WHEN THE PAST DISAPPEARS

[excerpt from Schacter's Searching for Memory]

Retrograde Amnesia and the Structure of Memory

We have already seen that patients with damage restricted to the medial temporal lobes have retrograde amnesias that obey Ribot's Law: such patients can remember many experiences from the distant past but flow from the recent past, just prior to their brain injuries. But when brain damage extends outside the medial temporal region, into areas of the cortex where engrams are actually stored or regions that are essential to retrieving memories, we see much more extensive retrograde amnesias that sometimes blanket nearly all of a patient's personal past. These retrograde amnesias underscore that our normally seamless awareness of episodes from our personal pasts and knowledge of the world masks a great deal of underlying complexity. Remembering one's wedding depends on a different brain network than knowing where a bar of soap can be found. Like recognizing a submarine requires different neural machinery than recognizing a spider. Each of these types of knowledge depends on the integrity of specific constellations of underlying brain structures and processes.

Gene helped teach me some of these lessons. He developed amnesia when he sustained a serious head injury during a 1981 motorcycle accident that damaged large sectors of his frontal and temporal lobes, including his left hippocampus.23 Thirty years old at the time of the accident, Gene, like Frederick, is now unable to recall day-to-day experiences except for a few isolated new facts (see chapter 4). In addition to his anterograde amnesia, however, Gene also shows a remarkable form of retrograde amnesia. Unlike amnesic patients who obey Ribot's Law, Gene is unable to recall a single specific episode from any time in his life.

Asking Gene about his personal past is an almost unnerving experience. He is a quiet, polite, and affable young man who always tries his best to come up with answers to questions that are posed to him. But no amount of prompting or cueing helps Gene recall specific past events, whether happy or sad, at school or at work, or including family or friends. Even when detailed descriptions of dramatic events in his life are given to him--the tragic drowning of his brother, the derailment, near his house, of a train carrying lethal chemicals that required 240,000 people to evacuate their homes for a week—Gene does not generate any episodic memories. Though he had been an avid motorcyclist prior to his injury, Gene no longer remembers any of the numerous trips he had made with his cycling buddies. Nor can he recall the frequent visits to bars he used to make with his friends. Whereas other amnesic patients usually can dredge up some episodes from the distant past, Gene remembers absolutely nothing. He looks at me with a puzzled expression, as if he understood that he should be able to provide a response to my questions. Gene appreciates that people are generally able to recall specific incidents from their pasts. Sitting quietly trying to come up with an episodic memory, he is apt to emit a nervous laugh--a sort of recognition that it is strange, almost silly, that he cannot come up with anything. Then there is usually a sigh of resignation as Gene acknowledges that nothing is going to come to him. Within a couple of minutes, this incident, too, vanishes into the black hole of his episodic memory.

A life without any episodic memory is psychologically barren—the mental equivalent of a bleak Siberian landscape. Nothing much hap-pens in Gene's mind or in his life. He has few friends and lives quietly at home with his parents. He performs many of the same routine activities again and again.

[excerpt from Clinical Disorders of Memory] 

While patients with penetrating head injury fall within a relatively homogeneous group, other head-injury patients represent a more heterogeneous population. The term 'blunt head injury' will be used to describe this broad category of patients, and I will be including within this category studies of patients with 'closed' head injury. While the cerebral pathology resulting from blunt head injury entails a significant degree of generalized brain damage, focal cerebral lesions may also be evident after this type of injury. This may result from localized skull fracture or from the development of a focal intracerebral or extracerebral haematoma. In addition, however, areas of focal cerebral contusion are frequently found in temporal lobe and basal-frontal regions after blunt head injury (Figure 5.1).

Pre-traumatic and post-traumatic amnesia

Memory for events around the time of the injury accompanies cases of blunt head injury where significant cerebral concussion has taken place. In most instances, loss of consciousness accompanies the injury, but in a few cases the patient may not lose consciousness yet will subsequently display both pre- and post-traumatic amnesia (Fisher, 1966: Haas and Ross, 1986) The mechanisms of this amnesia arc unclear, but it is possible that there is some overlap with phenomenon of transient global amnesia and that the head injury acts as a trigger to initiate a separate pathological process.

The following sections will consider memory loss for events immediately preceding and subsequent to the head injury, and also memory functioning during the early post-traumatic period.

Pre-traumatic amnesia

Pre-traumatic amnesia has traditionally been defined as memory loss for events which immediately preceded the head injury. The estimate of pre-traumatic amnesia is usually obtained months or years after the injury and is often difficult to obtain with precision, especially when the patient's daily schedule prior to the injury was fairly routine. It is also unclear whether the memory loss should be absolute or whether partial memory for events may also be subsumed under the period of pre-traumatic amnesia. Nor is it generally agreed whether such amnesia should include more general memory loss which may be present for events, months and years prior to the injury, but which may be discontinuous with a period of pre-traumatic amnesia. Pre-traumatic amnesia is frequently short in duration, and can usually be counted in minutes or hours rather than days or weeks (Eden and Turner, 1941). In many cases of mild blunt head injury, pre-traumatic amnesia may be absent, leaving an isolated period of post-traumatic amnesia (Crovitz, Horn and Daniel, 1983). Russell (1935) suggested that residual pre-traumatic amnesia seldom lasts more than a week, and that instances where the duration was in years probably had a hysterical basis. He also noted that in a few cases pre-traumatic events might be remembered in the early post-traumatic period but be lost from memory after the period of post-traumatic amnesia. Symonds (1962) described two patients with blunt head injury, one of whom had a pre-traumatic anmesia of 3 months and the other of one year, but he did not believe there was a hysterical basis for the amnesia in either of these patients.

