Two experiments measured the effects of temporally directed attention on intensity perception in a longer auditory stimulus.
In Experiment 1, listeners evaluated overall loudness of a 1200-ms wideband noise fluctuating in level. The stimulus consisted of 14 temporal segments. For this multiple observation task, the perceptual weight assigned to each segment was derived (Berg, 1989). The weights can be taken as indicators of auditory attention, allowing to measure the attentional pulse over several hundred milliseconds. In the Rhythmic Conditioner condition, the noise was preceded by five tone bursts (IOI = 500 ms). In the Control condition, no conditioners were presented. The hypothesis was that the conditioner rhythm introduces anticipatory attending, so that segments on rhythmically expected positions receive larger weights than in the Control condition. The stimuli were presented in a 1I, 2AFC procedure; listeners decided whether they had just heard a “soft” or “loud” noise. They were instructed to ignore the rhythm and received no trial-by-trial feedback. Perceptual weights were estimated out of 1000 binary responses obtained in each condition. Increased weights on rhythmically expected conditions were found for one of the five listeners only. As in previous studies, the largest weights were assigned to the initial portion of the stimulus. With the rhythmic conditioners, this primacy effect was slightly but significantly stronger than in the control condition.
A potential explanation for the absence of increased weights on expected positions is that in Experiment 1, rhythmic attention would have resulted in impaired performance: an ideal observer would weight the level information provided by each segment uniformly. In Experiment 2, a ‘temporal profile’ task was used. Listeners were required to detect a 4-dB level increment on the sixth of ten 50-ms noise segments. Directing attention to the target segment should improve performance in this task. In the On Beat condition, four rhythmic conditioners preceded the noise; the target segment was presented on a rhythmically expected position. In the Off Beat condition, the target segment occurred 100 ms before the rhythmically expected position. In the Control condition, no rhythm was presented. Listeners received trial-by-trial feedback and completed 1000 trials per condition. As expected, performance was significantly better in the On Beat than in the Control condition. There was no significant difference between the Off Beat and the Control condition. The segment following the target segment was more negatively weighted in the two rhythmic conditioner conditions than in the Control condition. This effect was marginally significant.