The Effect of Name Category and Discriminability on the Search Characteristics of Colour Sets

Abstract

Within (and between) cultures, people tend to agree on which parts of colour space are easiest to name and what the names for these regions are. Therefore it is likely that the manipulation of ease of naming (nameability) of colours should change performance in tasks where categorisation by colour name is important. More specifically, highly ‘nameable’ colour sets should lead to better performance than metrically equivalent but less categorically distinct sets, when the task requires categorisation. This hypothesis was investigated by testing observers on a name-based task, the naming and subsequent identification by name of colour sets with up to sixteen members. These sets were designed to be easy to name (nameable), maximally discriminable, or matched discriminable. The first were derived from previously generated data, the second by a standard algorithm to space colours widely in colour space, and the latter by closely matching their metric characteristics to those of an easy-to-name colour set. This final condition was metrically (but not categorically) equivalent to the nameable set. It was found that sets designed to be nameable did indeed lead to superior performance as measured by response times, confidence ratings, and response accuracy. Perceptual colour similarity, measured by a ΔE metric, did not predict errors. Nameability may thus be a valid, manipulable, aspect of sets of colours, and one which is not otherwise duplicated in the metric characteristics of such sets.