Abstract
Difference measurement is a little-used but potentially valuable method for studying suprathreshold vision. We used it to investigate the perceptual salience of borders between adjacent uniformly coloured regions. In these experiments, the observer viewed pairs of colours C1, C2 presented simultaneously in two rectangles, sharing a common border, centred on a neutral background. On each temporal 2AFC trial, the observer saw two colour pairs (C1, C2) and (C3, C4), and was asked to judge which of the two colour borders was more salient (stronger). In a single condition of the experiment, the colours C1, C2, C3, C4 were drawn from a set of ten colours constrained to lie along a fixed line segment in LMS space extending from the neutral background point (NBP) in a particular direction in LMS colour space.
Let D i denote the distance of C i from the NBP in LMS space. A difference measurement model accounts for such data if and only if there is an increasing ‘scale’ f(D) such that (C1, C2) is judged more salient than (C3, C4) precisely when f(D2) - f(D1) > f(D4) - f(D3). For each of three observers and eight directions, we found that estimated functions f(D) were concave, resembling power functions with exponents between 0.64 and 1.
Get full access to this article
View all access options for this article.