A Comparison of Partition Scaling and Magnitude Estimation for Brightness Scaling

Abstract

ABSTRACT Several psychophysical procedures exist to capture the relationship between the magnitude of an optical stimulus (e.g. luminance) and the corresponding perceived brightness. A commonly used psychophysical method is magnitude estimation (ME). However, some drawbacks of this method have been identified in literature: the method could be prone to some biases such as order effects, range effects and centering bias. On the other hand, partition scaling (PS) is one of the oldest, yet rarely used, psychophysical method to derive a similar luminance-brightness relationship. In the present study, the two methods are compared in terms of robustness and susceptibility to possible biases. Psychophysical experiments with simple achromatic discs were set up to obtain a brightness scale as a function of luminance using ME. Regarding PS, very similar experiments have been conducted in a previous study. The experiments were conducted for four luminance ranges: three subranges (low-, mid- and high-range of luminance) equally divided from 5 to 175 cd/m2 and one overlapping full-range. Results indicate that observers had difficulties in accurately estimating the mid- and high-ranges for both methods, because the brightness between discs was too similar. Perceptual brightness scales could be obtained for the full- and low-range, although strong evidence was found for a range bias using ME. When pooling the full- and low-range data, both ME and PS results converge. Results show that PS is a valid alternative compared to ME for deriving brightness scales, with the advantage that it is more reliable and resilient to range bias.