Masking of Tonal Signals

Many of the phenomena of masking can be explained on the basis of two models, one for monaural listening, and the other for binaural. The monaural model is the familiar narrow band-pass filter followed by a detector responsive to changes in output level. The binaural model is a series of coincidence detectors associated with a delay-network capable of matching delay in the stimulus with a delay in the neural path. The two models have proved helpful in understanding the phenomena of tonal masking, and have led to a number of predictions which have been subsequently verified by experiments reported here. Some of the new findings are related to monaural masking and some to binaural. Among the latter are the fact that masking-level differences can be observed in the masking of one pure tone by another when a short signal is employed, and that a binaural signal can be heard in the presence of uncorrelated noise at the two ears better than a monaural signal can be heard against noise presented monaurally, again provided that a short signal is employed. Many studies of the masking of tones have appeared during the past few years. They have been characterized by close agreement in results wherever they have had common conditions, and they provide a large body of dependable quantitative information. With this body of fact there also grew, at least in the minds of present writers, a considerable state of bewilderment. Many of the facts appeared to contradict inferences drawn from others. The present paper discusses a number of these facts and attempts to encompass them under a few explanatory principles.