Division of attention and the decision theory of signal detection

Abstract

Ten men were asked to listen to bursts of noise which were presented to one ear, and which might or might not contain a tone. The other ear received 6 digit numbers simultaneously. The listeners reported their degree of confidence that a tone was present; in one condition they ignored the numbers and in another condition they had to report them as well as their judgement about the tone. In the latter condition they reported the tone with confidence slightly less often when it was present, but also reported it more often when it was in fact absent. Analysis of the results, by a model which supposes the brain to detect signals by a statistical decision, shows that one parameter, $\beta $, is unchanged by division of attention. This parameter measures the subjective probabilities and values associated with signal as opposed to nonsignal. Another parameter, d$^{\prime}$, changes when attention is divided. This quantity measures the strength of the signal relative to the random variation within the system. It is concluded that diversion of attention away from a stimulus produces an effect resembling a reduction in the intensity of the stimulus. The ignored event is therefore not blocked altogether and under suitable conditions may nevertheless produce a response from an observer.