Inverse attribute functions and the proposed modifications of the power law

Three prominent modifications of Stevens’ power law, designed to account for the curvature at low stimulus intensities, were examined in the light of previously reported inverse attribute functions and five new experiments scaling loudness and softness by magnitude estimation. One hundred and seven students served as observers in the five experiments. Lack of curvature in the inverse functions at low stimulus intensities, plus the lack of parameter invariance, pointed to the inadequacy of the effective threshold and physiological noise modifications. The additive constant model of zero-point response bias, previously advanced by Irwin and Corballis (1968)and McGill (1960)was found more satisfactory. A discussion of the implications of the research findings for subsequent formulations of the modified power law was included.