Wasterlain (1971) suggested that patients with long periods of pretraumatic amnesia (which he defined as more than 10 minutes) tended to have more severe head injuries with evidence of considerable cerebral contusion. Russell and Nathan (1946) assessed pre-traumatic and post-traumatic amnesia in a large series of patients, including cases of gunshot wounds and cases of 'accidental head injury' (mostly blunt head injury). In the latter group of patients, pre-traumatic amnesia lasted less than 30 minutes in over 80% of the series, and was also found to be less than 30 minutes in about 50% of the patients who had post-traumatic amnesia of up to 7 days. A small number of patients 111/20111 had no pre-traumatic amnesia despite a post-traumatic amnesia of up to 7 days. There did not therefore appear to be much of a correlation between the duration of the pre-traumatic amnesia and that of post-traumatic amnesia, however, Blomert and Sisler (1974) examined patients shortly after a mild head injury and found a stronger correlation between duration of pre-traumatic and post-traumatic amnesia. When post-traumatic amnesia was less than one hour, pre-traumatic amnesia was always under one minute.

On the basis of their observations of head-injured patients Williams and Zangwill ( 1952) hypothesized that two types of pre-traumatic amnesia may exist: (1)A short and usually complete retrograde amnesia; (2) More diffuse and widely distributed disturbances of memory for pre-traumatic events'. The suggestion that a short pre-traumatic amnesia and a more general retrograde amnesia may be dissociable and rely on different neural mechanisms has not been systematically followed tip by subsequent research workers, even though it has been alluded to in other studies of patients with blunt head injury (Benson and Geschwind, 1967)

Some recent evidence has, however, been gathered to support the presence of a more general retrograde memory loss in head injury patients. Using a questionnaire relating to television programmes which had been broadcast in the years prior to the injury, Levin et al. (Levin, Grossman and Kelly, 1977; Levin, Papanicolaou and Eisenberg, 1984) found impaired memory during the post-traumatic period for programmes which had taken place over a period of around 12 years prior to the injury, and did not find any evidence of a temporal gradient in the memory loss. In a subsequent report (Levin et al., 1985), a similar impairment was also observed in patients who were assessed after the period of post-traumatic amnesia. The remote memory loss shown by patients during post-traumatic amnesia, was also evident for autobiographical events (e.g. name of primary/secondary school attended). However, when autobiographical memory was sampled, they did find better recall of earlier compared to more recent events by patients tested during the post-traumatic period. It is worth pointing out that the questionnaire method employed in the earlier studies (Levin, Grossman and Kelly, 1977; Levin, Papanicolaou and Eisenberg, 1984) used a recognition format and it is possible that recall procedures would have produced a temporal gradient in the retrograde amnesia. Further, head-injury patients tend to be relatively young subjects, and so time periods which can be validly sampled may not stretch back far enough for any gradient effect to be manifest. Crovitz (19861 has also outlined, in the form of a detailed single-case interview, the use of cue words to stimulate recall of autobiographical information by a patient who suffered a severe blunt head injury. A further manifestation of retrograde memory deficits after blunt head injury may be in the form of an inability to recognize faces which were familiar to the patient prior to his head injury. Levin and Peters 119761 have reported one such case, although this patient showed significant cognitive impairment in addition to his deficit in faces recognition.

Russell (1935) first reported the condition of 'shrinking' pre-traumatic amnesia find described a patient with a closed head injury who had fin initial pre-traumatic amnesia of at least 5 years which, over a period of 111 weeks, eventually shrank to a matter of a few minutes. Russell and Nathan 119461 confirmed this pattern and found that the recovery of pre-traumatic amnesia was in chronological order, with items in the distant past recovering first. Williams and Zangwill (1952) pointed to some degree of variability in the chronological sequence of remembered events, find indicated that cueing might often help patients to recover lost memories. Although shrinkage of pre-traumatic amnesia usually parallels the resolution of post-traumatic amnesia, the latter may terminate well before pre-traumatic amnesia shrinks to its final duration (Russell, 1971), and significant pre-traumatic amnesia may remain in a few isolated cases (Goldberg et al., 1981).

The converse to shrinking pre-traumatic amnesia has been noted in football players with minor concussion who can recall pre-traumatic information immediately after the injury but have amnesia for the same information if questioned minutes later (Lynch and Yarncll, 1973). The phenomenon of shrinkage of pre-traumatic amnesia has led some investigators (e.g. Benson and Geschwind, 1967) to suggest a retrieval deficit as the major mechanism underlying this form of temporary memory loss, although other explanations such as interference with normal rehearsal processes (Williams, 1969) may also be feasible.

Post-traumatic amnesia

Post-traumatic amnesia refers to the period subsequent to the head injury for which the patient has loss of memory. Russell (1932) alluded to some of the difficulties in equating loss of consciousness with loss of memory, but it was not until several years later that Symonds and Russell (1943) specifically defined the duration of post-traumatic amnesia as ending 'at the time from which the patient can give a clear and consecutive account of what was happening around him'. They emphasized the dangers of underestimating the duration of post-traumatic amnesia by using 'ishmds of memory' to indicate the end of the amnesia, or by assuming that if the patient is aware of things going on around him, he may necessarily be able to recall these at a later date. A number of authors (e.g. Whitty and Zangwill, 1977) have noted that a 'lucid interval' may be present early in the period of post-traumatic amnesia where there has been minimal initial concussion. This may be due to a highly emotional event which has remained imprinted in the patient's memory, or to a period preceding the delayed development of vascular lesions such as an extradural haematoma, which may subsequently result in a period of amnesia due to cerebral compression.

