Abstract
Ozgen and Davies (2002, JEP:G, 131, 477-493) provided evidence in favour of learned categorical perception of colour, both in terms of hue and lightness-defined categories. To throw further light on potential differences between hue and lightness-based category learning, their findings were reassessed in an experiment that closely followed their methodology, with hue and lightness learners training on a novel category boundary within the green region of colour space. After training, hue and lightness learners as well as untrained controls (15 participants per group; 45 in total) performed delayed same-different discrimination for both lightness and hue pairs. Ozgen and Davies used stimuli from the Munsell space. Here, stimuli were selected from CIE Lab space - equated in terms of chroma, spanning roughly the same area of lightness and a somewhat wider set of hues in the lime-green/green area of colour space than in Ozgen and Davies (147°-171° as opposed to 159°-173°, maintaining the same distance ΔE=6). In addition to discrimination data, we monitored errors made during learning and asked participants to report on any colour-labelling strategies. Main findings are as follows: 1) For lightness learners, mistakes during training accumulated evenly around the newly learned category boundary, but for hue learners they were distributed non-uniformly, in accordance with the Bezold-Brucke effect; 2) almost all learners reported using labelling strategies during both training and discrimination, with hue learners mainly using green/blue and lightness learners mainly using light/dark. The “blue” label was associated with the greener colours (i.e., the ones closer to unique green). The results indicate that existing colour labels can play a role in perceptual learning and that this labelling does not have to conform to everyday naming (i.e., blue is used for stimuli that would otherwise be named as green).