistically signi � cant gender differences that come to about three percent on average, (4) there are small but statistically signi � cant differences between Chinese and English respondents of about one-and-a-half percent, (5) there are differences in the semantic structure of the names of colors as compared to the judgments of the color samples that amounts to about � ve percent, and (6) there is about a three percent difference in the paired comparison task and the triads task. The results place strong constraints on theories relating to individual differences, linguistic relativity, and the relation of perceptual and semantic structures for colors. Universals, language, cognition.