On how P300 amplitude varies with the utility of the eliciting stimuli.

The bulk of the relevant evidence indicates that stimuli which elicit the P300 component of the human event-related potential (ERP) tend to be unexpected and task relevant. Yet, contradictory evidence has been presented which implied that P300 amplitude is modulated by the intensity difference between two equally relevant, equally probable feedback stimuli (Adams and Benson, 1973). The two experiments reported here were designed to clarify this contradiction. Subjects were instructed to press a button one second after a cue light. A tone at one of two intensities was presented 200 msec later indicating whether the time estimate was 'correct' or 'incorrect'. The magnitude of P300 varied as a function of the difference in intensity between S+ and S-; the smaller the difference, the smaller was P300, regardless of stimulus meaning (S+ or S-). We attribute this effect to the degree to which the subject chooses to utilize the tones in the performance of the time estimation task. In support of this hypothesis, we found that performance of the subjects as time estimators degraded as the difference between S+ and S- decreased. Moreover, a study of choice reaction times taken for the different pairings of S+ and S- showed that the stimulus pairs were not equally discriminable. It appears that as the S+/S- pairs become increasingly difficult to discriminate, their feedback value decreased and this decrease is positively correlated with a decrease in P300 amplitude. This interpretation received support from Exp. 2 in which the stimulus conditions were the same as in Exp. 1, but the subject's task was to count the "s+ stimuli. Here, when subject had no option but to process each individual stimulus, the P300 amplitude remained constant despite variations in S- intensity. Thus, it appears that task relevance must be defined in terms of the subject's behavior rather than by the experimenter's instructions to the subject.