Detectability of increments and decrements in spectral profiles

In auditory profile analysis, the subject typically has to discriminate a standard spectrum consisting of widely spaced equal‐amplitude sinusoids from a comparison, in which one or several components have been elevated relative to the others. According to the widely accepted multichannel model, however, it should not matter whether the spectral shape alteration consists of incrementing or decrementing a component by a small number of decibels. Thus, in order to investigate the generality of the hypothesized mechanism, and to explore the role of pitch salience cues, which might be thought to favor increments to single components, two experiments comparing the detectability of intensity increments and decrements in spectral profiles were conducted. In experiment 1, in which the number of components was varied from 3 to 15 within a fixed frequency range, decrements to the central 800‐Hz component produced significantly larger thresholds than did increments. Experiment 2, which employed three‐tone complexes spanning different frequency ranges, revealed that this effect is entirely due to the spectral density of the frequency components: As the distance between the variable signal component and the flanking reference sinusoids decreased, the increment/decrement disparity became larger. These effects may—at least qualitatively—be accounted for when peripheral excitation patterns are considered, suggesting that local decrements are constrained more than increments by the excitation intruding from flanking stimulus components.