The duration of post-traumatic amnesia will vary with severity of the head injury, and therefore 'typical' durations are somewhat misleading. Russell and Nathan (1946) noted that in a series of over a thousand 'accidental' head injuries, 70% of cases had post-traumatic amnesia lasting between one hour and 7 days. Even in patients with a post-traumatic amnesia lasting more than one week, as in the series reported by Roberts (1979), a relatively wide distribution of duration of post-traumatic amnesia was evident, with one-third of patients having a post-traumatic amnesia duration of 28 days or more. Russell and Smith (1961) observed that the distribution o1' length of post-traumatic amnesia in older patients was skewed towards prolonged periods in this age group. As Levin, Benton and Grossman (1~82) have pointed out, the total duration of post-traumatic amnesia will include some period of coma and a further period of impaired memory. The relative lengths of these may be variable, and in some patients such as the elderly head injured, a short period of coma may be followed by a relatively protracted period of poor memory, resulting in a long overall duration of post-traumatic amnesia (Von Wowern, 1966).

Dissatisfaction with retrospective estimates of post-traumatic amnesia has led investigators to seek more accurate observations of memory functioning during this period and to obtain concurrent measures of post-traumatic amnesia. In a longitudinal study of patients with mild head injury, Gronwall and Wrightson (1980) holed a dissociation between recovery of orientation (mainly orientation for time and place) and recovery from post-traumatic amnesia, and therefore disputed the functional equivalence of disorientation and post-traumatic amnesia. Similar findings were reported by Sisler and Penner (1975) in patients with more severe head injury. Somewhat different conclusions were reached by Fortuny et al. (19~;0) in their study of post-traumatic amnesia in head-injured patients. They supplemented questions regarding orientation for time and place with a picture recall task, and also tested patients' memory for the examiner's name. They defined the termination of post-traumatic amnesia as the point when the patient had three successive days of correct recall on such items, and found that this brief test yielded estimates of post-traumatic amnesia that correlated with retrospective estimates made by neurosurgeons.

Levin, O'Donnell and Grossman (1979) designed a test of orientation specifically for the study of head-injured patients - the Galveston Orientation and Amnesia Test (GOAT). It included orientation for time and place, and the patient's memory for events around the time of the injury, mode of transport to hospital, etc. Using this instrument to monitor post-traumatic amnesia and its termination, they found that in a series of 52 patients with significant closed head injury there was a median post-traumatic amnesia of 14 days. They found a correlation between duration of post-traumatic amnesia and ratings on components of the Glasgow Coma Scale. Post-traumatic amnesia which persisted beyond 2 weeks was associated with CT scan findings indicative of diffuse injury or bilateral lesions, whereas GOAT scores which improved more rapidly tended to be found in patients with focal mass lesions. In terms of correlation with the Glasgow Outcome Scale (Jennett and Bond, 1975), patients with post-traumatic amnesia of less than 2 weeks achieved good recovery, while longer intervals were associated with prolonged disability. They found that their instrument was particularly useful in patients with very brief periods of coma (one or 2 hours) which were then followed by prolonged confusional states that may have included periods of fluctuating memory functioning. Levin, Papanicolaou and Eisenberg (1984) reported some further data relating duration of coma to duration of post-traumatic amnesia defined by GOAT scores, and they found that there was a relatively weak relationship between the two, with some instances of prolonged post4raumatic amnesia following relatively brief periods of coma. They suggested that the factors which determine the duration of post-traumatic amnesia may be quite different from those variables which influence the duration of coma. Apart from considering memory functioning after blunt head injury as a means of defining the period of post-traumatic amnesia, other investigators have examined memory performance in its own right. Cronholm and Jonsson (1957) reported impaired memory scores on the Benton Visual Retention Test in patients examined within the first 24 days of a closed head injury, the severity of which was not indicated by the authors but stated to have 'varied considerably' between patients. Dunn and Brooks (1974) observed that patients' verbal recall scores during the period of post-traumatic amnesia could be improved with semantic cues, and suggested that difficulties in retrieval may characterize memory problem~ in the post-concussion phase.

In addition to general impairment in memory functioning in the post-traumatic period, confabulation and other forms of paramnesia may be present. As was pointed out earlier (see p. 20), there are difficulties in interpreting reports of confabulation due to the lack of standardized procedures for assessing confabulatory responses and the practical difficulties in ascertaining the extent to which patients' responses are based on real or imaginary past experiences. Whitty and Zangwill (1977) indicated that a mild form of paramnesia or confabulation may exist in most instances of post-traumatic amnesia, but that occasionally an instance of marked confabulation may occur in the presence of relatively normal behavior in other speheres of activity, as opposed to confabulation which may be present as part of a 'general traumatic confusion.' In the former cases, there is usually subsequent evidence of prolonged post-traumatic and pre-traumatic amnesia and other neurological signs of brain injury. A focal confabulator syndrome was noted to have been present in only 38 out of 1931 cases of blunt head injury observed during the Second World War. Six of these showed confabulation persisting for more than 2 or 3 days, with the content of the confabulation mostly comprising misrecall of real experiences, e.g., events placed in the wrong temporal/spatial context. There is a rather facile euphoria in most of these patients which tended to disappear as recovery continued.

In a study which included both patients with blunt head injury and those with penetrating head injury (gun-shot wounds), Weinstein and Lyerly (1968), found confabulation in 60% of patients. This was defined as invention of a story about a past event or gross distortion of an actual occurrence, and was elicited by direct questions about the patients' disability and the circumstances surrounding their admission to hospital. While florid, spontaneous confabulation is invariably restricted to the early stages of the post-traumatic period, in a few instances a patient examined several months after injury may confabulate in response to specific questions probing his memory for past events; these patients, such as the one described by Baddcley and Wilsoh (1986), usually have evidence of significant frontal lobe pathology as a result of their head injury.

