ABSTRACT
The hypothesis that attentive selection involves a "filter" is considered and rejected in favor of an interpretation based on analysis-by-synthesis. The hypothesis that auditory attention is the process of synthesis also explains how attention can be withdrawn from every external channel and focused on the subject's own train of thought. This common experience seems to pose grave difficulties for Broadbent's filter theory, and insuperable ones for the filter-amplitude theory of Treisman. So far as direct control over the listener's experience and behavior is concerned, the echo certainly disappears rather quickly. It is possible to suppose that echoic memory is discarded immediately after each segment of speech. All these feats apparently require a fine-grained or echoic memory of appreciable length that persists through a number of cognitive units. The most direct attempt to measure the duration of echoic memory is perhaps that of Guttman and Julesz.
