Toward a model of dichotic listening performance

Dichotic listening performance for different classes of speech sounds was examined under conditions of controlled attention. Consideration of the complex of target item and competing item demonstrated that, in general, targets were more accurately identified when the competing item shared no relevant features with it and less accurately identified when the competing item shared place, voice, or manner with the target item. Nasals as well as stops demonstrated a significant right-ear advantage (REA). False alarm rates were very similar for left and right attentional conditions, whereas intrusions from the right ear while attending to the left were far more common than intrusions from the left while attending to the right. Attention is viewed as serving to select the stimuli that will be reported, but at a late stage, and only after the right ear perceptual advantage has had its effect. A model of dichotic listening performance is proposed in which both the ease of localizing the item and the strength of evidence for the presence of the item are relevant factors.