Memory: A Neurosurgeon's Perspective

Joseph E. Bogen

Neurosurgeons commonly encounter a variety of amnesias (i.e., memory deficits), especially those ascribable to a specific event. Familiar amnestogenic events include head injuries and subfrontal hemorrhages as well as operations on either temporal lobe or in the vicinity of the third ventricle. Consequent to an amnestogenic event, deficits may include occurrences before (retrograde amnesia) and after (anterograde amnesia) the event.
Memory deficits can be from a failure to register some occurrence (e.g., from lack of attention), a failure to store it (by a process usually called consolida tion), or a failure to retrieve from previously stored memories. Retrieval failures include both inability to recall some event or object and, in more severe cases, inability even to recognize some recently seen object. Distinguishing among these alternatives has been a persistent concern in memory research, with considerable progress in the past few decades. Recent research has particularly emphasized the anatomical distinction between hippocampal-dependent and hippocampal-independent memory processes, in parallel with an associated clinical distinction between the amnestic syndrome and other types of memory problems (16, 26).
Whereas the term amnesia is loosely applied to many types of memory deficit, the term amnestic syndrome (also known as "classical amnesia") is more specific. It is characterized by a disability for learning new material (i.e., anterograde amnesia) despite good memory in other respects. It is now known that this disability results from a storage failure, characteristically the result of damage including the hippocampus. The difficulty in learning new material contrasts with the relatively good retrieval of old material; the subject can often recall circumstances before the amnestogenic event quite well. Moreover, the amnestic patient typically has a normal immediate memory, commonly tested by measuring the digit span--how long a string of digits the subject can repeat after hearing it once (normal being approximately seven to nine digits).
Short-Term Memory
It is common to distinguish between short-term and long-term memory. In each case, however, there are different types and durations. For example, shortterm memory includes immediate memory, as in testing for digit span. It also includes what is called "working memory," which can comprise a large variety of material currently kept in mind while it is being processed. One approach to studying the neuronal basis of working memory involves having a monkey, with implanted electrodes, make a choice when a signal (e.g., a light flash) follows a cue after a delay of several seconds. The correct choice depends on which of two cues is presented at the beginning of the delay. For example, saccade to the right at the signal if the visual cue was a circle but saccade to the left if the cue was a square. In these experiments, certain neurons will become active on presentation of the cue and will continue firing until the delay ends with the signal to respond. Such neurons are commonly found in prefrontal cortex. They can be found in other regions (premotor, parietal, temporal) depending on the nature of the cue, the signal, and the response required (8). The preserved working memory enables the amnestic subject to carry on a sensible conversation or to cooperate with extensive testing or training--little of which the subject will remember a short time later, much less the next day.
Short-term memory testing also includes exposing human subjects to material and asking them to respond 30 or 60 minutes later to see how much they have retained. This is usually done both with verbal material and with unnameable figures that the patient copies. Both are used because it is now generally recognized that verbal and pictorial memories tend to be differentially represented, tending toward left and right hemispheres, respectively, in right handers (12, 13). Memories stored both verbally and pictorially are examples of dual coding (15). To the extent that this involves both cerebral hemispheres, it helps to explain memory deficits that can occur after corpus callosotomy (25).
Procedural Memory
Surprisingly, the amnestic patient shows a benefit from nonremembered training sessions. For example, the subject will steadily improve with daily testing on a maze or a rotary pursuit problem despite having no recollection, when asked, of previous training sessions or indeed of ever having met the person who gave the daily training sessions. This striking dissociation has given rise to the distinction between episodic memory and procedural memory. The term declarative memory (or propositional memory) includes both episodic memory (recall of a specific episode) and semantic memory (recall of a cluster of undated information about some subject such as the meanings of words).
Declarative memory is tested by what the subject says (in humans, but see below regarding monkeys), whereas procedural memories are demonstrated by carrying out a procedure. Procedural memory includes various skills or learned habits that are usually acquired over multiple trials with gradual improvement; this is in contrast to declarative memory, which can be acquired more quickly, often on a single trial. Evidence has progressively accumulated to show that procedural memories (running a maze, playing a scale, and similar actions) depend on neocortical interaction with the striatum rather than with the hippocampus (10).
