Search

Short-term memory was assessed in two groups of amnesic patients. Six patients had confirmed or suspected damage to the hippocampal formation, and six patients had diencephalic damage as a result of alcoholic Korsakoff's syndrome. Verbal short-term memory was evaluated with seven separate administrations of the standard digit span test in order to obtain a precise measure of short-term memory. Nonverbal short-term memory was evaluated with four tests that assessed apprehension, retention, and the ability to manipulate nonverbal material--all within the span of immediate memory. One of these four tests assessed short-term memory for spatial location. Patients with damage to the hippocampal formation had a digit span equivalent to that of control subjects and also performed normally on the four tests of nonverbal short-term memory. The patients with Korsakoff's syndrome had a marginally low digit span and performed poorly on three of the four nonverbal tasks, a finding consistent with the deficits in attention and visuospatial processing previously described for this patient group. These deficits are likely due to the frontal lobe atrophy typically associated with Korsakoff's syndrome, rather than to diencephalic damage. The results support the view that short-term (immediate) memory, including short-term spatial memory, is independent of the hippocampus.

Amnesic patients and control Ss read the names of famous and nonfamous persons. Subsequently, both groups were more likely to designate a name as famous if it had been encountered previously. The facilitatory effect of prior presentation was similar for amnesic patients and control Ss and similar for famous and nonfamous names. For amnesic patients, the effect occurred despite severely impaired recognition memory for the names. In a 2nd experiment, recombining the first and last names that had been presented together did not diminish the facilitatory effect of prior presentation, which indicates that the effect does not depend on forming an association between first and last names. The results show that nondeclarative (implicit) memory can support the acquisition of information that is specific (e.g., names of persons) and that has no preexisting representation (e.g., nonfamous names).

Inhibition of the adrenocortical axis by glucocorticoids (GCs) occurs at both hypothalamic and suprahypothalamic sites. In the rat, the hippocampus has been shown to be an essential suprahypothalamic site. The present study shows that the hippocampal system serves a similar role in the nonhuman primate. Bilateral lesions that included the hippocampal formation and the parahippocampal cortex; the hippocampal formation, parahippocampal cortex, and the amygdala; or the fornix all produced GC hypersecretion in cynomolgus monkeys. The hypersecretion occurred throughout the day. Moreover, these lesions were also associated with dexamethasone resistance (i.e., GC hypersecretion following administration of the synthetic GC dexamethasone). The hypersecretion could not be attributed to acute surgical trauma, because neither circumscribed lesions of the amygdala nor conjoint lesions of the perirhinal and parahippocampal cortex produced adrenocortical abnormalities. Finally, in agreement with data derived from the rat, the GC hypersecretion following hippocampal lesions was transient. Secretory activity returned to normal levels by 6-15 months in all operated groups. Thus, the primate hippocampal system appears to share some neuroendocrine functions with the rodent.

Amnesic patients (n = 9) and normal subjects (n = 12) read lists of unique words, repeated words, unique nonwords, and repeated nonwords as quickly as possible. In the first experiment both groups of subjects read the lists of repeated items faster than the lists of unique items and improved at the same rate within each list. In the second experiment, subjects read four new lists of items and then reread the same four lists after a 10-min delay. The results replicated the findings from the first experiment and demonstrated in addition that the facilitated reading speed persisted across the delay. These results show that the acquisition of novel verbal information can be supported by nondeclarative (implicit) memory. It is suggested that facilitated reading speed for words and nonwords reflects changes in early-stage perceptual systems and that these changes occur independently of the brain structures damaged in amnesia.

Studies of human amnesia and studies of an animal model of human amnesia in the monkey have identified the anatomical components of the brain system for memory in the medial temporal lobe and have illuminated its function. This neural system consists of the hippocampus and adjacent, anatomically related cortex, including entorhinal, perirhinal, and parahippocampal cortices. These structures, presumably by virtue of their widespread and reciprocal connections with neocortex, are essential for establishing long-term memory for facts and events (declarative memory). The medial temporal lobe memory system is needed to bind together the distributed storage sites in neocortex that represent a whole memory. However, the role of this system is only temporary. As time passes after learning, memory stored in neocortex gradually becomes independent of medial temporal lobe structures.

