The Rise and Fall of `Volume': A Case Study in Psychophysical Measurement

This talk examines the changing norms of psychophysical measurement through the example of auditory “volume.” I argue that the trajectory of changing interest in volume is best explained by a change in attitudes toward our epistemic access to phenomenal attributes intimately connected to the introduction of multidimensional scaling techniques. In the early 20th century, it was discovered that subjects could consistently assign volume as an attribute of auditory stimuli distinct from loudness or intensity. This prompted speculation that volume might be a distinct auditory quality, or even a truly amodal perceptual quality assignable to stimuli for any sensory modality (might their be “smell volume,” for instance?). The lynchpin of research on volume was Stevens' (1934) discovery of a systematic relationship between volume, loudness, and auditory “density” (Boring, 1942). Stevens’ result motivated further systematic investigation of volume in the following decades, e.g. Thomas’ (1949) equal volume contours, Terrance and Stevens’ (1962) volume scale, and Gulick’s (1971) determination of subjective volume scales (Gulick, Gescheider, and Frisna, 1989).Nevertheless, by the late 20th century, research on volume as an auditory quality had largely disappeared. If volume is mentioned at all in contemporary psychophysical texts, it is as a mere historical footnote. While the legitimacy of the claim that volume is an attribute of sound experience is typically not questioned, neither is it taken to be of interest for a modern study of sound perception. What explains this change in attitude toward volume?Two possible explanations can be found in the classic criticisms of late 19th and early 20th century psychophysics: (1) measurement for measurement's sake (i.e. without a guiding theoretical framework) is inadequate for approaching truth—mere measurement of volume is no indication it is meaningful as a theoretical concept for understanding auditory perception; (2) overtraining of subjects contaminates data, producing “robust” but essentially meaningless effects—consistency in volume judgments was a mere artifact of overtrained subjects. I demonstrate that neither of these criticisms applies convincingly to volume research.I argue that the decline of interest in volume may be better understood by placing it in the context of the change in data analysis techniques for psychophysics, in particular the introduction of multidimensional scaling. The program of Titchener, Stevens, and other early 20th century psychophysicists required the experimenter to posit fundamental perceptual attributes before they could be measured through experiment. In contrast, multidimensional scaling allowed researchers to instruct subjects to merely make judgments of similarity—the perceptual attributes which guided those judgments could then be derived mathematically through data analysis. The upshot of this new technique was that access to perceptual quantities was (perceived as) no longer hypotheticodeductive, but rather purely empirical. Nevertheless, I conclude, the new empiricism of psychophysics fails to answer key questions about the nature of perceptual experience—for instance, the reduction of volume and loudness judgments to a single “dimension” via multidimensional scaling leaves unexplained the initial data point, that the two qualities are systematically distinguishable by subjects.