A longitudinal investigation of event-related potentials in depression.

A longitudinal investigation of the evoked potential to a warning stimulus, and of the CNV, was conducted with reactive depressives and controls under incentive and nonincentive conditions in a long interstimulus interval paradigm. A smaller group of endogenous depressives were recorded once. The application of multivariate analysis, which controlled for experimental and session effects, indicated that the CNV, as a multiple-component potential, is not capable of distinguishing reactive depressives from controls. Significant univariate interactions involving late CNV area and peak latency were obtained, possibly indexing psychomotor retardation. Analysis of the evoked potential did reveal significant group differences which were explained in terms of main and interaction effects in the P240 and interaction effects involving N120. The findings suggested that groups differed in their responsiveness to experimental demands, and modified their responses differently over time. Mann-Whitney U analysis of endogenous depressives produced no group differences in the CNV measures, but did for N120 amplitude.