Medial Temporal Lobe Amnesia
The forgoing distinctions have developed as the result of research that was particularly stimulated by studies of a single patient, H.M. In 1953, this man (aged 27 years at the time) had bilateral hippocarnpectomy together with removal of underlying cortex and amygdala. His resulting amnestic syndrome was first reported by Scoville and Milner (6a), and his psychological status has been carefully studied for many decades (7). His deficits and those of similar patients have led to the recognition that consolidation of episodic memory depends on hippocampal function, whereas hippocampal function is unnecessary for the retrieval of previously acquired memories. The hippocampal lesions needed to cause an anterograde loss of episodic memory can be quite small, in at least one case restricted to the CA, field bilaterally. For a more severe anterograde amnesia, the lesions typically include the underlying perirhinal and entorhinal cortices (26).
The distinctive amnestic syndrome in humans from bimedial temporal damage stimulated many attempts, mostly unsuccessful, to demonstrate a memory deficit in animals with similar lesions. Finally, in 1978, Mishkin described a successful test called delayed non-matching-to-sample (16). Neurosurgeons are aware that one of the most reliable deficits from a bilateral prefrontal lobectomy in monkeys is their inability to perform tasks involving a delayed response. The monkey sees two objects through a window. A reward (e.g., a banana chip) is placed under one of the objects. The window is then covered for a delay, after which the window is opened so the monkey can lift the correct object to obtain the reward. This is a test of working memory that, as described above, commonly involves prefrontal cortical neurons. The monkeys with bimedial temporal damage but intact frontal cortex have little difficulty with delay alone. However, if the test is compounded by using a large number of obj ects presented two at a time and requiring that the monkey choose the object that was not previously correct, the bimediotemporally lesioned monkey fails (16). It has been suggested that this deficit in monkeys corresponds to the declarative memory deficit in humans (26).
Role of the Amygdala
The likelihood that data being processed will be stored often depends on its affective load, that is, the accompanying emotional significance. Hence, the amygdalae and related structures are essential for certain kinds of memories (such as conditional autonomic responses) but not for others. A patient with selective bilateral damage of the amygdala did not acquire a conditioned skin conductance response when conditioning was attempted but had no deficit in declarative memory for the details of the experiment. By contrast, a patient with bilateral hippocampal damage did acquire the conditioned skin conductance response but without any recollection of the training sessions (2). Experiments using Pavlovian conditioning for somatic responses (e.g., eyeblink) have also implicated the cerebellum (17, 21).
Diencephalic Amnesia
An amnestic syndrome much like that from bimedial temporal lobe damage can also be caused by bilateral medial thalamic damage. The exact thalamic locations have been heatedly debated for many years (11). The most effective small lesions probably involve, simultaneously and bilaterally, the connections from amygdalae to the mediodorsal thalamic nuclei (MD) and the nearby mamillothalamic tracts (9). The nearby descending columns of the fornices may also be relevant.
In each hemisphere, the connections from amygdala to ipsilateral MD form one side of a triangular circuit, because the MD project to the same orbitefrontal cortex receiving amygdalocortical fibers. I)amage to the MD-orbital connections along with neighboring septal nuclei (and possibly nuclei basales of Meynert) is the likely explanation for memc,ry deficits associated with anterior communicating aneurysm hemorrhage and/or surgery. Deficits following disconnection within the third side of the triangle, amygdalofrontal connections, remain undetermined.
The extent to which isolated lesions of any one of the diencephalic structures alone can cause memory tleficits is debated with conflicting evidence. At present, the multiplicity of diencephalic memory circuitry even within a single hemisphere suggests that a grossly evident amnesia generally requires multiple lesions.
Neocortical Storage
Because hippocampal ablation minimally affects retrieval of older memories, it is clear that these memories are not stored in hippocampus. They are stored in neocortex, in those higher-order cortices where the processing of the percepts originally occurred (8,22).
The likelihood that information being processed will be stored depends not only on its emotional importance, but also on the condition of cortex as affected by a variety of modulatory substances. For example, the greatly reduced output to cortex from locus coeruleus of norepinephrine, which is typical of rapid eye movement sleep, may explain the rapid fading of dreams (6).
