‘Sensation’ is the subjective experience of a physical stimulus. This book traces the sources of two ideas how to measure sensation — Fechner's Law and Stevens' Power Law — and examines the evidence and arguments in their support. Fechner's law is based on discriminations between separate stimulus magnitudes (Weber's Law), together with the additional premise that just noticeable differences are subjectively equal. But such discriminations can be modelled, with equal precision, using Gaussian noise of power density proportional to the stimulus magnitude. Moreover, Gaussian noise accommodates a ... More

‘Sensation’ is the subjective experience of a physical stimulus. This book traces the sources of two ideas how to measure sensation — Fechner's Law and Stevens' Power Law — and examines the evidence and arguments in their support. Fechner's law is based on discriminations between separate stimulus magnitudes (Weber's Law), together with the additional premise that just noticeable differences are subjectively equal. But such discriminations can be modelled, with equal precision, using Gaussian noise of power density proportional to the stimulus magnitude. Moreover, Gaussian noise accommodates a wider range of phenomena and relates to the electrophysiology of sensory neurons. Stevens' Power Law describes the assignment of numbers to individual stimulus magnitudes. But there are empirical difficulties if one looks for a realization of the power law in elementary neural function, or uses it as a definitive measure of sensation, or merely as an operational characterization of how people use numbers. Each individual stimulus magnitude is judged relative to the stimulus presented on the preceding trial and the number assigned to it. The book finishes with a theory of sensory judgment, applicable to category judgment and absolute identification as well as magnitude estimation. The comparison of each stimulus in an experiment with its predecessor is little better than ordinal — ‘greater than’, ‘about the same as’, ‘less than’. This leads to great variability in estimates of stimulus magnitude. If the stimulus values lie in a geometric series and if participants judge ratios, the only meaningful relationship between stimulus and response is the linear regression of log numerical assignment on log stimulus magnitude, which equates to Stevens' Power Law. However, the ordinal nature of the comparison between one stimulus magnitude and another means that there is no metric for the measurement of sensation.