|
COPYRIGHT 2001 Massachusetts Institute of Technology
Jon S. Simons [1]
Kim S. Graham [1]
Adrian M. Owen [1, 2]
Karalyn Patterson [1]
John R. Hodges [1, 3]
Abstract
* Previous studies have suggested differences in the neural substrates of recognition memory when the contributions of perceptual and semantic information are manipulated. In a within-subjects design PET study, we investigated the neural correlates of the following factors: material type (objects or faces), semantic knowledge (familiar or unfamiliar items), and perceptual similarity at study and test (identical or different pictures). There was consistent material-specific lateralization in frontal and temporal lobe regions when the retrieval of different types of nonverbal stimuli was compared, with objects activating bilateral areas and faces preferentially activating the right hemisphere. Retrieval of memories for nameable, familiar items was associated with increased activation in the left ventrolateral prefrontal cortex, while memory for unfamiliar items involved occipital regions. Recognition memory for different pictures of the same item at study and test produced blood flow increase in left inferior t emporal cortex. These results have implications for our understanding of the neural correlates of perceptual and semantic contributions to recognition memory.
INTRODUCTION
Neuropsychological and functional imaging investigations have greatly advanced our understanding of human long-term memory in recent years. Interest among cognitive neuroscientists has traditionally focused on the important role of medial temporal (Tulving & Markowitsch, 1998; Vargha-Khadem et al., 1997; Cohen & Eichenbaum, 1993; Squire, 1987, 1992) and frontal (Moscovitch & Winocur, 1995; Shimamura, 1995; Wheeler, Stuss, & Tulving, 1995; Milner, Petrides, & Smith, 1985) lobe structures in memory processes. Although several early neuroimaging studies failed to find memory-related activation in the medial temporal lobe, more recent studies have documented activity differences in this region during the encoding and retrieval of new information (Schacter & Wagner, 1999; Schacter et al., 1997; Kelley et al., 1998; Dolan & Fletcher, 1997; Gabrieli, Brewer, Desmond, & Glover, 1997; Owen, Milner, Petrides, & Evans, 1996; Tulving, Markowitsch, Craik, Habib, & Houle, 1996).
The importance of areas in the prefrontal cortex for long-term memory has also been highlighted by functional imaging studies (McDermott, Buckner, Petersen, Kelley, & Sanders, 1999; Buckner & Koutstaal, 1998; Buckner, Kelley, & Petersen, 1999; Fletcher, Shallice, & Dolan, 1998; Fletcher, Shallice, Frith, Frackowiak, & Dolan, 1998; Kelley et al., 1998; Dolan & Fletcher, 1997). Evidence has accumulated to support the idea that the left and right frontal cortex are lateralized for encoding and retrieval of memories (Fletcher, Shallice, & Dolan, 1998; Fletcher, Shallice, Frith, et al., 1998; Dolan & Fletcher, 1997; Nyberg, Cabeza, & Tulving, 1996; Shallice et al., 1994; Tulving, Kapur, Craik, Moscovitch, & Houle, 1994). The results of some more recent studies, however, suggest that both left and right prefrontal areas are involved in both episodic encoding and retrieval, with the relative lateralization of activation depending instead upon the type of material being remembered (Lee, Robbins, Pickard, & Owen, 200 0; Kim et al., 1999; McDermott et al., 1999; Kelley et al., 1998; Klingberg & Roland, 1998; Wagner, Desmond, Glover, & Gabrieli, 1998; Wagner, Poldrack, et al., 1998).
Material-Specific Lateralization
Much recent interest in the neuroimaging literature has centered on the idea that memory for different types of material may be lateralized to different regions within the left and right hemispheres. Although such lateralization is an established concept in the neuropsychological literature (Baxendale, 1997; Warrington, 1984; Milner, 1970, 1972; Warrington & James, 1967; Gazzaniga, Bogen, & Sperry, 1962), it is only recently that functional imaging studies of memory have begun to explore regional activations associated with the verbal/nonverbal nature of the stimuli employed (Kim et al., 1999; McDermott et al., 1999; Kelley et al., 1998; Wagner, Poldrack, et al., 1998). Of the few studies that have directly compared the encoding of verbal and nonverbal material, the majority has demonstrated that verbal encoding involves prefrontal and medial temporal regions in the left hemisphere and nonverbal encoding analogous regions in the right hemisphere (McDermott et al., 1999; Kelley et al., 1998; Wagner, Poldrack, et al., 1998). This same verbal/nonverbal lateralization has also been identified in memory retrieval studies (Lee et al., 2000; Kim et al., 1999; McDermott et al., 1999; Wagner, Poldrack, et al., 1998).