A technique is described for producing accurate stereotaxic lesions of the hippocampus in monkeys. This technique overcomes the problem that the size and shape of the brain can vary considerably from monkey to monkey. Magnetic resonance imaging (MRI) is used to create an individual brain atlas for each monkey. The atlas is then used to derive coordinates for making stereotaxic radio frequency lesions of the hippocampus. There are two key features of this procedure. First, a specially-designed, acrylic, stereotaxic headholder was constructed that could be used safely with the MR magnet. Second, small glass beads, anchored to the skull of the monkey, served as common landmarks from which lesion coordinates were determined in the MR images and then again in neurosurgery. MRI techniques are also described for determining the extent of tissue damage postoperatively. This technique could also prove useful in other areas of neuroscience research that depend on accurate stereotaxic placement of electrodes (e.g., electrophysiological studies and neuroanatomic tracing studies).

The hippocampus has sometimes been proposed to function as a cognitive map, a memory system that stores information about allocentric space. Work with experimental animals and memory-impaired patients has raised difficulties with this view by showing that the hippocampus is not performing an exclusively spatial function. However, the possibility has remained that the hippocampus plays a special role in spatial memory or a disproportionately large role in spatial memory compared to other kinds of memory. This study compared spatial and nonspatial memory in amnesic patients with lesions of the hippocampal formation or diencephalon. Subjects studied an array of 16 toy objects and were subsequently tested for object recall, object recognition, and memory for the location of the objects. Control subjects were tested after long retention intervals in order to equate their object memory performance with that of the patients. The main finding was that, when the performance of amnesic patients on the object memory tests was matched to the object memory performance of control subjects, spatial memory performance of the amnesic patients also matched the spatial memory performance of the control subjects. The results were the same for the two groups of patients. These findings suggest that the hippocampus is not especially involved in spatial memory. Spatial memory is simply one instance of a broader category of memory that requires the hippocampus. While cognitive mapping in its most abstract sense may describe hippocampal function, our results support alternative formulation, suggesting that the hippocampus is necessary for the rapid acquisition of relational, configural, or declarative (as opposed to purely spatial) information.

Structures and connections in the medial temporal lobe of humans and nonhuman primates have long been recognized as important for normal memory and emotional behavior. The present study investigated memory and emotional behavior in normal monkeys and six groups of monkeys with lesions of the medial temporal lobe. Two groups had damage to the hippocampal formation (or adjacent perirhinal and parahippocampal cortex) but not the amygdaloid complex; two groups had either partial or complete damage to the amygdaloid complex but not the hippocampal formation; and two groups had damage to both the hippocampal formation and the amygdaloid complex. Memory was evaluated with three tasks sensitive to human amnesia: (1) delayed nonmatching to sample; (2) retention of object discriminations; and (3) concurrent discrimination learning. Emotional behavior was assessed by measuring the responsiveness of monkeys to 12 different stimulus situations. Damage to the hippocampal formation or anatomically related cortex impaired memory but did not affect emotional behavior. Partial or complete damage to the amygdaloid complex affected emotional behavior but not memory. These findings show that memory impairment and abnormal emotional behavior are anatomically dissociable and independent effects of damage to the medial temporal lobe.

