Mechanisms of Frequency Response Extension

Abstract

There is considerable evidence that subjects in a traditional psychophysical experiment act as narrow-band receivers tuned to the frequency of a known sine-wave stimulus. Obviously wave forms in real life are complex rather than simple, and the wave forms we hear can seldom be predicted exactly. A series of experiments was conducted attempting to arrive at a model for hearing when the sine-wave signal to be presented is not absolutely predictable, but might be one of two alternative frequencies. For each experimental run of 100 trials, one of two possible tones was randomly selected for presentation on each trial in a two-alternative forced-choice detection situation: one of the tones was always 1000 cy and the other had a frequency difference of 20, 100, 200, or 500 cy above or below, giving eight pairs of tones investigated. The intensities of the tones were equated for detectability in a standard single tone preliminary detection experiment. All eight pairs were run at two different levels of background noise using different signal intensities. Under these conditions differential predictions of observers' behavior with an uncertain signal can be made within the framework of the theory of signal detectability for a number of alternative models calculated to account for the extension of frequency sensitivity. The data are compared to predictions from a scanning model, a multiplex model, a threshold model, and a variable band model. (This research was carried out under sponsorship of the Operational Applications Laboratory, Air Force Cambridge Research Center and Evans Signal Laboratory, Signal Corps Engineering Laboratories, Fort Monmouth, New Jersey.)