Paul Thompson's Research Publications

Dynamic Changes in the Human Hippocampus During Memory Consolidation of Face-Name Pairs

International Human Brain Mapping Conference, Sendai, Japan, June 2002

Michael M. Zeineh, Stephen A. Engel, Paul M. Thompson, Susan Y. Bookheimer

UCLA Brain Mapping Division, Los Angeles, CA USA


Structures within the medial temporal lobe (MTL) play a crucial role in forming new episodic memories. The importance of the MTL is clear, but its functional organization is unknown. While some experiments suggest that the hippocampus itself is important for any kind of declarative memory, other studies demonstrate that MTL subregions serve different roles in memory processing. Using human brain imaging at high-resolution, we can resolve signal changes in hippocampal fields CA3, CA1, subiculum and the parahippocampal, perirhinal, and entorhinal cortices. By correlating signal change with memory performance, we begin to define the functional organization of memory within the MTL substructures. In this study, we have taken advantage of recently developed hippocampal unfolding techniques (1) combined with functional magnetic resonance imaging (fMRI) to provide high resolution maps of the evolution of memory for face-name pairs in the medial temporal lobe.

Subjects were scanned on a 3.0-T GE scanner. Hippocampal unfolding methods were identical to (1). The paradigm featured four blocks of learning and retrieval separated by a distraction condition. During the encoding condition, subjects viewed 8 pairs of black-and white faces and first names presented serially every 3.5 seconds and pressed a button for each pair. During the distraction condition, subjects focussed on a fixation cross and pressed a button when the cross changed to a filled circle, which occurred randomly every 2-5 seconds. During the recall condition, subjects viewed the 8 faces serially for 3.5 seconds without the accompanying names. They attempted to recall the name and pressed a button if they judged their recall successful. Subjects performed two to three runs with unique stimuli.

Measurement of subjects’ performance while scanning illustrated a learning curve, with 4.07, 6.60, 7.48, and 7.82 pairs successfully recalled for recall blocks one through four, respectively (12). From this performance curve, an incremental learning curve was calculated (4.07, 2.53, 0.88, 0.34) reflecting how much new learning and retrieving was occurring in successive learn and recall blocks. Using this incremental performance curve, we computed two statistical maps measuring how closely a pixel in our subject population tracks performance either during learning or during retrieval. Strikingly, the anterior CA 2 and 3 fields and dentate gyrus (DG) followed this curve only during encoding (Figure - Learn) but not during retrieval (Figure -Recall). The posterior subiculum, on the other hand, exhibited diminishing activity during retrieval but not during encoding. In contrast, the fusiform gyrus was active throughout both learning and recall, with an initial decrement attributable to priming. These effects were so robust that they were evident from the raw time series computed from all pixels in a subregion without statistical selection.

These findings illustrate that the subregions of the medial temporal lobe contribute to memory in unique ways. Specifically, the anterior CA fields 2 and 3, dentate gyrus were more involved in encoding while the posterior subiculum was more involved in retrieval; in contrast, the fusiform exhibited perceptual priming.

1. Zeineh et al., (2000). Neuroimage 11(6): 668-83. Zeineh et al., (2001) The Anatomical Record: The New Anatomist 265:111-120.

Related Publications

(back to main list)

Contact Information

  • Mail:

    Paul Thompson, Ph.D.
    Assistant Professor of Neurology
    4238 Reed Neurology
    UCLA School of Medicine
    710 Westwood Plaza
    Westwood, Los Angeles CA 90095-1769, USA.

  • E-mail:
  • Tel: (310)206-2101
  • Fax: (310)206-5518