Recently, Wood and his colleagues (1989) presented a case of childhood amnesia as evidence against the distinction between declarative and procedural memory that has sometimes been applied to human amnesia. Their argument was based on the observation that their patient showed some progress in school over the years, i.e., acquired some declarative knowledge, despite severely impaired day-to-day memory ability. We briefly review their case, together with a carefully studied second case of childhood amnesia not mentioned by Wood et al. Their argument is wrong in several ways. First, the utility of the declarative/procedural distinction for amnesia, or the utility of any other distinction between memory systems, depends on whether or not one kind of memory is impaired selectively, not on the severity of the impairment. In particular there is no requirement that one kind of memory be totally absent. Second, they have not provided the data necessary to support their argument; namely, data showing that the amount of declarative knowledge accumulated during years in school was better than would have been expected given the capacity for moment-to-moment or day-to-day memory. Indeed, the patient's moment-to-moment memory ability is better than represented, and the patient's progress in school was abnormally slow. Third, it is not clear that academic achievement scores provide a direct measure of declarative memory abilities (skill learning and recovery of function may also have contributed). We conclude that the evidence from childhood amnesia is fully consistent with the proposal that amnesia reflects a selective impairment in the formation of long-term declarative memory.

Amnesic patients were studied to determine whether the acquisition and retention of item-specific skills can be supported by nondeclarative (implicit) memory. In Experiment 1, subjects read 2 different passages 3 times in succession. Reading speed improved at a similar rate in both amnesic patients and normal subjects and was specific to the text that was read. In Experiment 2, amnesic patients and normal subjects read a passage 3 successive times and then reread the same passage after a 0-s, 10-min, 2-hr, or 1-day delay. In both groups, facilitation persisted for at least 10 min and disappeared within 2 hr. It is suggested that facilitated reading speed depends importantly on both semantic and perceptual information and that such information can be supported by nondeclarative memory.

Clinical and experimental studies have shown that the hippocampal formation and related structures in the medial temporal lobe are important for learning and memory. Retrograde amnesia was studied prospectively in monkeys to understand the contribution of the hippocampal formation to memory function. Monkeys learned to discriminate 100 pairs of objects beginning 16, 12, 8, 4, and 2 weeks before the hippocampal formation was removed (20 different pairs at each time period). Two weeks after surgery, memory was assessed by presenting each of the 100 object pairs again for a single-choice trial. Normal monkeys exhibited forgetting; that is, they remembered recently learned objects better than objects learned many weeks earlier. Monkeys with hippocampal damage were severely impaired at remembering recently learned objects. In addition, they remembered objects learned long ago as well as normal monkeys did and significantly better than they remembered objects learned recently. These results show that the hippocampal formation is required for memory storage for only a limited period of time after learning. As time passes, its role in memory diminishes, and a more permanent memory gradually develops independently of the hippocampal formation, probably in neocortex.

Studies of circumscribed amnesia have been a useful source of information about the organization of human memory functions. In an effort to obtain neuroanatomical information about the patients being studied, we have used 2 high-resolution protocols for imaging the human brain with magnetic resonance (MR). One provides images of the hippocampus, permitting visualization of the hippocampal formation in considerable detail. The second provides images of the mammillary nuclei. Four amnesic patients with alcoholic Korsakoff's syndrome had abnormally small mammillary nuclei that were barely detectable by MR in most cases. The temporal lobe, hippocampal formation, and parahippocampal gyrus were of normal size. In a second group of 4 (non-Korsakoff) amnesic patients, the opposite findings were obtained. The hippocampal formation was markedly reduced in size in the absence of a detectable change in the size of the temporal lobe. By contrast, the mammillary nuclei, though somewhat reduced in volume, were considerably larger than in the patients with Korsakoff's syndrome. The anatomical findings for individual patients generally corresponded to the severity of their memory impairment. The results show that neuroimaging techniques can distinguish between patients with diencephalic and medial temporal lobe amnesia. These techniques can therefore provide useful information about the crucial issue of patient classification, and they set the stage for exploring possible quantitative and qualitative differences between patients.

