"I heard it before … or not": time-course of ERP response and behavioural correlates associated with false recognition memory.

Electrophysiological and behavioural correlates of true and false memories were examined in the Deese/Roediger-McDermont (DRM) paradigm. A mass univariate approach for analysing event-related potentials (ERP) in the temporal domain was used to examine the electrophysiological effects associated with this paradigm precisely (point-by-point) and without bias (data-driven). Behaviourally, true and false recognition did not differ, and the predicted DRM effect was observed, as false recognition of critical lures (i.e., new words semantically related to studied words) was higher than false alarms of new (unrelated) words. Neurally, an expected old/new effect was observed during the time-range of the late positive component (LPC) over left centro-parietal scalp electrodes. Furthermore, true recognition also evoked larger LPC amplitudes than false recognition over both left centro-parietal and fronto-central scalp electrodes. However, we did not observe LPC-related differences between critical lures and new words, nor between correct rejections of critical lures and new words. In contrast, correct rejections of critical lures were accompanied by higher activation of a sustained positive slow wave (SPSW) in right fronto-central electrodes beyond 1200 ms. This result reveals a key role of post-retrieval processes in recognition. Results are discussed in light of theoretical approaches to false memory in the DRM paradigm.