This topic was considered in some detail by Hodges (1995) in the first edition of this Handbook, and it has also been reviewed in detail elsewhere (Kopelman 1993, 2000a; Kapur, 1999).
There is little doubt that large temporal lobe lesions produce an extensive RA (e.g. Cermak &O'Connor, 1983; Wilson &Wearing, 1995). Cermak&O'Connor (1983) studied a patient, S.S., who had suffered herpes encephalitis resulting in bitemporal pathology and a severe anterograde amnesia. On a test of famous faces (Albert et al., 1979), he showed a marked impairment in performance but relative sparing of early memories (a "temporal gradient"). During follow-up over a number of years, there was substantial improvement. Asked about events from his past life (1930s-1970s) on a questionnaire, he showed impairment for the two most recent decades only. Given cue-words, and asked to describe specific episodes from his past relating to those words (the so-called "Crovitz test"), S.S. seemed to display only a "personal pool of generalized knowledge about himself, i.e. his own semantic memory". On the other hand, his past knowledge about physics and laser technology (his profession) appeared to be intact, although he was not able to retain information encountered in a new article about the subject. A further study showed that he was impaired in recalling and recognizing the meaning of words which had come into the language only since the onset of his amnesia (Verfaellie et al., 1995a). Hodges (1995) argued that there was not a temporal gradient in herpes encephalitis patients, but data from the Cermak & O'Connor study and that of Kopelman et al. (1999) suggest that there is, in fact, a weak or "gentle" one. Fujii et al. (2000) placed particular emphasis on the importance of damage to the temporal pole in such patients, arguing that such pathology disconnects pathways between the hippocampi and the temporal lobe neocortex.
Large frontal lobe lesions can also produce retrograde memory loss, particularly if bilateral. Baddeley & Wilson (1986) described impoverished retrieval of autobiographical memories in two such patients and florid confabulation in the retrieval of autobiographical memories in two others. Levine et al. (1998) reported a patient who had a right frontal lesion with a severe and disproportionate retrograde amnesia. These authors emphasized damage to the uncinate fasciculus, causing disconnection between the frontal and temporal lobes. Group studies of patients with neuroradiologically delineated frontal lesions have also demonstrated severe impairments in autobiographical memory retrieval (Della Sala et al., 1993; Kopelman et al., 1999) and knowledge of famous faces or news events (D'Esposito et al., 1996; Mangels et al., 1996; Kopelman et al., 1999).
With respect to diencephalic lesions, there have been many investigations of Korsakoff patients, and there is little doubt that they generally show an extensive RA with a relatively steep temporal gradient (Zola-Morgan et al., 1983; Butters & Cermak, 1986; Kopelman, 1989; Kopelman et al., 1999). Kopelman (1989) found an RA extending back approximately 25 years, and that Korsakoff patients had a steeper temporal gradient than did Alzheimer patients across several different types of remote memory test. Squire et al. (1989) compared the performance of seven patients with the alcoholic Korsakoff syndrome with that of five patients of more acute onset on six tests of remote memory. In both groups, there was a temporally-graded RA extending across a period of about 15 years, and the temporal gradient was closely similar between the two groups. Verfaellie et al. (1995b) examined remote memory for semantic information in Korsakoff and other amnesic patients, using a test of vocabulary for words that had come into the language between 1955 and 1989. They found an impairment in the Korsakoff patients in the recall of these words, and that there was a temporal gradient such that their knowledge of recent words was more impaired than that of remote words. However, the concomitant presence of some degree of frontal lobe atrophy in Korsakoff patients appears to contribute to their RA. This was first postulated on the basis of correlations between RA scores and frontal/executive test performance by Kopelman (1991), who showed that 68% of the variance on remote memory tests could be accounted for by a regression based on executive test scores. Verfaellie et al. (1995b) also obtained a significant correlation between performance on their remote memory test and a "composite" frontal test index.