Patients with frontal lobe lesions, amnesic patients with Korsakoff's syndrome, other (non-Korsakoff) amnesic patients, and control subjects were given tests of memory for temporal order. In the first experiment, subjects were presented with a list of 15 words and then asked to reproduce the list order from a random array of the words. In the second experiment, they were asked to arrange in chronological order a random display of 15 factual events that occurred between 1941 and 1985. In both experiments, patients with frontal lobe lesions were impaired in placing the items in the correct temporal order, despite normal item memory (i.e. normal recall and recognition memory for the words and facts). The two groups of amnesic patients exhibited impaired memory for temporal order as well as impaired item memory. Patients with Korsakoff's syndrome exhibited poorer temporal order memory than the other amnesic patients, despite similar levels of item memory. These findings demonstrate that patients with frontal lobe lesions have difficulty organizing information temporally. Patients with Korsakoff's syndrome, who have both diencephalic and frontal damage, have memory impairment together with a disproportionate deficit in memory for temporal order.

The performance of three kinds of amnesic patients and control subjects was assessed using four methods for testing memory: free recall, recognition, cued recall, and word completion. Whereas amnesic patients were impaired on free recall, recognition, and cued recall, they were normal on word completion. Moreover, performance on the word-completion test declined at a normal rate reaching chance after about 120 min. The word-completion test resembled the cued-recall test in that the initial letters of previously presented words were given as cues. It differed from cued recall only in the instructions, which directed subjects away from the memory aspects of the test and asked them to complete each three-letter cue with the first word that came to mind. The present results offer an explanation of conflicting findings that have been obtained with amnesic patients on tests of the cued-recall type. The results are considered in terms of a process (activation or procedural learning), which is spared in amnesia and not dependent on the integrity of the damaged brain regions.

Electrical stimulation with brief pulses can produce a seizure requiring less energy than conventional sine-wave stimulation, and it has been suggested that brief-pulse stimulation might reduce the memory loss associated with electroconvulsive therapy (ECT). The authors evaluated the effects of electroconvulsive shock (ECS) on memory in mice by using various waveforms, current intensities, training-ECS intervals, pulse widths, and stimulus durations. When equated for ability to produce seizures, low-energy, brief-pulse stimulation caused as much amnesia as sine-wave stimulation and sometimes more. In the absence of comparisons of the amnesic effects of brief-pulse and sine-wave stimulation in humans, the use of brief pulses for administering ECT is unwarranted.

Inhibition of cerebral protein synthesis impairs long-term memory in a variety of species and tasks. Recently it was reported that subcutaneous injection of the protein synthesis inhibitor cycloheximide impaired short-term retention, measured 10 min after training in a passive avoidance task. To examine the possibility that inhibition of cerebral protein synthesis may sometimes disrupt short-term memory, mice were injected subcutaneously with cycloheximide (120 mg/kg) or anisomycin (150 mg/kg), or bitemporally with cycloheximide or anisomycin (100 mug/side) and given one training trial in a passive avoidance box. Subcutaneously injected cycloheximide reduced step-through latencies 10 min after training as reported previously, but anisomycin or bitemporally injected cycloheximide did not. All 4 drug groups exhibited impaired long-term memory. Since the results obtained at short intervals after training varied depending on the drug and route of injection, the impairment produced by subcutaneous cycloheximide at 10 min after training cannot be attributed to inhibition of cerebral protein synthesis. It is suggested that performance at short intervals after training reflects drug side effects on step-through behavior. By contrast, the impairment obtained at long intervals after training is consistent with the hypothesis that cerebral protein synthesis is required for formation of long-term memory.

