Comparative electrophysiological and hemodynamic measures of neural activation during memory-retrieval

The spatial and temporal characteristics of the brain processes underlying memory retrieval were studied with both event-related potentials (ERP) and positron emission tomography (PET) techniques. Subjects studied lists of 20 words and then performed episodic (old/new judgment) or semantic (living/n onliving decision) retrieval tasks on multiple four-item test lists, each l asting 10 sec. The PET and ERP measurements at test were assessed in relati on to both the task (episodic vs. semantic) and the item (old vs, new or li ving vs. nonliving). Episodic retrieval was associated with increased blood flow in the right frontal lobe (Brodmann Area 10) and a sustained, slowly developing positive ERP shift recorded from the right frontopolar scalp. Se mantic retrieval was associated with increased blood flow in the left front al (Area 45) and temporal (Area 21) lobes but no clear ERP concomitant. The two retrieval tasks also differed from each other in the ERPs to single it ems in an early (300-500 ms) time window. Item-related comparisons yielded convergent results mainly if the retrieved information was relevant to the given task (e.g., old/new items during episodic retrieval and living/nonliv ing items during semantic retrieval). Episodically retrieved old items were associated with increased blood flow in the left medial temporal lobe and a transient increase in the amplitude of the late positive component (500-7 00 ms) of the ERP. Semantically retrieved living items were associated with increased blood flow in the left frontal cortex and anterior cingulate and a transient late frontal slow wave (700 -1,500 ms) in the ERPs. These resu lts indicate that the brain regions engaged in memory retrieval are active in either a sustained or transient manner. They map task-related processes to sustained and item-related processes to transient neural activity. But t hey also suggest that task-related factors can transiently affect early sta ges of item processing. (C) 2001 Wiley-Liss, Inc.