Long-term patterns of change in ERPs across repeated measurements.

Ten subjects were tested for eight sessions using an auditory oddball paradigm at 1-week and 1-month intervals to examine the stability of multiple event-related potential (ERP) components, in terms of trait (between session) and state (within session) reliability. The between session reliabilities presented as intraclass correlations (r') were 0.54 for the P300 amplitude and 0.47 for the P300 latency. Pearson correlation coefficients, representing within-session reliabilities, ranged from 0.54 to 0.68 for P300 amplitude and 0.12 to 0.46 for P300 latency. Statistical analysis of results from the eight sessions demonstrated that session-dependent reduction and recovery of the P300 amplitude occurred during the testing sessions. Other ERPs measurements such as N100-P300, N200-P300, N100, and N200 amplitudes did not exhibit any significant changes. The between- and within-session reliabilities of P300 components suggest that P300 may be a reproducible biological marker of state and trait even across repeated measurements. We speculate that the observed long-term patterns of change in P300 may have reflected habituation and dishabituation of certain processes in the human central nervous system.