Information about which structures and connections in the brain are important for MEMORY has come from studies of amnesiac patients and from systematic experimental work with animals. Work in animals includes studies which assess the effects of selective brain lesions on memory, as well as studies using neurophysiological recording and stimulating techniques to investigate neural activity within particular brain regions (for discussions of the latter two approaches, see OBJECT RECOGNITION, ANIMAL STUDIES; FACE RECOGNITION; SINGLE-NEURON RECORDING). An important development that has occurred in the area of memory during the past two decades was the establishment of an animal model of human amnesia in the monkey (Mahut and Moss 1984; Mishkin 1982; Squire and ZolaMorgan 1983). In the 1950s, Scoville and Milner (1957) described the severe amnesia that followed bilateral surgical removal of the medial temporal lobe (patient H.M.). This important case demonstrated that memory is a distinct cerebral function, dissociable from other perceptual and cognitive abilities.
In monkeys, surgical lesions of the medial temporal lobe, which were intended to approximate the damage sustained by patient H.M., reproduced many features of human memory impairment. In particular, both monkeys and humans were impaired on tasks of declarative memory, but fully intact at skills and habit learning and other tasks of non-declarative memory. This achievement set the stage for additional work in monkeys and for work in rodents that has identified structures in the medial temporal lobe that are important for declarative memory. These structures include the hippocampal region (i.e., the cell fields of the hippocampus, the dentate gyrus, and the subiculum), and adjacent cortical areas that are anatomically related to the hippocampal region, namely, the entorhinal, perirhinal, and parahippocampal cortices (Zola-Morgan and Squire 1993).
The midline diencephalon is another brain area important for memory, although less is known about which specific structures in this region contribute to memory function. Findings from work in animals, including the development of an animal model of alcoholic Korsakoff's syndrome in the rat (Mair et al., 1992), have been consistent with the anatomical findings from human amnesia in showing the importance of damage within the medial thalamus, especially damage in the internal medullary lamina, for producing memory loss. Lesions in the internal medullary lamina would be expected to disconnect or damage several thalamic nuclei, including intralaminar nuclei, the mediodorsal nucleus, and the anterior nucleus (Aggleton and Mishkin 1983; Mair et al. 1991; Zola-Morgan and Squire 1985). However, the separate contributions to memory of the mediodorsal nucleus, the anterior nucleus, and the intralaminar nuclei remain to be explored systematically with well-circumscribed lesions in animals.
A major criterion for demonstrating that an animal has a memory deficit is to show that performance is impaired at long-delay intervals, but is intact at short-delay intervals, that is, no impairment in perception, attention, or general intellectual function. A successful strategy for demonstrating intact short-term memory and impaired long-term memory has involved training normal monkeys and monkeys with medial temporal lobe lesions on the delayed nonmatching -to-sample task, a recognition memory task sensitive to amnesia in humans. In this task, the monkey first sees an object, and then after a prescribed delay the animal is given a choice between the previously seen object and a novel one. The key feature of this experimental approach is the use of very short delay intervals (e.g., 0.5 sec). The absence of an impairment at a delay of 0.5 sec would indicate that the medial temporal lobe lesions do not affect short-term memory. Using this strategy, Alvarez-Royo, Zola-Morgan, and Squire (1992) and Overman, Ormsby, and Mishkin (1990) showed that medial temporal lobe lesions impair memory at long delays, but not at very short delays. Studies in rats using delayed nonmatching-to-sample as well as a variety of other memory tasks have also demonstrated that long-term memory is impaired while short-term memory is spared following lesions that involve the hippocampal region (Kesner and Novak 1982; for recent reviews of work in rats, see Further Readings). These findings underscore the idea that medial temporal lobe lesions reproduce a key feature of human amnesia, that is, the distinction between intact short-term memory and impaired long-term memory.
It was originally supposed that damage to the amygdala directly contributed to the memory impairment associated with large medial temporal lobe lesions (Murray and Mishkin 1984). Subsequent work showed that monkeys with virtually complete lesions of the amygdala performed as well as normal monkeys on four different memory tasks, including delayed nonmatching-to-sample task (ZolaMorgan et al. 1989). other experiments with rats and mon keys suggest that the amygdala is important for other kinds of memory, including the development of conditioned fear and other forms of affective memory (see emotion and the animal brain). These and other findings (Murray 1992) focused attention away from the amygdala toward the cortical structures of the medial temporal lobe, that is, the perirhinal, entorhinal, and parahippocampal cortices, in addition to the hippocampal region itself.
Direct evidence for the importance of the cortical regions has come from studies in which circumscribed damage has been done to the perirhinal, entorhinal, or parahippocampal cortices, either separately or in combination (Moss, Mahut, and Zola-Morgan 1981; Zola-Morgan et al. 1989; Gaffan and Murray 1992; Meunier et al. 1993; Suzuki et al. 1993; Leonard et al. 1995). For example, monkeys with combined lesions of the perirhinal and parahippocampal cortices exhibited severe, multimodal, and long-lasting memory impairment (Zola-Morgan et al. 1989; Suzuki et al. 1993). More limited lesions of the cortical regions also produce memory impairment. For example, several studies found that monkeys with bilateral lesions limited to the perirhinal cortex exhibit long-lasting memory impairment (Meunier et al. 1993; Ramus, Zola-Morgan, and Squire 1994). Additionally, a large number of individual studies in monkeys and in rats with varying extents of damage to the medial temporal lobe, together with work in humans, has led to the idea that the severity of memory impairment increases as more components of the medial temporal lobe memory system are damaged.