Findings in patients with diencephalic lesions from vascular aetiology or tumours are more variable. Winocur et al. (1984) did not find RA in a thalamic infarction patient, neither did Graff-Radford et al. (1990) in the majority of their cases, nor did Guinan et al. (1998) and Kopelman et al. (1999) in pituitary tumour cases (despite anterograde memory disorder in pituitary tumour patients, which was sometimes severe). On the other hand, Stuss et al. (1988) reported severe and extensive RA in a patient with bilateral paramedian thalamic infarction, and Hodges & McCarthy (1993) described severe autobiographical memory loss with relatively preserved knowledge of famous people in a somewhat similar patient who showed severe impairments on tests of executive function. Hodges (1995) attributed such RA to a disconnection between frontal (retrieval and initiation) and posterior (storage) components of the memory system.
Most controversial is the issue of whether damage apparently confined to the medial temporal lobes produces an extensive RA. Following a bitemporal lobectomy, patient HM appeared in initial studies to have an RA of only 2-3 years (Milner, 1966, 1972). This appeared to be confirmed on tests of famous faces and famous news events (Marslen-Wilson & Teuber, 1975; Gabrieli et al. 1988), although performance on the Crovitz test suggested a memory impairment extending back 11 years before the operation (Corkin, 1984). Zola-Morgan et al. (1986) found a 2 year RA in a patient with moderately severe AA following hypoxic brain damage to the CA1 regions of the hippocampi bilaterally. In contrast, Nadel & Moscovitch (1997) reviewed reports which suggested that hippocampal pathology alone can produce an extensive RA. Unfortunately, many of the studies they reviewed involved patients with extensive temporal lobe pathology, confounding the interpretation of the findings: this issue will be taken up again below (see section on Theories).
Kopelman et al. (1999) carried out the first study comparing groups of patients with temporal lobe, frontal lobe or diencephalic lesions across several different RA tasks, including recall of autobiographical incidents, personal semantic facts and famous news events. As already mentioned, Korsakoff patients (with combined diencephalic and frontal pathology) showed severe RA across all tasks with a relatively steep temporal gradient, whereas patients who had been treated for pituitary tumours extending into the diencephalon showed moderate or severe AA but no evidence of RA. Patients with temporal lobe pathology also showed a severe RA, although their temporal gradients appeared to be "flatter" than those of Korsakoff patients. Patients with frontal lobe lesions showed severe impairment in the recall of autobiographical incidents and famous news events, but were relatively intact in the retrieval of well-rehearsed personal semantic facts, i.e. this latter group seemed to be particularly impaired where "effortful" or organized retrieval processes were required for reconstructing "old" memories. In general, patients with bilateral frontal lesions performed worse than patients with unilateral frontal lesions. Subsequent analysis has shown significant correlations with quantitative MRI measures of the volume of specific brain structures in these patients, such that 60-68% of the variance on autobiographical memory tasks could be accounted for by changes in these regional brain volumes.
In summary, large temporal lobe or frontal lesions can produce an extensive RA. More controversial are the specific effects of isolated diencephalic or medial temporal lobe pathology. Diencephalic lesions appear to contribute to an extensive RA when there is concomitant frontal pathology or frontal/executive dysfunction. The contribution of medial temporal lobe pathology in isolation remains controversial.
DISSOCIATIONS IN RETROGRADE AMNESIA Autobiographical vs. Semantic Remote Memory
Autobiographical memory refers to a person's recollection of past incidents and events, which occurred at a specific time and place. Episodic memory is sometimes used in a somewhat broader sense, encompassing autobiographical memory as well as performance on certain learning tasks (e.g. recall of a word list). However, the terms "autobiographical" and "episodic" are often used interchangeably. In the RA literature, autobiographical memory is commonly tested on tasks such as the Crovitz and the AMI. Semantic memory is commonly defined as referring to knowledge of language, concepts, and facts that do not have a specific time or location: they may once have been learned at a particular time and place, but these contextual aspects are not retained. The more semantic aspects of RA are commonly assessed on tests of famous faces, famous names, or famous news events. However, it is important to note that there may be a continuum of knowledge across these domains. First, "personal semantic" facts, such as the names of past schoolteachers or acquaintances, fall midway between the more purely autobiographical and semantic aspects of knowledge. Second, performance on many existing retrograde memory tests, involving famous faces or news events, may involve both autobiographical and more purely semantic knowledge (e.g. a picture of the mangled Mercedes in which Princess Diana died may conjure up simultaneously both the "semantic" facts of the accident and autobiographical information about what you were doing at the time). Third, it is difficult to develop equivalent tests of autobiographical and semantic remote memory for comparative purposes (Kopelman & Kapur, 2001).