The term "nonverbal" covers many different types of stimuli, such as faces, textures, and objects. The characteristic pattern of activation in right hemisphere regions has been demonstrated when faces (Kim et al., 1999; McDermott et al., 1999; Kelley et al., 1998) and textures (Wagner, Poldrack, et al., 1998) are contrasted with words, but only one study has directly compared memory for objects and words within the same participants. Kelley et al. (1998) investigated the encoding of object, face, and word memories and found that, while faces and words activated the expected right and left prefrontal and medial temporal regions respectively, object encoding activated these regions bilaterally. The authors interpreted this result in terms of dual-coding theory (Paivio & Csapo, 1973), suggesting that the encoding of nameable objects can draw upon both verbal and nonverbal information, hence the involvement of both hemispheres.
There are, to the best of our knowledge, no published studies reporting activations associated with the retrieval of different types of "nonverbal" material. The first aim of the present study, therefore, was to compare the retrieval of faces and objects within the same participants. Based on the findings described above, we hypothesized that the retrieval phase of recognition memory for faces compared with objects would activate predominantly right hemisphere regions (Kim et al., 1999; McDermott et al., 1999) within the prefrontal cortex and temporal lobe. Retrieval of memory for objects relative to faces, however, was expected to produce a bilateral pattern of activation in these areas, similar to that reported for related studies of encoding (Kelley et al., 1998).
Manipulating Semantic Knowledge
It has been argued that the bilateral pattern of activation produced when nameable objects are remembered reflects the involvement of both verbal (i.e., left hemisphere) and nonverbal (i.e., right hemisphere) processes. If true, then one might reasonably expect unfamiliar (i.e., novel) and, therefore, unnameable objects to produce a pattern of activation that is more strongly lateralized to the right hemisphere. On the same grounds, while memory for unfamiliar faces produces activations that are predominantly right-sided (Kim et al., 1999; McDermott et al., 1999; Kelley et al., 1998), familiar faces, which may be explicitly or implicitly named in the course of processing, may yield a less strongly lateralized pattern of activation.
In the only memory study of which we are aware that directly compared encoding of familiar and unfamiliar faces, encoding of famous faces did indeed produce bilateral activation (Kelley et al., 1999). A similar pattern was also reported recently in a comparison between a fame judgement task using famous faces and recognition memory for recently encoded unfamiliar faces (Leveroni et al., 2000). The second aim of the present study was to provide a direct comparison of recognition memory retrieval of famous and unfamiliar faces, and of nameable and novel objects, within the same participants. We hypothesized that compared to unfamiliar items, the additional verbal processing possible with familiar items might produce activation involving left prefrontal cortex areas implicated in the strategic processing of semantic information, and perhaps in the left temporal lobe, thought to be associated with semantic representations (Mummery et al., 1999; Wagner, Poldrack, et al., 1998; Dolan & Fletcher, 1997; Thompson-Sch ill, D'Esposito, Aguirre, & Farah, 1997; Vandenberghe, Price, Wise, Josephs, & Frackowiak, 1996). Conversely, compared to familiar items, the predominantly nonverbal, perceptual processing required for recognition memory of unfamiliar items should be associated with more selective right hemisphere activation, perhaps including more posterior, visuoperceptual areas in the occipital lobes.
Perceptual Similarity at Study and Test
When familiar items are used as stimuli, both perceptual and semantic factors may contribute to successful recognition memory (Graham, Simons, Pratt, Patterson, & Hodges, 2000; Bruce, 1982; Paivio & Csapo, 1973). Indeed, it has been argued that the slight decrement in performance observed when different pictures of objects or faces are used at study and test (Srinivas, 1995; Cooper, Schacter, Ballesteros, & Moore, 1992; Bruce, 1982) may reflect the fact that the value of the perceptual information available from seeing...
Read the full article for free courtesy of your local library.
|
Bookmark this article
Print this article
Link to this article
Email this article
Digg It!
Add to del.icio.us
RSS