Focal cortical storage has long provided an explanation for the loss of material-specific memories (e.g., the nonrecognition of faces in prosopagnosia, loss of memories of once familiar places, or cases of anemia for animate but not inanimate objects) following focal cortical lesions (14). However, the reality is not that simple, as will be discussed in the following section.
Inhibition of Recall
Memories ordinarily available can be unavailable momentarily or for even longer periods. Because the memories are intact (as evidenced by recall at other times), the retrieval process would appear to be temporarily inhibited (1). How this occurs is suggested by some simpler cases in which there is inhibition of performance. One example is the loss of the contact placing reaction in the cat forepaw contralateral to a sensorimotor cortex ablation. This result could (and did) suggest that the competence (or the "memory") for paw contact placing resided in the sensorimotor cortex. However, a subsequent removal of the ipsilateral sensorimotor cortex was abruptly followed by reappearance of paw contact placing in the affected forepaw (4, 5). Analogously, loss of responsiveness to visual stimuli in the contralateral half-field is routinely observed after extensive unilateral posterior corticectomy in the cat; However, destruction of the contralateral superior colliculus or substantia nigra abruptly restores responsiveness to visual stimuli in the affected half-field (20, 24).
These experiments in cats, using a second lesion, showed that a performance deficit could be the result of unbalanced inhibition rather than simply loss of the underlying competence. That is, after the first lesion has resulted in loss of performance, a second lesion strategically placed is abruptly followed by reemergence of the performance. Would this work with loss of cortical function from strokes in humans? Recent results with the syndrome of hemineglect suggest that it might.
A patient with hemineglect from a right parietal lesion ignores stimuli in left hemispace. In a simple but eloquent experiment by Bisiach and Luzzati (3), patients with right hemisphere lesions were asked to imagine themselves at one end of the Piazza del Duomo in Milan and describe all the places of business on the plaza. They failed to recall shops, cafes, etc., on the left. Remarkably, when imagining themselves at the other end of the plaza, they named the previously neglected places but omitted those recalled before. It is clear that the memories of each side of the plaza were either available to consciousness or were not, depending on their relation to the subjects' imagined body-centered coordinates. A likely explanation, in physiological terms, is that access to consciousness of memories to the left was actively inhibited. This inhibition, although present even in the normal state, is not apparent because it is overcome by facilitatory influences from the cerebral regions damaged in hemineglect patients.
We might suppose that a second operation (contralateral frontal?) could restore the hemineglect patients' attention to the previously neglected half of body or extrapersonal space. It is unlikely that this will soon be tried. However, an analogous "experiment of nature" was recently reported by Vuilleumier et al. (23). Their right-handed patient had the sudden onset of a right parieto-occipital infarct. The patient's left hemineglect "remained unvaryingly severe" for 10 days, until an angiogram resulted in a left frontal infarct with abrupt disappearance of all signs of left hemineglect.
Once we understand that inaccessibility of memories consequent to cortical lesions can reflect imbalance rather than loss, we can better appreciate the idea, put forward by Sherrington (18) in his 1932 Nobel lecture, that recovery from central nervous system deficits can often be attributable to subsidence of inhibition.
Concluding Remarks
By "memory," we usually mean that an individual's behavior in a particular context reflects previous experience with some aspect of that context. The many ways in which this occurs depend on a variety of different mechanisms. Studies distinguishing among these progressively emphasize more on anatomy than on introspection.
A hemispherectomized person (assuming a good residual hemisphere, as in the case of Smith and Sugar [191) possesses the full panoply of memory types and their underlying mechanisms Hence, the anatomically intact individual has all of these in duplicate, plus the mnestic function of the cerebral commissures. This suggests the likely presence of multiple (probably bilateral) lesions in the event of a major memory deficit. Simultaneously, it suggests that partial deficits can be overlooked in the absence of expert testing. As increased sophistication enables us to be more specific in the description of memory loss, correlation with the underlying neuronal systems can also be expected to progress. For expert evaluation of our patients' memory problems, we need the help of memory specialists. Our own increasing sophistication will enable us to work better with both the specialists themselves and their data.

