Abstract
Aim: A question that continues to engage vision researchers is whether contrast transduction noise is additive (fixed with respect to stimulus contrast) or multiplicative (increases with stimulus contrast). Previous attempts to answer this question have been based on an analysis of contrast discrimination thresholds, i.e. JNDs. JNDs however are determined by both the slope of the transducer function as well as the magnitude of internal noise at the point of discrimination. It is difficult to tease these two factors apart, even when employing both the threshold and slope of the psychometric function. One solution is to measure the shape of the transducer function directly using a suprathreshold scaling procedure that is robust to whether internal noise is additive or multiplicative. This will result in a set of contrasts defined at equal perceptual intervals, or ‘equal-perceived-differences’ (EPDs). If the functions relating JNDs and EPDs to contrast magnitude superimpose when suitably scaled this suggests that transduction noise is additive, whereas if the JNDs increase more rapidly than EPDs, this suggests that transduction noise is multiplicative. Method and Results: Whittle (Vis. Res., 32, p1493–1507, 1992) asked subjects to adjust the luminances of a series of patches on a uniform background to form an equal-interval brightness scale, i.e. a set of EPDs. Although not aiming to address the additive-versus-multiplicative noise question, Whittle compared the EPDs with JNDs obtained from a previous study (Whittle, Vis. Res., 26, p1677–1691, 1986). He found that when the JNDs were suitably scaled they neatly superimposed on the pattern of EPDs. Conclusion: Contrast transduction noise for incremental and decremental patches on uniform backgrounds is additive not multiplicative.
Supported by a Canadian Institute of Health Research (CIHR) grant # MOP-11554 given to F.K.