Division of attention: Evidence for a between-trial component of impaired performance

Detection and recognition studies are reviewed, which show a dependency between errors on successive trials. They suggest the impairment accompanying the presentation of simultaneous signals may also involve a similar “between-trial” component. The hypothesis was tested using a two-choice recognition paradigm with bimodal signals. Dimensions presented simultaneously were interleaved on alternate trials with a single dimension presented alone. Interleaving two signals in the first experiment showed no effect on the following signal presented alone, compared to a control condition in which the signal always preceded itself. Interleaving four signals in the second experiment produced a significant impairment. Accounts of the results based on the length of the response interval, fluctuations of state variables such as arousal and decay of the memory trace are rejected. Accounts based on the impairment and set-size of preceding simultaneous signals are retained. Both suppose part of the impairment accompanying simultaneous signals to be a between-trial phenomenon. Implications of the results for work on divided-attention are considered and a method of experimental control proposed.