Article Text

This article has a correction. Please see:

Download PDFPDF

Basal forebrain amnesia: does the nucleus accumbens contribute to human memory?
  1. Georg Goldenberg,
  2. Uwe Schuri,
  3. Olaf Grömminger,
  4. Ursula Arnold
  1. Neuropsychological Department, Bogenhausen Hospital, Munich, Germany
  1. Dr G Goldenberg, Neuropsychologische Abteilung, Krankenhaus München Bogenhausen, Englschalkingerstrasse 77, D 81925 München, Germany. Telephone 0049 89 9270 2106; fax 0049 89 9270 2089; emailGeorg.Goldenberg{at}


OBJECTIVE To analyse amnesia caused by basal forebrain lesions.

METHODS A single case study of a patient with amnesia after bleeding into the anterior portion of the left basal ganglia. Neuropsychological examination included tests of attention, executive function, working memory, recall, and recognition of verbal and non-verbal material, and recall from remote semantic and autobiographical memory. The patient’s MRI and those of other published cases of basal forebrain amnesia were reviewed to specify which structures within the basal forebrain are crucial for amnesia.

RESULTS Attention and executive function were largely intact. There was anterograde amnesia for verbal material which affected free recall and recognition. With both modes of testing the patient produced many false positive responses and intrusions when lists of unrelated words had been memorised. However, he confabulated neither on story recall nor in day to day memory, nor in recall from remote memory. The lesion affected mainly the nucleus accumbens, but encroached on the inferior limb of the capsula interna and the most ventral portion of the nucleus caudatus and globus pallidus, and there was evidence of some atrophy of the head of the caudate nucleus. The lesion spared the nucleus basalis Meynert, the diagnonal band, and the septum, which are the sites of cholinergic cell concentrations.

CONCLUSIONS It seems unlikely that false positive responses were caused by insufficient strategic control of memory retrieval. This speaks against a major role of the capsular lesion which might disconnect the prefrontal cortex from the thalamus. It is proposed that the lesion of the nucleus accumbens caused amnesia.

  • amnesia
  • basal forebrain
  • nucleus accumbens

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Research in human amnesia has identified three core regions where localised brain damage can cause a lasting amnesic syndrome: the medial temporal lobe, the thalamus, and the basal forebrain.1 However, the weight of the evidence for a selective correlation between circumscribed lesions and amnesia is different for these three locations. There are many case reports showing that circumscribed lesions of the medial temporal lobes and of the medial and anterior portions of the thalamus can cause amnesia.2-7 The evidence for a selective role of the basal forebrain in human memory is weaker and more indirect. It has originally been derived from degeneration of cholinergic cells within the basal forebrain in Alzheimer’s disease.8 9 As amnesia is a core symptom of Alzheimer’s disease it has been concluded that basal forebrain lesions can cause amnesia. This conclusion is not compelling as neuronal degeneration in Alzheimer’s disease affects medial temporal structures as well.10 11 More direct evidence for basal forebrain amnesia is provided by patients with amnesia after circumscribed damage to only the basal forebrain.12-22 In most of them, the lesion was caused by haemorrhage from anterior communicating artery aneurysms and subsequent infarction of perforating arterial branches or Heubner’s recurrent artery.12-14 16 17 19-22However, subarachnoidal or intraventricular haemorrhage may cause hydrocephalus and diffuse brain damage. This raises the possibility that at least some of the symptoms of these patients are due to diffuse brain damage rather than to the basal forebrain lesion.23

The anatomicoclinical relation of basal forebrain amnesia is further complicated by the proximity of the basal forebrain to the frontal lobe. Lesions from anterior artery aneurysms often affect the orbital and medial frontal lobe.12 17 24 25 Disinhibition or unconcern related to frontal lobe pathology may distort or conceal the amnesic syndrome.16 Particularly, confabulations which are a frequent feature of basal forebrain amnesia have been related to accompanying frontal lobe pathology rather than to the basal forebrain lesion itself.26

The uncertainty of the anatomicoclinical relation increases when localisation within the basal forebrain is considered. Inspired by the parallel with Alzheimer’s disease, some authors have considered damage to cholinergic cells in the septum, the diagonal band, or the nucleus basalis of Meynert as being the cause of amnesia.12 18 An alternative account holds that lesions of fibre tracts rather than cells are crucial for amnesia.15 20 The basal forebrain is traversed by the inferior and anterior thalamic peduncles which connect the dorsomedial nucleus of the thalamus with the amygdalae and with the prefrontal cortex. Amnesia may be caused by interruption of an amygdala-thalamic-prefrontal loop rather than by basal forebrain damage itself.1 15 27

In this paper we describe a patient with amnesia from a circumscribed basal forebrain lesion which did not extend into the frontal lobes and which largely spared structures rich on cholinergic cells.

Case report

The patient was a 55 year old right handed engineer who held a managing position in an international computer company. He had arterial hypertension. In July 1996 he became somnolent and confused. On emergency admission, CT showed bleeding into the anterior portion of the left basal ganglia with invasion into the left frontal horn of the lateral ventricle. No source of bleeding could be identified on angiography. During the subsequent weeks the medical records noted mild right sided weakness, word finding difficulties, and paraphasia. His ability to store new information was severely defective and he was disoriented to time.