The condition known as 'reduplicative paramnesia', which was earlier discussed in Chapter I, has also been observed in head-injured patients. Paterson and Zangwill (1944) reported spatial disorientation, which included features of 'reduplicative .paramnesia' in two patients who incurred severe head injury. The patients disorientation for place, which lasted 1-2 weeks, included confusing their present location with a place in which they had previously been. Benson, Gardner anti Meadows (1976) reported an even longer duration (several months) of reduplicative paramnesia in their three patients with severe closed head injury. These patients consistently mislocated the hospital to another site which had been significant to them at an earlier stage in their lives, even after memory for everyday events seemed to have returned to normal. Benson, Gardner and Meadows implicated right hemisphere and frontal pathology in mediating the form of spatial disorientation present in their patients. Staton, Brumback and Wilson (1982) reported a case who showed reduplicative amnesia which included features of the 'Capgras syndronlc', following a severe head injury which resulted in a number of cerebra lesions, including right hippocampal and temporal lobe damage. As part of this syndrome, their patient failed to recognize familiar people even family members, and also believed that the real person had been replaced by a double. He also had a number of deja vu experiences.

Anterograde memory deficits

Memory symptoms and impaired memory test performance are among the most common sequelae of blunt head injury, even in those patients who otherwise appear to have made a good recovery from their injury (Stuss et al., 1985). In view of the most frequent sites of cerebral contusion in blunt head injury (see Figure 5.1), namely the temporal and basal-frontal regions, it is not surprising that memory impairment is the most marked and persistent impairment after blunt head injuries. Everyday memory symptoms which are present several years after a severe head injury may often be more reliably reported by an observer than by the patient himself (Sunderhmd, Harris and Gleave, 1984), perhaps due in part to the head-injured patient sometimes forgetting instances of his own memory lapses. Memory symptoms described by patients or their close family members arc in general similar to 'normal' memory symptoms, ranging from difficulties in remembering people's names to forgetting where articles have been put. in general, there is only a slight correlation between memory symptoms and memory test performance (Kaput and Peason, 1983), and where a correlation has been found it has been limited to certain memory tasks such as story recall, and to patients who are assessed several years after the head injury (Sunderland, Harris and Baddeley 1983). Russell (1932) and Russell and Smith (1961) pointed to clinical evidence for the presence of memory deficits after the initial period of traumatic amnesia, but did not report any detailed memory testing. Recently there has been a mushrooming of investigations of memory function after blunt head injury, with a large number of these emanating from Brooks and his colleagues in Glasgow. In a study of patients who were mostly seen within 2 years of injury, Brooks (1976) found normal or near normal performance on digit span but impaired recall on reverse span, impairment in immediate and delayed recall of a short story, and in paired-associate learning. Thomsen (1977) assessed verbal learning within 3 years of injury in patients with severe closed head injury, 19 of whom were still dysphasic. Both dysphasic and non-dysphasic patients were worse than controls on learning a complete sentence, learning 10 unrelated words, or recalling a short story; on most measures, patients with dysphasia were worse than patients without. Brooks (1975) showed that in word-list recall it was the long-term memory component of the task which was most impaired in head-injured patients. Lower levels of clustering and subjective organization were noted to be present in the verbal free recall of patients with blunt head injury who were examined during a course of rehabilitation after severe head injury (Levin and Goldstein, 1986). Fodor (1972) reported that in her patients, who were studied shortly after their head injury, delayed recall of related material was one of the more sensitive measures of their memory deficit. However, this study has been criticized for confounding measures of premorbid and current intellectual level (Schacter and Crovitz, 1977), and it also suffered from floor effects for delayed recall of unrelated material.

Brooks (1974b) examined continuous recognition memory performance with non-verbal stimuli in patients seen one to 32 months after injury. Interpreting his data in terms of signal detection theory, he suggested that head-injured patients were significantly more unwilling to guess than control subjects on the test. However, Richardson (1979a) disputed Brooks' use of signal detection theory and reinterpreted his data as indicating no change in response bias after head injury. Hannay, Levin and Grossman (1979) found that the number of false positive responses by head-injured patients on a similar continuous recognition memory task was greater than that by control subjects, but that a combined measure of hits and false positives, i.e. the total number of correct responses, was the factor which best discriminated head-injured from control subjects. Using the selective reminding technique (Buschke and Fuld, 1974), Levin et al. (1979) assessed word-list recall of severely head injured patients seen at least 6 months after injury. They found that memory test scores showed a fairly close relationship with severity of disability, as measured by the Glasgow Outcome Scale. Brooks (1984) pointed out that this sample of patients was not random, since it was restricted to the ages of 16 50 years and excluded patients with a history of alcoholism, drug abuse or neuropsychiatric disorders.

Patients with relatively mild head injury (usually with post-traumatic amnesia between a few minutes and a few days) have also been found to show impaired memory functioning (Gronwall and Wrighlson, Barth et al., 1983), although this evidence is subject to some qualifications. In general, such deficits appear to have been limited to the first few months after injury, and one study (McLean et al., 1983) only found significant memory impairment if this was tested a few days after injury. Another investigation (Gentilini et al., 1985) did not find any evidence of memory impairment m patients with mild head injury who were assessed one month after injury. Richardson (1979b) found a deficit in recall of concrete but not of abstract words after mild head injury, and subsequently obtained similar findings in a group of 30 patients, 22 of whom had incurred a mild blunt head injury and eight of whom had a moderately severe blunt head injury (Richardson and Shape, 1984). A similar pattern of memory deficits was observed by Richardson and Barry (1985) in a group of patients with mild blunt head injury who, nonetheless, performed normally on tests of memory for faces and memory for pictures. These researchers also observed that if blunt head injury and control patients were instructed to use visual imagery to remember the words, the former no longer showed a deficit relative to control patients. Richardson (1984) also observed that in patients with blunt head injury the pattern of intrusion errors during recall of a word-list was unrelated m the concreteness of the material in the word-list, whereas such a relationship was readily apparent in the performance of control subjects. It is important to point out that in all of these studies by Richardson, patients were examined within the first few days after their head injury, and so such deficits should not necessarily be regarded as reflecting permanent sequelae of a minor blunt head injury.