Nevertheless, the most common distinction employed in the RA literature is that between autobiographical (episodic) memory and semantic memory (e.g. Kopelman, 1993). Indeed, Kapur (1999) has employed this distinction as the fundamental division within an hierarchical model of retrograde amnesia, in which there are further subdivisions between "pre-ictal" and "extended" episodic RA and between "semantic" knowledge of people and of events.
De Renzi et al. (1987) reported the case of a 44 year-old woman who, following an episode of herpes encephalitis, displayed a severe impairment of semantic knowledge, contrasting with normal memory for autobiographical events. She was impaired at tasks demanding the retrieval of words or of their meaning, e.g. on a confrontation naming test, a sentence verification test, and a test requiring classification of items to categories. More pertinently, she was impaired at knowledge of famous people (including Hitler, Mussolini and Stalin) and at knowledge of public events, e.g. she was unable to provide any information about either the Second World War or the recent (at that time) assassination of the Italian Prime Minister. Cueing helped her in some instances, but she was never able to achieve detailed recollection of these public events. By contrast, not only did she remember personal incidents that had occurred before and after the acute stage of her illness, but she was well informed on current issues in her family, and she could recall the bulk of what had been done from testing session to testing session. A 20-item questionnaire was constructed about autobiographical memories, and her performance was generally very satisfactory. An MRI scan showed a large irregular area of increased signal density extending over the inferior and anterior part of the left temporal lobe, above and lateral to the temporal horn (which was enlarged), involving the amygdala, the uncus, the hippocampus and the parahippocampal gyrus. The frontal lobes and the language areas of the temporal and parietal lobes were spared. In the right hemisphere, there were only minimal signs of increased signal density in the white matter of the inferior temporal lobe.
Subsequent studies have reported disproportionate impairment of knowledge of public events, compared with autobiographical memories, in other patients with left hemisphere pathology, including: a large left parietal lesion following head injury (Grossi et al., 1988); left temporal lobe epilepsy and/or temporal lobectomy (Barr et al., 1990; Kapur et al., 1989); and bilateral irradiation necrosis of anterior/inferior temporal lobe structures (Kapur et al., 1994; Yasuda et al., 1997). Patients with semantic dementia resulting from left temporal lobe atrophy also show disproportionate semantic memory loss. Some studies report preservation of "recent", but not more distant, autobiographical memories in semantic dementia (Snowden et al. 1996; Graham & Hodges, 1997), whereas other investigations find a more uniform impairment of autobiographical memory retrieval, largely secondary to their semantic memory deficit (Moss et al. 2002).
There are other patients who have been reported to show the opposite pattern of performance, i.e. disproportionate autobiographical memory impairment. Dalla Barba et al. (1990) described a female Korsakoff patient with severe episodic memory problems, but who performed well when asked questions about famous people or events. O'Connor et al. (1992) described a patient who had extensive damage to right temporal lobe structures following herpes encephalitis: this resulted in a disproportionately severe impairment in the recall of autobiographical incidents, relative to remote semantic information. This patient also exhibited severe visuoperceptual deficits, and the authors argued that she might have had a particular difficulty in eliciting the visual images necessary for the retrieval of past autobiographical experiences. Ogden (1993) described a head injury patient who also had a severe autobiographical memory loss, associated with prosopagnosia and visual agnosia, with relative preservation of remote semantic knowledge. In this case, the pathology was more posterior, but projections from the right occipital to the right temporal lobe were disrupted. Ogden (1993) suggested, like O'Connor et al. (1992), that a failure in visual imagery might be contributing to the autobiographical memory loss, and both patients manifested a severe impairment in visual, anterograde memory. Rubin & Greenberg (1998) have reviewed a series of similar cases, in whom "visual memory-deficit amnesia" gave rise to apparently disproportionate impairments in autobiographical memory.
