Memory and the Brain: What's Right and What's Left?

Abstract

A recent study by Kelley et al. 1998xKelley, W.M, Miezin, F.M, McDermott, K.B, Buckner, R.L, Raichle, M.E, Cohen, N.J, Ollinger, J.M, Adbudak, E, Conturo, T.E, Snyder, A.Z, and Petersen, S.E. Neuron. 1998; 20: 927–936Abstract | Full Text | Full Text PDF | PubMed | Scopus (596)See all ReferencesKelley et al. 1998 provides an important rapprochement of the neuropsychological and imaging approaches to hemispheric specialization in memory encoding. This study used fMRI to examine brain activation during encoding into memory of three types of stimuli: words (which obligatorily engage verbal processing), line drawings of common objects (which engage both verbal processing and form-based visual processing), and unfamiliar faces (which are difficult to label verbally and rely upon processing of visual form). Images obtained during the encoding of each of these stimuli were compared to a baseline task in which subjects fixated visually on a crosshair. Brain regions demonstrating enhanced fMRI signal (denoting increases in blood oxygenation and, by inference, increases in synaptic activity) during the memory-encoding periods compared to the fixation period included the prefrontal cortex and MTL. Notably, the lateralization of activation during encoding was determined by the nature of the stimulus materials. Consonant with the results of lesion studies, the left posterior prefrontal cortex (Brodmann's area [BA] 44/6) was most active during encoding of words, the right posterior prefrontal cortex (BA 44/6) was most active during encoding of faces, and both hemispheres were active during encoding of nameable objects. A similar pattern of lateralized activation was found in the MTL.The Kelley et al. 1998xKelley, W.M, Miezin, F.M, McDermott, K.B, Buckner, R.L, Raichle, M.E, Cohen, N.J, Ollinger, J.M, Adbudak, E, Conturo, T.E, Snyder, A.Z, and Petersen, S.E. Neuron. 1998; 20: 927–936Abstract | Full Text | Full Text PDF | PubMed | Scopus (596)See all ReferencesKelley et al. 1998 study provides important information about the lateralization of memory encoding, but did not examine memory retrieval. A study by Klingberg and Roland 1998xKlingberg, T and Roland, P.E. Cereb. Cortex. 1998; 8: 73–79Crossref | PubMed | Scopus (63)See all ReferencesKlingberg and Roland 1998 extended this result by examining both encoding and retrieval for pairings of nonverbal patterns and sounds. Subjects learned pattern–sound pairings by trial and error, and brain activity was measured using PET during a period early in training (measuring memory encoding) and a period later in training (measuring memory retrieval). Klingberg and Roland found encoding-specific activation in the right posterior prefrontal cortex (BA 44/6), which was significantly greater than activation in the same region at retrieval. This finding suggests that some regions may exhibit specificity both to the type of materials (verbal or nonverbal) and to the memory processing (encoding or retrieval).One important question is why these results differ from those of the previous studies, which had led to formulation of the HERA model; particularly of interest is why previous studies of face encoding found left hemisphere prefrontal activation. Differences in tasks may have resulted in selective use of verbal or nonverbal processes during face encoding in these studies, which could have resulted in either left or right hemisphere activation, respectively. Evidence for this account comes from a study by Haxby and colleagues (Haxby et al. 1995xHaxby, J.V, Ungerleider, L.G, Horwitz, B, Rapoport, S.I, and Grady, C.L. Hum. Brain Mapping. 1995; 3: 68–82CrossrefSee all ReferencesHaxby et al. 1995), which examined the retention of unfamiliar faces across time. Retention of faces for 1 s resulted in both right and left hemisphere frontal activation, whereas retention of faces for 6 s resulted only in left hemisphere activation. This suggests that the increased amount of encoding time allowed in previous studies may have shifted the subjects' strategy toward verbal labeling of facial features. The findings of Klingberg and Roland 1998xKlingberg, T and Roland, P.E. Cereb. Cortex. 1998; 8: 73–79Crossref | PubMed | Scopus (63)See all ReferencesKlingberg and Roland 1998 using nonsense patterns suggest that the right hemisphere is the locus of memory encoding when such verbal labeling is discouraged.The recent results reviewed here represent a significant step in the understanding of the neural architecture of memory encoding because they integrate findings from lesion and neuroimaging research. Major questions, however, remain to be answered. The HERA model, and the foregoing review, treat memory encoding as if it were a single process mediated by a single frontal lobe region. The actual encoding of a memory, however, undoubtedly involves multiple cognitive processes, including assembling a representation of the stimulus, detecting that the stimulus is novel, relating the stimulus to existing knowledge, and laying down a lasting memory trace of the assembled representation. Indeed, there is already evidence that different regions of the left frontal lobe are specialized for different kinds of verbal encoding processes (1xPsychonomic Bull. Buckner, R.L. Rev. 1996; 3: 149–158See all References, 5xGabrieli, J.D.E, Poldrack, R.A, and Desmond, J.E. Proc. Natl. Acad. Sci. USA. 1998; 95: 906–913Crossref | PubMed | Scopus (573)See all References). Further, recent findings by Dolan and Fletcher 1997xDolan, R.J and Fletcher, P.C. Nature. 1997; 388: 582–585Crossref | PubMed | Scopus (302)See all ReferencesDolan and Fletcher 1997 have suggested separate roles for the left prefrontal cortex and MTL in verbal memory encoding, with the MTL processing stimulus novelty and the prefrontal cortex integrating new stimuli with existing knowledge. Additional studies will continue to characterize the specific contribution of multiple frontal and MTL brain areas to different aspects of memory encoding. Finally, there is still a need to better reconcile lesion and neuroimaging evidence concerning right prefrontal activations for memory retrieval. One approach has been to interpret those activations as short-term memory processes involved in the evaluation of memory judgments, rather than representing long-term memories per se (Wagner et al. 1998xWagner, A.D, Desmond, J.E, Glover, G.H, and Gabrieli, J.D.E. See all ReferencesWagner et al. 1998).The most exciting aspect of these recent results is that they begin to demonstrate neural bases for fundamental results from cognitive psychology. One such result is the “picture superiority effect,” which refers to the fact that pictures of real objects are remembered far better than words. The results of the Kelley et al. 1998xKelley, W.M, Miezin, F.M, McDermott, K.B, Buckner, R.L, Raichle, M.E, Cohen, N.J, Ollinger, J.M, Adbudak, E, Conturo, T.E, Snyder, A.Z, and Petersen, S.E. Neuron. 1998; 20: 927–936Abstract | Full Text | Full Text PDF | PubMed | Scopus (596)See all ReferencesKelley et al. 1998 study offer a simple explanation for this result in neural terms: the involvement of additional right hemisphere brain regions during object encoding that are not active during word encoding results in greater memory for objects. Further research that combines functional brain imaging with lesion studies should continue to provide new insights into the organization of cognitive processes in the brain.