Standard loudness analysis discards response variability that probabilistic models can exploit

Evidence

A listener’s CLS function, generally computed from the median stimulus level for each response category, is the measure typically used in analysis. This standard approach treats the variability in the response data as something to be removed. We propose to instead model the probabilistic nature of listener responses; these probabilistic representations can be used to improve loudness measurements and further our understanding of the mechanisms underlying this perception. — Dijk, Pim van, Physiology, Psychoacoustics and…, 2513

Categorization

• SEED: Angle: the information lost when scientists average away noise in perceptual measurement — a general essay on why within-subject variability carries signal.
• XREF: Connects to the role of institutions / information-literature theme that averaging discards distributable information.