P3 latency in threshold signal detection.

The EOG, EEG(Cz, Pz) and response latencies were obtained from ten subjects in an auditory threshold detection task with a signal probability of 0.5. In order to minimize any interaction of signal condition (signal-presence vs. signal-absence) and uncertainty about the moment of signal occurrence (temporal equivocation), the observation period was marked with a cue light. The single-trial latencies of the P3 component associated with correct signal-present and signal-absent decisions were assessed on the basis of the cross-covariance functions of the EEG epochs and a template of the P3. Trials with values of covariance smaller than a criterion value were discarded. The mean and dispersion of the P3 latency proved larger for signal-absent than for signal-present trials, while the mean values of covariance were smaller for signal-absent than for signal-present trials. In addition, the mean and dispersion of the P3 proved smaller for high confidence than for medium confidence signal-present trials. Nearly all these effects also proved significant for the response latency. The single-trial P3 latencies were positively correlated with the response latencies of the same trials. The observed latency variations of the P3 were interpreted in terms of changes in the detectability of the stimulus events. It was concluded that the signal condition had a significant effect upon the latency of the P3 associated with signal-present and signal-absent decisions over and above that due to temporal equivocation.