A long-standing and controversial issue in work on memory has been whether the hippocampal region is disproportionately involved in spatial memory, or whether spatial memory is simply a good example of a broader category of memory that requires the hippocampal region. One view of the matter comes from earlier work with monkeys (Parkinson, Murray, and Mishkin 1988). Monkeys with lesions that involved the hippocampal formation (hippocampus plus underlying posterior entorhinal cortex and parahippoc-ampal cortex) were severely impaired in acquiring an object-place association task, whereas lesions that involved the amygdala plus underlying anterior entorhinal cortex and perirhinal cortex were only mildly impaired. The authors suggested that the hippocampus has an especially important role in spatial memory, an idea developed originally by O'Keefe and Nadel (1978), based mostly on rat work. It was unclear from this monkey study, however, whether the observed spatial deficit was due to hippocampal damage, the adjacent cortical damage, or both. Additional work from both humans and animals suggests another view. In one formal study (Cave and Squire 1991), spatial memory was found to be proportionately impaired in amnesiac patients relative to object recognition memory and object recall memory. The same (nonspatial) view of hippocampal function has also been proposed for the rat, based, for example, on demonstrated deficits in odor memory tasks after ibote-nate hippocampal lesions (Bunsey and Eichenbaum 1996). The role of the hippocampus in spatial memory remains unclear. Recent commentaries on the issue of the hippocampus and spatial memory can be found under Further Reading.
Uncertainty about the function of the hippocampus has been due, in part, to the inability until recently to make circumscribed lesions limited to the hippocampal region in experimental animals. Studies in which selective lesions of the hippocampal region could be accomplished became possible only with the development of (a) a technique for producing restricted ibotenate lesions of the hippocampus in the rat and (b) a technique that uses magnetic resonance imaging to guide the placement of radiofrequency or ibotenic acid stereotaxic lesions of the hippocampal region in the monkey. Monkeys with bilateral, radiofrequency lesions of the hippocampal region, which spared almost entirely the perirhinal, entorhinal, and parahippocampal cortices, exhibited impaired performance at long delays (ten minutes and forty minutes) on the delayed nonmatching-to-sample task (Alvarez, Zola-Morgan, and Squire 1995).
Ibotenic acid lesions cause cell death but, unlike radio-frequency lesions, spare afferent and efferent white matter fibers within the region of the lesion. If it should turn out, after systematic study, that ibotenic acid lesions of the hip-pocampal region do not impair performance on the delayed nonmatching task, the interpretation of such studies should not be overstated. The results concern recognition memory, not memory in general, and only the kind of recognition memory measured by the nonmatching-to-sample task itself. The delayed nonmatching task has been extraordinarily useful for evaluating the effects on visual recognition memory of damage to the medial temporal lobe memory system and for measuring the severity of recognition memory impairment. However, in the case of human memory, recognition memory tests are known to be rather easy and not as sensitive to memory impairment as other tests, for instance, tests of recall or cued recall. The issue of task sensitivity is crucially important. Other kinds of recognition memory tasks, for example, the paired comparisons task (a task of spontaneous novelty preference; Bachevalier, Brick-son, and Hagger 1993) and tasks that are thought to be more sensitive than tasks of simple recognition memory, for example the transverse patterning, the transitive inference, and naturalistic association tasks, have recently been developed to assess memory in animals.
An important question with respect to the components of the medial temporal lobe memory system is whether these structures all share similar functions as part of a common memory system, or do they have distinct and dissociable functions? In this regard, one must consider the neuroanatomy of the medial temporal lobe system and its pattern of connectivity with association cortex. An extensive anatomical investigation by Suzuki and Amaral (1994) showed that different areas of neocortex gain access to the medial temporal lobe memory system at different points. Visual information arrives preferentially to perirhinal cortex. Approximately 65 percent of the input reaching the perirhinal cortex is unimodal visual information, mostly from TE and TEO. By contrast, about 40 percent of the input reaching parahippocampal cortex is visual, mostly from area V4. Cortical areas that are believed to be important for processing spatial information project preferentially to parahippo-campal cortex. Approximately 8 percent of the input to para-hippocampal cortex originates in the parietal cortex, whereas virtually none of the input to perirhinal cortex originates in the parietal cortex. These anatomical considerations lead to the expectation that perirhinal cortical lesions might impair visual memory more than spatial memory and that the reverse might be true for parahippocampal cortex. Furthermore, because both the perirhinal and the parahippocampal cortices project to the hippocampus, one might expect that hippocampal damage will similarly impair visual memory and spatial memory. The establishment of new, more sensitive behavioral tests and the development of new techniques for producing selective brain lesions have now made it possible to address these possibilities and to systematically clarify the separate contributions to memory of structures in the medial temporal lobe and the diencephalon.
WORKING MEMORY, NEURAL BASIS OF —Stuart Zola
Was this article helpful?