Boxers represent a unique subset of the poputalion of patients with mild blunt number of minor head injuries rather than a single episode of concussion (Roberts, 1969). In considering this group of patients, it is important to make allowance for any history of alcohol abuse and also any limitations in the subject's premorbid intellectual level. Thomassen et al. (1979) found evidence for a deficit in a word-list learning task in a group of former boxers, but this deficit was no longer evident when allowance was made for educational and vocabulary level of the boxers. More recently, McLatchie et al. (1987), in a more detailed investigation of memory functioning, found that a subgroup of active boxers (7 out of 16 for whom data were available) showed impaired memory functioning. In summary, there appears to he no doubt that significant impairment off a range of memory tasks, but especially those with a long-term component, can occur as a sequel of blunt head injury. What is of greater interest, however, is the extent to which residual memory deficits can be predicted on the basis of particular features of the head injury or the head-injured patient. This will he examined in the following sections.

Features of the comatose period

Interpretation of studies reporting correlations between the comatose state of patients after head injury and subsequent memory performance, need to take account of the difficulties in obtaining valid and reliable measures of head-injured patients' cerebral function while they are in a comatose state, especially where such measures have been made on a retrospective basis. Such difficulties have to some extent been alleviated with the introduction of scales such as the Glasgow Coma Scale (Teasdale and Jennett, 1974). Levin, Grossman and Kelly (1976) and Hannav, Levin and Grossman (1979) found a relationship between coma duration and short-term recognition memory lot visual patterns, but this relationship was somewhat variable - patients with severe head injury performed worse than those with mild or moderate head injury in the study by Levin et al., whereas in the investigations by Hannay et al., patients with moderate and severe head injury did not differ in memory performance, although both groups were more impaired than a group with mild head injury. Lezak (1979) did find some relationship between coma duration on auditory verbal learning; patients with coma of longer than 2 weeks were much more likely than those with coma of shorter duration to show impaired memory 3 years after injury. Brooks et al. (1980), using the Glasgow Coma Scale, did not find any significant relationship between coma duration and performance on a range of memory tasks. Levin et al. (1979) reported a close relationship between features of the comatose state, namely, presence of oculomotor deficit and impaired motor functioning in the early stages of recovery,and impaired performance on both storage and retrieval measures of word list learning. Dye, Milby and Saxon (1979) also found poor tactile memory performance in patients with coma and additional motor and ocular signs, compared to those with less severe neurological deficit after injury. In general, therefore, the relationship between coma duration and residual memory functioning has been shown to be relatively inconsistent, although it appears that some features of the comatose patient, e.g. particular neurological signs, may be predictive of future memory performance.

Features of the post-traumatic period

In his study of continuous recognition memory, Brooks (1974b) found that the duration of post-traumatic amnesia showed a significant negative correlation with efficiency of recognition memory, but in a later study (Brooks, 1975), he found that the post-traumatic amnesia showed a weak relationship with short-term and long-term memory for word list material. In a subsequent larger-scale study of 82 patients, Brooks (1976) found that performance on the story recall and paired-associate learning subtests of the Wechsler Memory Scale was significantly related to the duration of postraumatic amnesia, and that patients with more than 7 days of amnesia performed significantly worse than controls. In another investigation by Brooks et al. (1980) using paired-associate learning, story recall and visual design recall (recall of the Rey-Osterreith figure), a further effect with regard to post-traumatic amnesia and memory/learning tasks was found, and this relationship was particularly marked in the case of delayed relearning of paired-associate items.

Smith (1974) did not find any significant relationship between post-traumatic amnesia and memory performance 10-20 years after injury, although her data suggest that patients with shorter post-traumatic amnesia, 21 days or less, performed somewhat better than those with longer amnesia on story recall. Van Zomeren and Van Den Berg (1985) noted that a cut-off point of post-traumatic amnesia of 13 days was most reliable in distinguishing those patients with blunt head injury with symptoms of everyday memory difficulty. Parker and Serrats (1976) used a classification of post-traumatic amnesia based on assessment of memory functioning during the post-injury period. This assessment comprised 'a simple question and answer test given daily', but no further details of the lest were provided. They estimated that 92% of patients with post-traumatic amnesia of 24 hours or less reached normal memory by 2 years, whereas only 50% of those with post traumatic amnesia longer than one month recovered normal memory functioning. Over shorter periods of post-traumatic amnesia (from a few seconds to a few days}, Gronwall and Wrightson (199l) found that durations of the amnesia which were less than one hour were similar to those for one to 24 hours in terms of their effects on Wechsler Memory Scale performance, but patients with post-traumatic amnesia of more than one day were impaired compared to those with a shorter duration. A further relationship between duration of post-traumatic amnesia and memory for word list material, more especially with regard to storage than retrieval measures of performance, was observed in a further group of patients whose duration of amnesia varied between several hours and several days.