However, group studies have indicated that this simple left/semantic vs. right/ autobiographical distinction, postulated in single-case reports, does not necessarily hold good. Kopelman et al. (1999) did indeed find particularly severe autobiographical memory loss in patients with right-sided temporal lobe damage from herpes encephalitis, relative to patients with left-sided pathology, consistent with several of the studies cited above (O'Connor et al., 1992; Ogden, 1993; Rubin & Greenberg, 1998). But it should also be noted that the right-sided patients were, in addition, impaired on a measure of famous news events, particularly when this test involved perception of famous faces, suggesting that there were both "episodic" and face recognition components to the test. Left-sided patients were particularly impaired when they had to "complete" the names of famous people from the past from word-stems: this was interpreted as reflecting a deficit in the lexical-semantic labelling of remote memories. There was a statistically significant right/left double dissociation between performance on the autobiographical incidents and famous name-completion tasks (see Figure 9.1). By contrast, Eslinger (1998) also examined patients with left or right temporal pathology, finding that left medial temporal lesions caused time-limited retrograde autobiographical memory changes. More extensive left temporal lobe pathology impaired personal semantic memory, but did not affect recollection of autobiographical incidents. However, right temporal lobe lesions did not appear to affect either personal semantic or autobiographical incident recall. Bilateral temporal lobe lesions seemed to be required to cause extensive autobiographical memory deficits, and Eslinger (1998) postulated that interactions between prefrontal cortex and diverse temporal lobe regions were involved in autobiographical memory retrieval.
In summary, there is some evidence that the more semantic remote memories are dependent upon left temporal lobe function, and that the retrieval of autobiographical incidents is more dependent upon the integrity of right temporal lobe structures. However, this distinction is by no means clear-cut, perhaps reflecting the fact that performance on many remote memory tasks involves aspects of both autobiographical and semantic memory. For example, there is evidence that tests involving identification of famous faces or pictures of famous news events may be sensitive to damage in the right temporal lobe, and that it is tests involving the lexical-semantic labelling of remote memories which are particularly sensitive to left temporal lobe damage.
A different type of dissociation in remote memory was reported by Warrington & McCarthy (1988) and McCarthy & Warrington (1992). They described a 54 year-old man who had suffered herpes encephalitis, resulting in bilateral medial temporal lobe damage as well as generalized signal alteration and atrophy of the right temporal lobe neocortex. This patient showed extensive retrograde amnesia for autobiographical episodes and also for knowledge of public events, famous faces and famous names. Despite this, he performed within normal limits at a word-completion task for famous names, requiring him to give the "completed" name to a picture and a name-stem, and at familiarity judgements for famous faces. The authors proposed a dual system for the semantic representation of names and faces: a vocabulary-like fact memory, which was preserved in this patient, and a cognitive mediation memory system, which was impaired. The implication seemed to be that the former was analogous to so-called "implicit" memory, which is preserved in anterograde amnesia.
Several subsequent studies have employed similar tests. Eslinger et al. (1996) compared the performance of two post-encephalitic patients. One had sustained left inferior and anterior medial temporal lobe damage, together with a small right temporal polar lesion, and
AUTOBIOGRAPHICAL INCIDENTS CUED NAME COMPLETION FOR FACES
„ x Unilateral left lesions
Predominantly right lesions
Predominantly right lesions
Unilateral left lesions
Childhood Early adult Recent Time period
Figure 9.1 Herpes patients with right-sided lesions are severely impaired at recalling autobiographical incidents (left panel) but not at a more semantic remote memory task (right panel). Herpes patients with left-sided lesions are only mildly impaired in recalling autobiographical incidents (left) but are severely impaired at the more semantic task (right). The recall of autobiographical incidents was determined using the Autobiographical Memory Interview (AMI). The more semantic task involved naming famous faces to word-stem cues. Two herpes patients with unilateral left temporal damage were compared with three herpes patients with predominantly right temporal lobe damage the other showed right-sided inferior and anterior medial temporal lobe pathology. The left-sided patient was substantially impaired at the name-completion task, when the cue was paired with a famous face, whereas the right-sided patient showed only very mild impairment on this test. Similarly, Kopelman et al. (1999) found that two patients with unilateral left temporal lobe pathology from herpes encephalitis were severely impaired at this test, whereas herpes patients with predominantly right-sided temporal lobe pathology were virtually intact on this test (Figure 9.1), consistent with the Warrington & McCarthy (1988) and Eslinger et al. (1996) results. However, Reed & Squire (1998) found impairment in patients with temporal lobe lesions at a more difficult test, in which only the names were presented for completion (in the absence of famous faces), indicating that task difficulty (as well as the side of lesion) may need to be taken into account in evaluating such findings.
