On the number of trials needed for P300

Abstract

The P300 event-related brain potential (ERP) was elicited in n=24 subjects using auditory and visual stimuli with target probabilities of 0.20 and 0.80 in different conditions. A total of 40 target stimulus ERPs were obtained from each subject and were averaged in two different ways: (1) single-trial averages were obtained by averaging each subsequent pairs of trial blocks (1–2, 3–4, 5–6, etc.); (2) successive averages were obtained by averaging successive blocks of two trials (1–2, 1–4, 1–6, etc.). Statistical analyses demonstrated that for the single-trial averages, P300 measures are variable but not significantly different from one another across trial pairs, whereas successive averages demonstrated systematic decreases in amplitude and slight increases in peak latency. Auditory and visual stimuli produced similar patterns for the two averaging methods, as did target probability. The findings suggest that P300 amplitude stabilizes with approximately 20 target trials for all conditions, and that peak latency changes relatively little during this time. The results are discussed in terms of applied/clinical contexts. © 1997 Elsevier Science B.V.