In an effort to bring into correspondence the findings from human amnesic patients and the findings from monkeys with surgical lesions of those brain regions thought to be affected in the human cases, we have addressed in three experiments the implication of findings that human amnesia spares motor and cognitive skills. In the first experiment, monkeys with conjoint lesions of hippocampus and amygdala (H-A), which reproduced the surgical removal sustained by the noted amnesic case H.M., were only mildly impaired in learning relatively difficult pattern discrimination tasks. Monkeys with lesions of temporal stem matter (TS) were severely impaired on the same tasks, due to an apparent deficiency in visual information processing. In the second experiment, monkeys with H-A lesions were severely impaired at learning relatively easy discrimination tasks that could be acquired rapidly by normal monkeys. Monkeys with TS lesions were not impaired. In the third experiment, monkeys with H-A lesions exhibited normal acquisition of two motor skill tasks. These data can be understood in the light of a distinction between kinds of memory, founded in recent studies of the neuropsychology of human amnesia. These studies have led to a distinction between the learning of skills or procedures, which is spared in human amnesia, and the learning of facts and episodes, which is impaired. Monkeys with H-A lesions are normal at skill learning like human amnesic patients with similar lesions. This conclusion depends in part on the argument developed here that pattern discrimination learning, as accomplished by monkeys, has a large skill-like component. These results bring into correspondence the behavioral data from human amnesic patients and operated monkeys and set the stage for identifying precisely what brain structures must be damaged to produce amnesia.

Recent successes in developing an animal model of human amnesia in the monkey have made it feasible to try to identify with certainty the specific structures in the diencephalon and medial temporal region that cause amnesia when damaged. Monkeys with small lesions restricted largely to the posterior portion of the mediodorsal nucleus of the thalamus were given a test of memory sensitive to human amnesia and a second test that is analogous to the skill-based tasks performed normally by amnesic patients. The monkeys exhibited a marked impairment on the first test and performed normally on the second. The results show that circumscribed lesions of the mediodorsal nucleus can cause substantial amnesia.

Using a task known to be sensitive to human amnesia, we have evaluated two current hypotheses about which brain regions must be damaged to produce the disorder. Monkeys with bilateral transections of the white matter of the temporal stem were unimpaired, but monkeys with conjoint amygdala-hippocampal lesions exhibited a severe memory deficit. The results indicate that the hippocampus, amygdala, or both, but not the temporal stem, are involved in memory in the monkey and suggest that a rapprochement between the findings for the human and the nonhuman primate may be close at hand.

Mice were given a single training trial and then received either sham treatment or electroconvulsive shock (ECS; four treatments at hourly intervals) at 1 of 7 times (1-70 days) after training. Retention was always tested 2 weeks after treatment. Control animals exhibited gradual forgetting with increasing retention intervals. Mice given ECS exhibited severe retrograde amnesia, which diminished as the interval between training and ECS increased from 1 to 21 days. ECS given 21-70 days after training had no effect on memory. The finding of long, temporally graded retrograde amnesia in mice establishes continuity between the results for laboratory animals and those for humans and indicates that the neural changes involved in the consolidation of memory can continue for a significant portion of the lifetime of a memory.

Whether frontal lobe pathology can account for some of the cognitive impairment observed in amnesic patients with Korsakoff's syndrome was investigated. Various cognitive and memory tests were given to patients with circumscribed frontal lobe lesions, patients with Korsakoff's syndrome, non-Korsakoff amnesic patients, and control Ss. Patients with frontal lobe lesions were not amnesic. Nevertheless they exhibited 2 deficits that were also exhibited by patients with Korsakoff's syndrome but not by other amnesic patients: (a) impairment on the Wisconsin Card Sorting Test and (b) impairment on the Initiation and Preservation subscale of the Dementia Rating Scale. Thus, frontal lobe pathology can explain some of the cognitive deficits observed in patients with Korsakoff's syndrome.