In addition, several studies of psychogenic amnesia have addressed the issue of whether there is evidence of preserved "implicit" memory for remote facts or events in the presence of a severe "explicit" retrograde amnesia (see Chapter 21). However, the interpretation of findings in psychogenic amnesia is very difficult in the absence of any very definitive knowledge of whether there is preserved "implicit" RA in organic amnesia.
In summary, it is an attractive idea that there may be preserved implicit memory in retrograde amnesia, analogous to what has been found in anterograde amnesia. However, it is much harder to produce unequivocal evidence of this in RA than in AA. The few studies to date remain vulnerable to alternative interpretations; in particular, that they may have simply demonstrated a cued response in explicit memory.
As already mentioned, Kapur (1999) distinguished two classes of episodic RA, which he called "preictal" and "extended" RA. The case for qualitatively different types of RA, broadly related to differing time-spans, has also been made by other authors (e.g. Symonds, 1966; Squire et al., 1984). Kapur (1999) proposed four arguments in favour of this distinction:
1. There are qualitative discontinuities in the density of pre-injury memory loss reported by patients, e.g. following head injury, there is commonly a short, virtually complete RA lasting a matter of minutes or (sometimes) hours. In some patients, there is also a far less dense loss of memory for incidents or events over the preceding few weeks (Russell & Nathan, 1946; Williams & Zangwill, 1952).
2. Following a closed head injury, an extensive RA characteristically shrinks to a much briefer period, which may be a matter of minutes, hours or days, depending upon the severity of the injury (Russell & Nathan, 1946; Williams & Zangwill, 1952; Wasterlain, 1971). Kapur (1999) also noted that, following episodes of transient global amnesia (TGA), there is commonly some residual "preictal RA", lasting a matter of minutes or (exceptionally) hours (Fisher, 1982; Hodges, 1991; Kapur et al., 1998).
3. There is a delayed onset to certain types of brief RA. For example, Lynch & Yarnell (1973) studied American footballers who had incurred a mild head injury. These footballers were initially able to describe what had happened just before the blow, but, when re-interviewed some minutes later, they were unable to recall these events. This has usually been interpreted as a failure of memory consolidation following the blow, resulting in a period of brief RA lasting a matter of minutes. Consistent with this, Russell & Nathan (1946) also noted that, in some patients, pre-traumatic events were briefly recalled in the first few minutes following a head injury, but were then rapidly forgotten.
4. Some experimental studies also support this distinction. Electrical stimulation (under local anaesthetic) of temporal lobe regions in epileptic patients with complex partial seizures produces a period of preictal RA, ranging from a few minutes to a few days or weeks (Bickford et al., 1958). Electroconvulsive therapy (ECT) also gives rise to a brief RA lasting a matter of days (Squire et al., 1981) and complaints of memory loss which may go back 2-3 years (Squire & Slater, 1983).
A fifth argument is that there are at least some patients with lesions confined to the diencephalic/medial temporal lobe structures who have an RA which extends back 2-3 years but no further (Milner, 1966; Zola-Morgan et al., 1986; Dusoir et al., 1990; Graff-Radford et al., 1990; Guinan et al., 1998). This is a controversial topic but, as will be discussed below, one view is that cortical damage seems to be required for an extensive RA going back years or decades, e.g. as seen in the Korsakoff syndrome, herpes encephalitis, or Alzheimer or Huntington's dementia (e.g. Albert et al., 1981; Kopelman, 1989; Wilson et al., 1995; Kopelman et al., 1999).
In summary, the present observations indicate that the nature of RA may vary according to whether it covers a matter of (a) seconds, minutes, or hours; (b) days, weeks, or months up to a period of 2-3 years; or (c) an extensive retrograde memory loss covering years or decades. The precise boundaries of these different types of RA, and whether they should be differentiated into two, three, or more sub-types, remain unclear. To date, most of the neuropsychological literature has concentrated upon an understanding of the nature of extensive RA, but the briefer components are also important.
Was this article helpful?