Measurement of low-signal suppression and other parameters of the forced-choice psychometric function

Empirical forced-choice psychometric functions often depart in shape from those based on the theory of normal sensory-excitation distributions and optimal decision processes. The typical departure is that the detectability of small signals is too low compared to that of high signals. Some possible explanations were reviewed and it was emphasized that the source of the effect is not known with certainty. An extra parameter resembling a threshold was introduced for the measurement of the effect, and a method of moments was proposed for estimating the two parameters of the resulting two-alternative forced-choice psychometric function. The method was tried out with success in the task of detecting a 0.1-sec luminance increment to a steady adapting luminance of a large visual target. There were large amounts of low-signal suppression, indicating the advisability of reviewing some current psychophysical practices.