Quantitative analyses were performed on computer tomography (CT) scans from 7 patients with Korsakoff's syndrome, 7 age-matched alcoholic subjects, and 7 age-matched healthy control subjects. CT values were used to estimate tissue density and fluid volume in specified brain areas. Tissue density was assessed by averaging CT values in small (5 x 5 mm) areas sampled bilaterally in 6 specified areas--thalamus, head of the caudate nucleus, putamen, anterior white matter, posterior white matter, and centrum semiovale. We assessed fluid volume using a semiautomated computer algorithm that estimated the proportion of fluid in 7 brain regions--total ventricular space, third ventricle, interventricular region, frontal sulci, peri-Sylvian region, medial cerebellum, and vertex. For the patients with Korsakoff's syndrome, we also assessed the correlation between CT measures and performance on 6 cognitive and 12 memory tests. Compared with alcoholic subjects and healthy control subjects, patients with Korsakoff's syndrome had lower CT density values bilaterally in the region of the thalamus and had greater estimated fluid bilaterally in the region of the third ventricle. Alcoholic and healthy control subjects did not differ on these measures. Significant cortical atrophy in frontal sulcal and peri-Sylvian areas was detected both in patients with Korsakoff's syndrome and in alcoholic subjects. For patients with Korsakoff's syndrome, impairment on behavioral tests, and on memory tests in particular, was correlated with low-density values in the thalamus and with high fluid values in the region of the frontal sulci. Damage to diencephalic and frontal areas may especially contribute to the memory and cognitive impairment exhibited by patients with Korsakoff's syndrome.

A group of tasks sensitive to human amnesia were used to characterize the severity and duration of memory impairment in monkeys following bilateral damage to the hippocampal formation, fornix, or mammillary nuclei. Monkeys with hippocampal formation lesions (which included the hippocampus proper, dentate gyrus, subiculum, posterior entorhinal cortex, and much of the parahippocampal gyrus) exhibited a substantial and lasting memory impairment. Monkeys with fornix transection or bilateral damage to the mammillary nuclei were impaired on the first task administered after surgery (delayed nonmatching to sample). However, they performed all the other tasks normally and were unimpaired when the delayed nonmatching to sample task was re-administered 18 months after surgery. The findings are consistent with reports that damage limited to the human hippocampus can produce a clinically significant and permanent amnesia. Because fornix transection or mammillary lesions produced only transient memory impairment, it seems unlikely that similar damage in humans can cause a severe or permanent amnesia.

The authors evaluated the effects on memory of ECT given with either unilateral or bilateral electrode placement and with brief-pulse or sine-wave stimulus waveform. Clinical criteria determined the mode of ECT and the treatment parameters. As expected, right unilateral ECT produced less memory impairment than bilateral ECT. Brief-pulse ECT resulted in less memory impairment than sine-wave ECT during the first hour after treatment but had similar effects on memory after the first hour. Brief-pulse ECT might produce less memory impairment than conventional sine-wave ECT; however, this can probably be achieved in clinical practice only if treatment parameters that keep stimulation close to the seizure threshold are developed individually for each patient.

N.A. has been amnesic since 1960 when at the age of 22 years he sustained a penetrating brain injury with a miniature fencing foil. The amnesia primarily affects verbal material and occurs in the absence of other detectable cognitive deficits. Previous CT scans demonstrated a lucency in the region of the left mediodorsal thalamic nucleus, but no additional damage was revealed. Beginning in 1986 when he was 48 years old, N.A. was evaluated with a series of magnetic resonance imaging (MR) studies. Three major areas of damage were identified. In the left thalamus there is a prominent 3- to 4-mm-wide linear lesion that approximates the position and orientation of the internal medullary lamina. The defect extends for approximately 20 mm anteroposteriorly and likely involves the rostral group of intralaminar nuclei (central medial, paracentral, central lateral, rhomboid, and reuniens nuclei), the caudal group of intralaminar nuclei (centrum medianum and parafascicular nuclei), the ventral aspect of the mediodorsal nucleus, and the ventral lateral and ventral anterior nuclei. It also likely interrupts the trajectories of the mammillothalamic tract and postcommissural fornix. The posterior hypothalamus is markedly disrupted and the mammillary nuclei appear to be missing bilaterally. Finally, the right anterior temporal lobe is damaged for a distance of about 3.5 cm from the pole to midway through the amygdaloid complex. This damage probably occurred during exploratory neurosurgery done at the time of N.A.'s injury. The hippocampal formation appears intact on both sides. A comparison of these findings with those from other patients with diencephalic amnesia suggests that amnesia can result when several diencephalic structures are damaged conjointly, including the internal medullary lamina, the intralaminar nuclei, the mediodorsal nucleus, and the mammillothalamic tract. Whether amnesia as severe as N.A.'s would result from selective damage to any one of these structures remains to be determined.