REFERENCES

1. Anderson MC, Bjork RA. Mechanisms of inhibition in long-term memory: a new taxonomy. In: Dagenbach D, Carr TH, eds. Inhibitory processes in attention memory and language San Diego: Academic Press, 1994:265-326.
2. Bechara A, Tranel D, Damasio H, et al. Double dissociation of conditioning and declarative knowledge relative to the amygdala and hippocampus in humans. Science 1995;269: 1115-1118.
3. Bisiach E, Luzzatti C. Unilateral neglect of representational space. Cortex 1978;14:129-133,
4. Bogen JE, Campbell B. Recovery of foreleg placing after ipsilateral frontal lobectomy in the hemicerebrectomized cat. Science 1962;135:309-310.
5. Bogen JE, Suzuki M, Campbell B. Paw contact placing in the hypothalamic cat given caffeine. Neurobiology 1975;6: 125-127.
6. Cirelli C, Pompeiano M, Tononi G. Neuronal gene expression in the waking state: a role for the locus coeruleus. Science 1996;274:1211-1215.
6a. Seville WE, Milner B. Loss of recent memory after bilateral hippocampal lesions. J Neurol Neurosurg Psychiatry 1957;20: 11-21.
7. Corkin S. Lasting consequences of bilateral medial temporal lobectomy: clinical course and experimental findings in H.M. Semin Neurol 1984;4:249-259.
8. FusterJM. Memory in the cerebral cortex. Cambridge: MIT Press, 1995.
9. Graff-Radford N, Trane D, Van Hoesen GW et al. Diencephalic amnesia Brain 1990;113:1-25.
10. Knowlton B, MangelsJA, Squire LR. A neostriatal habit learning system in humans. Science 1996;273:1399-1402.
11. Markowitsch HJ, Yves von Cramon D, Schuri U. Mnestic performance profile of a bilateral diencephalic infarct with preserved intelligence and severe amnesic disturbances. I Clin Exp Neuropsychol 1993;15:627-652.
12. Milner B. Interhemispheric differences in the localization of psychological processes in man. Hr Metl Rtrll 1971;27: 272-277.
13. Milner B. Complementary functional specializations of the human cerebral hemispheres. In: Levi-Montalcini K, ed. Nerve cells, transmitters and behavior. Rome: Pontifica Academia Sci entarium, 1980.
14. Nielsen JM. Agnosia, apraxia aphnsia: their value in cerebral localization. 2nd ed. New York: Paul B. Hoeber, 1946.
15. Paivio A. Mental representations: a dual coding approach New York: Oxford University Press, 1990.
16. Petri H, Mishkin M. Behaviorism, cognitivism and neuropsychology of memory. Am Scientist 1994;82:30-37.
17. Raymond JL, Lisberger SG, Mauk MD. The cerebellum: a neuronal learning machine? Science 1996;272:1126-1131.
18. Sherrington CS. Inhibition as a coordinative factor. In: Nobrl lectures in physiology or medicine 1 922- 1 94 1. Amsterdam: Elsevier, 1965.
19. Smith A, Sugar O. Development of above-normal language and intelligence 21 years after left hemispherectomy. Neurology) 1975;25:813-818.
20. Sprague JM. Interaction of cortex and superior colliculus in mediation of visually guided behavior in the cat. Science 1966; 188 :: 1544-1547.
21. Thompson RF, Logan CG. Memory and learning. In: Greger P Windhorst V, eds. Comprehensive human physiology. Berlin: Springer-Verlag 1996.
22. Ungerleider LG. Functional brain imaging studies of cortical mechanisms for memory. Science 1995;270:769-775.
23. Vuilleumier P, Hester D, Assal G, et al. Unilateral spatial neglect recovery after sequential strokes. Neurology 1996;19: 184-189.
24. Wallace SF, Rosenquist AC, Sprague JM. Ibotenic acid lesions of the lateral substantia nigra restore visual orientation behavior in the hemianopic cat. I Comp Neurol 1990;296: 222-252.
25. Zaidel DW. Memory and spatial cognition following commissurotomy. Handbook Neuropsychol 1990;4:151-166.
26. Zola-Morgan S, Squire LR. Neuroanatomy of memory. Annual Rev Neurosci 1993;16:547-563.