Magnitude estimation and sensory matching

Abstract

Scaling procedures are commonly used to generate sensory matches, in which each match is defined as the set of stimuli that yield a constant average judgment. Five experiments evaluated the principle of "matching-by-scaling" in the magnitude estimation of loudness: (1) Matches derived from magnitude estimates of the loudness of qualitatively different sounds (500-Hz and 2500-Hz tones) change dramatically with changes in the relative intensity levels (contexts) of the stimuli being judged. (2) Small changes in matches apparently can arise from sequential processes of judgment, even when the stimuli are qualitatively identical (all 500 Hz). (3) Even when the effects of sequential processes are minimized by averaging, changing the context induces large changes in loudness matches derived from magnitude estimates of 500-Hz and 2500Hz tones. (4) Changing contextual sets of qualitatively identical stimuli (all 500 Hz) presented to different ears has little effect on interaurally matching values. (5) Matches between 500-Hz and 2500-Hz tones shift with changes in the proportions of presentations of various intensity levels, even when the intensity levels themselves do not change. The results imply that judgments of perceived intensity of qualitatively different stimuli do not always obey the principle that equal sensations yield equal response, and, accordingly, that experiments using scaling methods to generate matches should compare stimuli whose levels of subjective magnitude are roughly the same. The results can be described by a quantitative model containing a parameter that depends on qualitative similarity and that may therefore reflect the degree of overlap in activated sensory channels. © 1988 Psychonomic Society, Inc.