Event-related potentials reveal processing differences in honest vs. malingered memory performance

Twenty-two undergraduate students completed a recognition memory test while event-related potentials (ERPs) were recorded. During the testing phase, subjects distinguished old from new words in a forced-choice format. There were two counterbalanced within-subject conditions, one in which subjects performed to the best of their abilities, and another with instructions to feign memory impairment. Test scores and response latencies differed significantly between the two conditions. Analysis of PCA-defined epochs revealed that old words were more positive than new in the control condition, with this difference confined to frontal regions and interpreted as reflecting familiarity-based recognition judgements. In the malingering task, this old/new word difference emerged earlier and was broadly distributed across the scalp. A discriminant function analysis using reaction time and ERP measures resulted in 82% correct classification of honest and simulated performance, with 79% correct on cross-validation. (C) 2002 Elsevier Science B.V. All rights reserved.