Certain questions about memory address a relatively global, structural level of analysis. Is there one kind of memory or many? What brain structures or systems are involved in memory and what jobs do they do? One useful approach to such questions has focused on studies of neurological patients with memory impairmerit and parallel studies with animal models.

Memory impairment sometimes occurs as a circumscribed disorder in the absence of other intellectual deficits 1-7. In such cases, the memory impairment occurs in the context of normal scores on conventional intelligence tests, normal immediate (digit span) memory, and intact memory for very remote events. The analysis of memory impairment can provide useful information about the organization of memory and about the function of the damaged neural structures. Clinically significant memory impairment, ie amnesia, can occur for a variety of reasons and is typically associated with bilateral damage to the medial temporal lobe or the diencephalic midline. The severity and purity of the amnesia can vary greatly depending on the extent and pattern of damage. Standard quantitative tests are available for the assessment of memory and other cognitive functions, so that the findings from different groups of study patients can be compared 8-1. The deficit in amnesia is readily detectable in tests of paired-associate learning and delayed recall. Indeed, amnesic patients are deficient in most tests of new learning, especially when they try to acquire an amount of information that exceeds what can be kept in mind through active rehearsal or when they try to retain information across a delay. This deficit occurs regardless of the sensory modality in which information is presented and regardless whether memory is tested by recall or recognition techniques. Moreover, the memory impairment is not limited to artificial laboratory situations, where patients are instructed explicitly to learn material that occurs in a particular episode and then are later instructed explicitly to recall the material. For example, patients can be provided items of general information with no special instruction to learn (eg Angel Falls is located in Venezuela); and later they can simply be asked factual questions without any reference to a recent learning episode (eg Where is Angel Falls located?). In this case, amnesic patients are impaired both in tests of free recall as well as in tests of recognition memory, in which the correct answer is selected from among several alternatives 11. These aspects of amnesia show that the deficit is a pervasive one, affecting the new learning of both specific episodes and facts.