The performance of amnesic patients was assessed on five tasks, which have figured prominently in the development of animal models of human amnesia in the monkey. The amnesic patients were impaired on four of these tasks (delayed nonmatching to sample, object-reward association, 8-pair concurrent discrimination learning, and an object discrimination task), in correspondence with previous findings for monkeys with bilateral medial temporal or diencephalic lesions. Moreover, performance of the amnesic patients correlated with the ability to verbalize the principle underlying the tasks and with the ability to describe and recognize the stimulus materials. These tasks therefore seem to be sensitive to the memory functions that are affected in human amnesia, and they can provide valid measures of memory impairment in studies with monkeys. For the fifth task (24-hour concurrent discrimination learning), the findings for the amnesic patients did not correspond to previous findings for operated monkeys. Whereas monkeys with medial temporal lesions reportedly learn this task at a normal rate, the amnesic patients were markedly impaired. Monkeys may learn this task differently than humans.

Memory for past events was assessed in 43 patients who had been prescribed bilateral electroconvulsive therapy (ECT) for relief of depressive illness. Four memory tests of personal or public events were administered before ECT, shortly after the fifth treatment, one week after completion of treatment, and about seven months later. The results indicated that ECT can initially disrupt recall of events that occurred many years previously, but recovery of these memories was virtually complete by seven months after treatment. It was also clear that persisting memory loss for information acquired only a few days before treatment can occur. For information acquired one to two years prior to treatment, recovery was substantial, but the results suggested that some memory problems might persist for events that occurred during this time period.

Amnesic patients and control Ss performed similarly on 2 memory tests. In Experiments 1A and 1B, amnesic patients exhibited intact adaptation-level effects: An experience lifting and judging a group of weights influenced their judgments of a 2nd group of weights 20-25 min later. The effect did not depend on peripheral accommodation, because Ss used 1 hand during their 1st encounter with the weights and the opposite hand during their 2nd encounter. In Experiment 2, amnesic patients acquired at a normal rate the ability to perceive binocular depth using random-dot stereograms. In both experiments, amnesic patients benefited from recent experience, despite the fact that they could not remember their prior experience accurately. The preserved memory abilities demonstrated here appear to be examples of implicit, or nondeclarative, memory.

Recent work with amnesic patients has revealed a preserved capacity for acquiring and retaining new skills despite otherwise profound anterograde impairment. In addition to their anterograde impairment, amnesic patients also have retrograde memory loss for some information acquired prior to the amnesic event. The present experiment addresses for the first time the question of whether preservation of memory for skills is also a feature memory impairment. To determine the susceptibility of a recently learned skill to retrograde amnesia, we taught patients to read mirror-reversed words before and during the early part of a prescribed course of electroconvulsive therapy (ECT) and then tested retention of the skill after the course of treatment had been completed. Patients prescribed bilateral or right unilateral ECT and depressed patients not receiving ECT acquired the mirror-reading skill at the same rate and then retained it at the same level. For the patients prescribed ECT, intact learning and retention of the skill occurred despite retrograde amnesia for the previous testing sessions and for the words that they had read previously.

