Task constraints and visual feature binding.

Tel.: 00 44 1904 433135; fax: 00 44 1904 433181; e-mail: ptql@york.ac.uk. Experimental variations on the paradigm described by Kahneman, D., Treisman, A., Gibbs, B.J., 1992. Cognitive Psychology 24, 175-219 are reported. With these techniques two coloured letters are briefly presented, in a preview field. Following a short blank interval, a single coloured letter (the target) is presented. In the first experiment, reported here, subjects named the target letter, and in the second they named the target's colour. In neither of the experiments did the colour relationships between the target and preview items influence performance, but in both cases the shape relationships were important. Further experiments explored the notion that performance in the tasks was only an imperfect reflection of underlying perceptual processes because of the constraint of having to name the letters. When subjects were presented with unfamiliar characters and named the target's colour, then the colour relationships in the displays were important; residual effects of shape preview were also found. In final studies visually similar letters were used and different groups of subjects either named the target letter or the colour of the target. Of main interest was that the colour relations between the items exerted an effect when subjects named the target's colour. The theoretical and methodological implications of these results are discussed and the limitations of the original experiments are considered in the light of the reported findings.