Roughness reflects rate of change, so fast modulation roughens while slow change does not

Evidence

Because masking level is an effective measure for determining excitation, it can be used to estimate the excitation-level differences produced by amplitude modulation. This procedure incorporates the two main effects already discussed, i.e. frequency and temporal resolution. The temporal masking patterns outlined in Figs. 4.24, 4.25, and 4.27 show the temporal effects and indicate values of 11£ that can be used as a measure to estimate differences between the maximum and the minimum of the temporal masking pattern. This temporal masking depth, 11£, becomes larger for lower modulation frequency. If roughnesses were determined only by this masking depth, then one would expect the largest roughness at the lowest modulation frequencies. This is not the case, indicating that roughness is a sensation produced by temporal changes. A very slow change does not produce roughness; a quick periodic change does, however. This means that roughness is proportional to the speed of change, i.e. it is proportional to the frequency of modulation. — Professor Dr.-Ing. Hugo Fastl, Professor Dr.-Ing. Eberhard, Psychoacoustics- Facts and Models, 1461

Categorization

• DEEPTHINKING: The argument that roughness is a derivative-like sensation — proportional to speed of change, not magnitude of change — is a genuinely elegant and counterintuitive way to build a perceptual metric.
• XREF: Connects to the roughness/fluctuation-strength models and to masking-depth measures; also parallels how fluctuation strength vs roughness are distinguished in this same book.