The neurology of memory has been illuminated by parallel studies of patients with circumscribed memory impairment and animal models of human amnesia. Human amnesia can occur as an isolated cognitive deficit that impairs the ability to learn new facts and episodes. In addition, memory can be affected for material learned many years prior to the onset of amnesia. The finding that some memory abilities are intact in amnesia (e.g., skill learning, word priming, and adaptation-level effects) has suggested that memory can be divided into two or more separate processes. Declarative memory affords the ability to store information explicitly and to retrieve it later as a conscious recollection. This form of memory depends on the integrity of the structures damaged in amnesia. Other, non-declarative kinds of memory afford the ability to change as the result of experience, but the information is available only through performance. Recent studies of a favorable human case provided strong evidence that the hippocampus is a critical component of the declarative memory system. Extensive convergent and divergent projections link the hippocampus to many areas of neocortex where processing and storage of new information is likely to occur. It is perhaps by way of these connections that the hippocampus operates upon and participates in declarative representations.

Four patients with subcortical lesions and either aphasia or amnesia were compared to four patients with cortical lesions and aphasia. Each patient had [18F] fluorodeoxyglucose positron-emission computed tomography, language, and memory evaluations. Marked metabolic depression was found in the thalamus and caudate in all patients. The subcortical patients had only mild left-cortical changes, while the cortical patients showed marked cortical metabolic changes in the left hemisphere. Language changes were mild in the subcortical patients, while moderate to severe in the cortical patients. All patients showed severe verbal memory dysfunction. The only common abnormalities in the two groups were metabolic changes in the thalamus and severity of verbal memory dysfunction. These findings suggest a relationship between verbal memory and thalamic function.

We assessed priming of new associations in amnesic patients and healthy control subjects in a paradigm developed by Graf and Schacter (1985). Subjects were presented unrelated word pairs embedded in sentences (e.g., A BELL was hanging over the baby's CRADLE) and were asked to rate how well the sentences related the two words. Subjects were then given a word completion test. They were shown three-letter word stems and were asked to complete the stem with the first word that came to mind. In the same context condition, each word stem was presented together with the word that had appeared in the same sentence during study (e.g., BELL-CRA--). In the different context condition, each stem was presented together with a new word that had never been presented (e.g., APPLE-CRA--). Control subjects completed more words in the same context condition than in the different context condition. In contrast, amnesic patients did not complete any more words in the same context condition than in the different context condition. Indeed, across two experiments none of the amnesic patients exhibited consistent priming of new associations. Thus, although amnesic patients do exhibit entirely normal priming of preexisting memory representations, they do not appear to exhibit priming of new associations in this paradigm.

Spontaneous firing rates and changes in firing rate in response to an intravenously administered dose of apomorphine were measured after various electroconvulsive shock (ECS) treatment regimens from dopaminergic cells of the substantia nigra in urethane-anesthetized rats. Similar measurements were obtained from noradrenergic neurons of the locus coeruleus before and after intravenous injection of clonidine. A significant decrement in the inhibition of spontaneous firing in response to intravenous administration of these agonists was observed following multiple or single ECS treatment in both substantia nigra and locus coeruleus cells. There was a consistent but nonsignificant tendency for cells in both areas of the brain from treated animals to display higher rates of spontaneous firing than their respective sham-shocked controls. Both the effects on base-line rates of spontaneous activity and on the depression of firing rate in response to drug administration were found to be independent of repeated treatment. A significant negative correlation was obtained between base-line firing rate and percentage depression to the autoreceptor agonist, but this correlation alone was insufficient to account for the observed differences in the drug response. These results are discussed with respect to possible mechanisms of action of electroconvulsive therapy in the treatment of depression.

Certain features of abnormal memory, which have figured prominently in theoretical treatments of the amnesic syndrome, were assessed in patients with Korsakoff's syndrome, in Case N.A., and in patients receiving electroconvulsive therapy. Patients with Korsakoff's syndrome differed from the other patients by (a) failing to exhibit release from proactive interference, and (b) being disproportionately impaired in the ability to make judgments about the temporal order of recent events. These deficits appear to be related to frontal lobe damage and are superimposed on a more basic memory disorder. Theories of amnesia should be founded on the basic memory disorder and not on deficits such as these, which have no obligatory relationship to amnesia. Dissociations between aspects of memory, revealed by the study of amnesia, can also shed light on the organization of memory in the brain.