Features of the cerebral injury

A number of studies have failed to find any significant effect of the presence of skull fracture on residual memory deficit (c.g. Klovc and Cleeland, 1972; Brooks, 1976; Brooks et al. 1980). Brooks (1975) did find an effects on one measure of memory performance - 20 seconds delayed recall of word list material - with greater impairment in patients with skull fracture. Smith (1974) found that the site of skull impact affected the degree of recovery of memory function. She found that patients with right-sided impact performed worse than those with left-sided impact on tests of story recall and recall of visual designs, but that on the later test patients with left-frontal impact performed particularly poorly. She interpreted some of her findings as reflecting contre-coup effects, but it is unclear why these should have appeared to occur for some impact sites and not for others. Brooks (1984) also disputed some interpretations of Smith's data and in his own study (Brooks et al., fracture or whether it was on the left or right side, although patients with operated haematomas performed better than non-operated patients, possibly reflecting more limited, focal damage in the operated patients. Left and right sided haematoma cases did not show any differences in pattern of memory performance.

In general, patients with focal intracranial haematomas tend to perform more poorly than patients with blunt head injury with normal CT scans or small ventricles (Henming, Orrison and Mikula, 1982). Patients with subdural haematomas, when compared to those without haematomas, have been found to perform worse on memory tests, and this may be related in part to the greater degree of cortical atrophy which accompanies such haematomas and to the frontal distribution of such lesions (Cullum and Bigler, 1985). Milner (1969) found that patients with left temporal epileptic foci after blunt head injury were particularly impaired on delayed recall of story and paired-associate material. Warrington and Shallice (1991) reported an isolated verbal memory deficit in a patient who had a left parietal subdural haematoma following a blunt head injury. Their patient showed significant short-term, mainly repetition, verbal memory deficits in the context of normal performance on long-term verbal memory tasks. Lezak (1979), using verbal memory tasks, noted that in the early recovery period, patients with greater left-hemisphere involvement tended to have lower test scores than other patients.

Brooks (1976) found that patients with persistent neurological signs did not score worse on the Wechsler Memory Scale. In the study by Levin et al. (1979) referred to earlier, patients showing signs of brainstem damage performed poorly on memory tasks, possibly reflecting greater severity of diffuse rather than focal brain (, 1,~) found that the presence of persisting damage. Klovc and Cleeland neurological signs was significantly associated with poor performance on a tactile memory task. Smith (1974) reported that patients with abnormal motor responses soon after injury (for example, decerebrate response), were more likely to show severe verbal learning deficits at follow-up, although this was only found in right-sided impact cases, but Brooks (1984) questioned whether the raw data in fact warranted such conclusions. Levin et al. (1979) reported that early hemiparesis, bilateral mm-reactive pupils, and more particularly, oculovestibular deficit were associated with poor memory storage and retrieval months or years after injury. Levin and Eisenberg (1979c) carried out a study (reported in Levin, Benton and Grossman, 1982) in which they classified head-injured patients into mild, diffuse, complicated by a lateralized mass lesion, or associated with a bilateral mass lesion. Patients were assessed after they had come out of post-traumatic amnesia or at least had reached a stable plateau of recovery. Thus, mildly injured patients were seen a shorter period (median 16 days) after injury than patients with more severe injury (median 42 days). They administered a word-list recall test and found significantly better performance in patients with mild blunt head injury and those with right hemisphere mass lesions, compared to patients with more diffuse injury. Patients with left hemisphere lesions, especially those involving the left temporal lobe, performed less well than other patients. Within the right hemisphere group, there was no evidence of more marked memory impairment with temporal lobe involvement, and Levin, Benton and Grossman (1982) also indicated that they did not think a measurable language deficit was a necessary or sufficient condition for impaired performance on the word-list recall task. Jetter et al. (1986) presented evidence implicating frontal lobe damage after blunt head injury in impaired delayed (24 hours) recall of verbal material, but interpretation of their study is complicated by the presence of some non-traumatic aetiologies in their sample, and by the presence of bilateral pathology in some patients with frontal lesions, but not in any of the patients with lesions outside the frontal area.

Evidence of more diffuse cerebral pathology in blunt head injury is more difficult to quantify, but some attempts have been made to relate severity of memory impairment with measures of ventricular dilatation, which may often follow cases of blunt head injury. Levin et al. (1981) found a relationship between degree of ventricular dilatation and performance on tests of verbal recall and picture recognition memory. Bigler et al. (1984) also reported a significant correlation between memory performance and degree of veutricular dilatation in patients with blunt head injury, and later studies (Cullum and Bigler, 1986; Massman et al., 1986) showed a relationship between severity of memory impairment, and measures of both cortical atrophy and ventricular dilatation. Verbal paired- associate learning showed the strongest correlation with these indices of cerebral injury and, in the case of cortical atrophy, it appeared that left frontal cortical atrophy accounted for much of the relationship with severity of memory impairment.

A single-case study of a patient with residual left frontal and right temporal pathology after a severe blunt head injury (Zatorre and McEntee, 1983) yielded evidence of a marked, relatively pure, amnesic syndrome. Some of the features of this syndrome resembled those found in studies of patients with alcoholic Korsakoff's syndrome (e.g. evidence of impaired semantic encoding of verbal information during verbal memory performance). Bauer (1~2) has also reported a case of bilateral cerebral lesions following a severe blunt head injury, although in this patient the intracerebral haematomas, which were sited in both posterior temporal regions, developed after a minor surgical procedure carried out 6 days after injury, and so it is not possible to relate the pathology specifically to the effects of the head injury. The patient showed marked memory impairment, which included inability to recognize familiar faces, together with topographical memory loss and impaired memory test performance, more especially for visual, non-verbal material. These memory deficits were still apparent about 18 mouths after the injury.

