The design, analysis and conceptual framework of experiments on attention
Due to the adoption of information theory by psychology, and especially to the popularity of the model developed for the human operator by Broadbent (1958), work in the last decade has frequently made use of the idea of the 'channel' over which messages arrive, and has described attention as the selection of one 'channel' and the rejection of another 'channel'. Because many of the experiments were auditory and since there has been particular interest in dichotic stimulation (where the listener receives one message through one ear and another message through the other ear) this seemed conceptually reasonable. However, later work has led to increasing confusion in the use of this word as pyschologists have used a large number of 'channels', some of which are very difficult to define in terms either of physical identity or even functional capacity. It is reasonable at first sight to talk of the left and right ears as providing two input channels. But supposing that more than two positions in space are used as sources (Triesman, 1964b; Moray, Bates and Barnett, 1965), is each one a channel? Is a language, for example, French, a different channel from English, if the two messages are merely translations of one another and come from the same location and are spoken by the same voice? (Treisman, 1964a) Is voice quality a channel in this sense? (Moray and Barnett, 1965)
Although the term 'channel' is still in use to describe the presentation of different messages, and while it is a convenient indeed almost irresistible shorthand for those familiar with the field, it seems that a more sophisticated treatment is now needed. For example, describing an experiment in which the ability to select on the basis of type of stimulus material was measured, Broadbent and Gregory (1964a) write
... a different kind of attention, to a class of item rather than to a source of stimulation.