The human auditory system is adept at handling concurrent and overlapping sound sources-such as multiple speakers or speech heard in background clutter-using their spatial, temporal, and spectral features to segregate them into separate auditory streams for further analysis (Bregman, 1990;Sussman, eponiené, Shestakova, Näätänen, & Winkler, 2001). The subject of our present inquiry is what happens when the human auditory system is confronted with a single message that is alternated rapidly between two spatial sources.The first demonstration of the consequences of this effect was by Cherry and Taylor (1954), who used an electronic switch to rapidly alternate speech back and forth between the two earpieces of dichotic earphones. They reported that intelligibility progressively declined with increasing rates of alternation, becoming poorest at approximately 3 to 5 complete switching cycles per second (cps), a rate corresponding to about 167 to 100 msec per ear. As the alternation rate was increased beyond this point, intelligibility paradoxically improved, thus yielding a V-shaped function of intelligibility decline and intelligibility recovery over successively increasing alternation rates. The size of the decrement at this critical rate is smaller when the speech materials are predictable enough to allow missed elements to be inferred from the linguistic context (cf. Speaks & Trooien, 1974;Wingfield & Wheale, 1975a) and larger when perceptual processing is made more difficult by mixing masking noise with the speech signal (Rupf, Hughes, & House, 1971). The basic V-shaped intelligibility function remains, however, as interaural alternation rates are progressively increased.The intelligibility loss at the critical rate of 3 to 5 alternating cps is intriguing because the speech is always present, either in one ear or the other. Cherry and Taylor (1954) attributed this loss to the time it takes a listener to switch attention from one ear to the other in order to follow the shifting signal. If each shift of attention from one ear to the other is accompanied by a perceptual "dead time," during which no usable information is available from either ear, as alternation rates increase, so would the cumulative loss of information. Cherry and Taylor (1954) argued that the point of minimal intelligibility in the V-shaped intelligibility function marks the point at which the listener's switching attention is completely out of phase with the shifting signal: By the time the listener's attention has shifted from one ear to the other to follow the signal, the speech would already have returned to the original ear. They accounted for the improved intelligibility when alternation rates are increased beyond this point by suggesting that, at these very rapid rates, listeners begin to concentrate on the input from only one ear. The result would be an interrupted signal to the attended ear in which the frequent but brief interruptions would result in a perceptual "picket fence" effect, with the broad speech envelope still recognizable. T...