Distinctive features of memory disorders in blunt head injury

While a large amount of data has been collected on memory functioning in patients with blunt head injury, few direct comparisons have been made with memory disorders in other neurological conditions. Although many other conditions tend to affect an older age group than head-injured patients, some illnesses, such as subarachnoid haemorrhage, overlap with blunt head injury in the population which is affected. This section will, therefore, include more general observations on indirect comparisons which can be made on the basis of existing evidence in the literature. In the case of traumatic amnesia, i.e. loss of memory for events around the time of the injury, some comparisons can be made between such memory loss in blunt head injury and analogous observations which have been made in the case of penetrating head injury. These have been summarized in the chapter dealing with penetrating head injury (see p. 93), and will not be repeated here. Schacter and Crovil7 (lq77) pointed to the desirability of comparing the memory loss in traumatic amnesia with that seen in conditions such as transient global amnesia, which usually have a cerebrovascular basis. While both conditions share features such as marked memory impairment with some degree of retrograde amnesia during the post-ictal period, there are a few distinguishing features of memory functioning. First, the most striking difference is the alert state of the patient with transient global amnesia - the patient is usually able to maintain attention and carry out skills which do not tax his memory. In contrast, the patient with blunt head injury is invariably rendered unconscious by the cerebral injury. Second, in traumatic amnesia, marked anterograde memory impairment may extend for a longer period (up to several months) than the maximum length of memory difficulty seen in transient global amnesia, which usually does not exceed a few hours. Third, long-term recovery of memory functioning is at a much higher level in transient global amnesia than blunt head injury. Fourth, paramnesic phenomena, such as florid confabulation, tend to be more common in the post-traumatic period after blunt head injury than in the post-ictal period during an episode of transient global amnesia. Fifth, as Fisher (1982a) has pointed out, the duration of permanent pre-ictal amnesia tends to be somewhat longer in transient global amnesia compared with blunt head injury.

Turning to residual memory deficits after blunt head injury, few studies have formally compared such deficits with those found in other aetiologies. Exceptions to this include the investigation by Corkin et al. (1985) in which they compared a small group (n = 5) of selected patients with blunt head injuries with those of other aetiologics. Their patients were selected such that they might display a relatively pure. global amnesic syndrome. On verbal and non-verbal recurrent recognition memory tasks, the patients with blunt head injury performed worse than those with subarachnoid haemorrage or with herpes simplex encephalitis. Other memory tests (e.g. delayed recall of verbal and non-verbal material, paired-associated learning) did not reveal any significant differences in performance relative to other patient groups. Massman et al. (1986) found that in patients with blunt head injury, severity of memory impairment showed a relationship with degree of structural brain damage (as indicated by measures of cortical atrophy and ventricular dilatation), but that such a relationship was absent in a group of patients with primary degenerative dementia.

E. MEMORY DISORDERS FROM HEAD TRAUMA

[excerpt from Memory Disorders in Clinical Practice]

Memory disorders in the survivors of head injury are only one aspect of the enormous disability that many of these patients suffer. Posttraumatic changes in personality, social relations, and in emotional adjustment influence the patient's motivation and performance on memory tests.

Memory disorders in head injury are discussed under the following main headings: (1) posttraumatic amnesia; (2) retrograde amnesia; and (3) residual memory impairment (recovery of memory).

Posttraumatic Amnesia

A slight blow on the head may cause a momentary loss of consciousness. A severe head injury may render a person unconscious for hours, days, or weeks before recovery occurs. Emergence from the state of unconsciousness or coma is generally taken to be marked by obeying spoken commands and the return of speech or an equivalent signal from the patient. A period of disorientation and confusion follows. During this period, the patient may be docile, aggressive, talkative, shouting, or delusional. The patient is usually able to retain memories of certain small events occurring during this period, such as a visit from a friend or relative. These are only islands of memory without the context of time or sequence of events. The period of confusion may last three to four times longer than the duration of unconsciousness or coma. The end of the disorientation period generally occurs gradually, but sometimes suddenly, and is marked by the recovery of continuous memory for day-to-day events. The time interval between the injury and recovery of continuous memory is called the period of posttraumatic amnesia (PTA). Thus PTA includes the period of coma and period of disorientation. These two periods are not clearly distinct and gradually merge into one another. At the termination of PTA, the patient should be well oriented and be able to recall most of the daily events and answer questions intelligently. This criteria, however, would apply to a patient with good recovery. A patient with severe disability may not show a clear-cut termination of PTA.

Types of Memory Disturbances during PTA

There is no recall of events from the period of coma. The following disturbances of memory occur during the period of disorientation.

Islands of Memory.

Memory of special events such as a visit from a friend, an operative procedure, or transfer to a different room of the hospital may be retained. This is likely to occur when the patient is less confused and is able to converse sensibly for short periods of time. The memories are best characterized as "islands" in the midst of a "sea of forgetfulness." These islands of memory lack a temporal sequence and coherence. They do not necessarily imply an early termination of PTA. The islands of memory may go on occurring for several days before the termination of PTA. Russel (1971) noted that in 13 cases showing an early island of memory during PTA, the final duration of PTA was 1 to 24 hr in four cases, I to 7 days in 4 cases, and over 7 days in five cases.

Delayed PTA.

In some cases all the details of the events of head injury are clearly recalled, but then follows a period of amnesia. This period of lucid interval between the injury and delayed amnesia usually occurs in cases where there is little or no loss of consciousness from the injury. The later occurrence of PTA frequently results from vascular complications of the injury such as extradural hematoma. Russel (1946) reported that 2.5% of 1029 cases of accidental bead injury suffered from delayed PTA. Survivors of gunshot wounds also tend to report a higher incidence of lucid interval (14% in 327 cases reported by Russel). This increased incidence may be the result of more localized brain damage from the gunshot and the absence of diffuse injury.

Paramensia.

Some patients during PTA may give a detailed account of their injury and it may seem, at first, that they had a long lucid interval after the injury. But on close scrutiny these details may turn out to be the details of a previous injury that occurred in a similar setting. These patients are still in PTA and are confused and disoriented for time and place.

Confabulation.

Confabulation refers to filling the gaps in memory by fabrications that are without any basis of fact, but are described as actual occurrences. A florid confabulatory syndrome may occur during PTA and may last for several days. The content of the confabulation appears to be actual experiences from the past but altered in details and context. It is usually associated with a facile euphoria. Russel (1935) reported a patient who was involved in a motorcycle accident caused by a dog. He subsequently described his injuries resulting from an attack by the dog and further elaborated by suggesting that the dog's owner had also attacked him. These delusions are generally short-lived, but they may persist in some cases leading to false accusations and legal difficulties.

Traumatic Automatism.

A blow on the head normally causes concussion. The person falls on the ground, becomes unconscious and motionless. If the blow is minor, recovery is quick. In some cases, a blow on the head may interrupt the memory of the event partially or completely, but does not cause the loss of consciousness or motor activity. The person may continue the routine or repetitive activity. It is determined only sometimes afterward that the person has no memory of the events occurring for some period of time after the blow. This phenomenon is not uncommonly reported by the boxer and football players who may get hit on the head enough to cause a temporary impairment of memory, but they may continue to box or play football.

Changes in Duration of PTA.

The period of PTA may vary from a few minutes to weeks or months. It correlates well with the severity of head injury. Minor head injury causes a brief period of PTA, while severe injury results in a prolonged PTA.

In uncomplicated cases of closed head injury, the duration of PTA is related to the degree of diffuse brain damage. However, secondary complicating factors such as intracranial hemorrhage, brain edema, and intracranial hypertension frequently prolong the duration of PTA.

Although the duration of PTA is more or less permanent after recovery, occasionally it is reduced by the appearance of continuous memories extending backward into the period of PTA. Posttraumatic amnesia may also be lengthened in the early days after recovery. Even after the acquisition of continuous memories, lapses of memory may occur.

Cognitive Deficits during PTA

While the duration of PTA correlates well with a variety of parameters relating to sensory, motor, memory, and psychiatric disturbances, its correlation with deficits on tests of intelligence is somewhat ambiguous. Mandleberg and Brooks (1975) administered the Weschler Adult Intelligence Scale to two groups of head-injured patients, one in PTA and the second out of PTA. Patients tested during the early portion of PTA were able to make appropriate responses to verbal tasks, but they performed poorly on performance tasks, indicating that verbal abilities are fairly intact during PTA while nonverbal skills are disrupted. These performance tasks are to be differentiated from simple repetitive motor skills that are generally intact. The performance tasks of intelligence tests require abilities to integrate complex sets of visual, spatial, and motor skills.

Some patients when tested later during PTA performed substantially better on the WAIS. Their scores were comparable to those of the patients tested at the termination of PTA, indicating that the recovery of the intellectual functions occurs sooner than the recovery of other mental functions.

Retrograde Amnesia

Head-injured patients frequently suffer from a loss of memory for events before the injury. The duration of retrogade amnesia (PA) is taken as the time between the injury and the last clear memory that the patient recalls before the injury. False shortening of the RA duration may result if the accident occurs in the midst of a daily routine, such as driving from home to work. The details may be recalled from past experience, blocking out the actual period of amnesia. Similarly, a false lengthening of PA would occur if the person was preoccupied with his own thoughts and did not pay attention to specific events before the injury. In the great majority of cases, PA is usually of a very brief duration. A few instances of no RA following severe head injury have been recorded.

Cases of RA of long duration (days and weeks or months) are extremely rare and usually require exploration of psychological causes. Symonds (1962) reported three patients with persistent, long PA who were examined at intervals of one or more years after injury: patient 1 suffered RA of 24 hr with PTA of 48 hr; patient 2 had RA of 3 months with PTA of 4 weeks; and patient 3 had RA of one year with PTA of 5 weeks. Symonds felt that in none of these cases was there any evidence to indicate that the long duration of PA was based on psychological causes.

Retrograde amnesia may be selective for certain types of memory. It appears that despite an inability to recall other events occurring during PA, the person is generally aware of the skills and experience learned during the period of PA. Many persons try to fill up the period of PA with islands of memories displaced from the past.

Retrograde amnesia may be separated into two classes: (1) a temporary long period that gradually shrinks during the recovery period, and (2) a permanent RA that is of short duration and persists after recovery.

Shrinkage of RA

During the period of traumatic disorientation, RA appears to be of much longer duration. However, there is a shrinkage of the RA period as the confusion clears up. During this process, memories do not return in strict chronological order. Rather, islands of memory emerge in a haphazard manner and are only gradually related to each other in a chronological fashion. Sometimes, recall of certain important links "brings them all together." This phenomenon may seem like a "spontaneous shrinkage of RA.'

Retrograde amnesia usually clears up after recovery from head injury except in cases of long PTA. There appears to be some relationship between the severity of head injury and the duration of RA, severe injury causing long RA. Retrograde amnesia, however, could result from minor head injury with concussion. Crovitz el al. (1983), in a retrospective study of 1000 college students, found that 24% of the males and 16% of the females had suffered head injury with loss of consciousness sometime in their life, usually in childhood. Thirteen males and 14 females in this group claimed still to have RA of brief durations. There were indications that RA had shrunk with time.

Mechanism of RA

One widely accepted explanation of RA is that a short time is required for the physiological process of memory consolidation to occur, and head injury interferes with the process. This theory is supported by the occurrence of a very short duration of RA in the majority of cases. Other mechanisms such as damage to the storage site may be the basis of a longer duration of RA. The shrinkage of permanent RA with time has not been documented precisely. It is usually explained on the basis of accelerated forgetting (Williams, 1966) and by displacement of past memories to fill the amnesic period of RA.