He was admitted as an outpatient to our department in October 1996. The hemiparesis had resolved and conversation was inobstrusive. He was fully oriented, but complained of memory problems: He rated himself as very poor at remembering names, places, and the content of conversations. His spouse confirmed that he often enquired about details which had been talked about only a few minutes ago. In addition he noted an impoverishment of his autobiographical memories and a loss of professional knowledge. He could not remember geographical facts, names of other companies, mathematical formulae, or phone numbers which he had known by heart before. He said that he had lost his previous knowledge of foreign languages. He had to translate laboriously even English which had been the everyday language in his job. Visiting a place where he had spent several holidays, his wife and he noted that he had difficulties in finding familiar routes. As well as his memory problems the patient complained of a loss of libido and of a flattening of emotional reactions. He said: “There is nothing which makes me really happy or really sad. It’s all far away”.

Neurological examination showed only a slight reduction of fine motor skill of the right hand, which disappeared completely during the subsequent weeks.


Brain MRI (figure) showed a sharply demarcated, slit-like, lesion which ran from the bottom of the left frontal horn in a slightly curved way laterally. The caudal limit of the lesion was at the level of the anterior commissure. The lesion destroyed a large portion of the nucleus accumbens and encroached on the inferior limb of the capsula interna and the most ventral portions of the anterior nucleus caudatus and globus pallidus. There was no general enlargement of ventricles, but the wall of the left frontal horn was slightly widened and flattened, indicating some atrophy of the head of the caudate nucleus.

(A) T 1 weighted MRI of the lesion. (B) Anatomical scheme of the centre of the lesion, corresponding to the leftmost image of the bottom row of the MRI. The right side of the figures corresponds to the left side of the brain. GP=globus pallidus; Cd=caudate nucleus; Acb=nucleus accumbens; CI=capsula interna; DB=diagonal band.

Tc-99-HMPAO SPECT showed hypoperfusion of left basal ganglia (92% of homologuous right brain region), left frontal cortex (91%) and left basal and medial temporal cortex (91% and 90%).



The patient scored well within the normal range on a battery of computerised tests probing reaction time, speed of visual scanning, divided attention, and shifting of attention.28 His attention/concentration index value on the WMS- R29 was 111.


In German, his native language, speech was well articulated, fluent, and syntactically correct. Only when talking about difficult and complex matters did his expression become imprecise and circumstantial with occasional confusions of low frequency words. He perfectly named 20 highly familiar items from the Snodgrass-Vanderwart pictures30 31 but made two errors for 20 unfamiliar items (“perhaps a goose?” for ostrich, and skittle for spinning top). Knowledge of second languages was not assessed.


On a multiple choice lexical decision test with words of decreasing frequency,32 reflecting premorbid intelligence level, his score was equivalent to an IQ of 124. On the German WAIS-R33 his performance IQ was 121 and his verbal IQ was 100. In the verbal part the poorest result was on arithmetic (scaled score 6) whereas the other subtests yielded scaled scores between 10 and 13. In the performance part scaled scores on picture completion and block design were superior (18 and 15) and the other scaled scores ranged from 10 to 12.

Executive functions

He obtained normal scores on the modified Wisconsin card sorting test,34 the Tower of London test,35 and the six elements test.36 Design fluency37 was normal (25/3 minutes: 40th percentile of normal controls). Fluency for words with a given initial letter was lower albeit still within the normal range (20/3 minutes: 15th percentile).

Primary and working memory

Visual span was above average (WAIS-R: forwards 98th percentile; backwards 90th percentile) and digit span average (forwards 51st percentile, backwards 31st percentile). A weakness of verbal working memory manifested itself in tests which demand both maintainance and processing of verbal information—for example, reading span38 (span=2, 6th percentile) or a computerised test of verbal working memory28 (level 3, 1st percentile). The poor result on WAIS-R arithmetic may also relate to this weakness.

Anterograde memory

A selection of tests of anterograde memory is given in the table. In accordance with the patient’s complaints, the reports of his spouse and our findings, the Rivermead behavioural memory test39confirmed ecologically significant memory impairment. Recall and recognition of verbal information were severely defective. By contrast, recall and recognition of non-verbal visual information were normal. On the Berlin amnesia test,40 which provides normative data for comparing recall and recognition, recognition of words seemed to be less affected than free recall. A story with emotional content was better recalled than a neutral one. A notable feature of his performance on tests of free recall of words was the production of intrusions. For example, in recall after interference of a German adaptation of the California verbal learning test (CVLT)41he produced more intrusions than list words. Likewise, the poor scores on verbal recognition memory were largely due to false positive responses. Indeed, he recognised roughly equal numbers of faces and words, but whereas the total score on face recognition was normal, the score on verbal recognition fell to near chance levels because of the many false positive responses. However, he confabulated neither in story recall nor in daily life.

Anterograde memory

Semantic memory

A preservation of basic semantic knowledge was manifested by his average score on the WAIS-R subtest information and by the absence of significant naming difficulties. However, he did have some word finding difficulties for unfamiliar items (see above) and categorical fluency42 was mildly impaired with the exception of musical instruments (animals: 13/1 min; birds: 4/1 min; household items: 15/1 min; musical instruments: 12/1 min). Drawing familiar objects from memory was good. We could not assess his professional knowledge but he credibly assured us that it had become severely defective. On a famous faces test which presents faces of 15 famous persons from each decade and asks for the famous persons’ names and for semantic information about them his scores were at the 5th percentile of age matched controls for persons famous between 1966 to 1975, below the 1st percentile for 1976 to 1985, and between the 5th and 25th percentile for 1945 to 1965 and after 1986. In this test he never produced a wrong name and only once wrong semantic information (“American Secretary for Foreign Affairs” for Winston Churchill). All other errors were constituted by “don’t know” responses.

Autobiographical memory

The patient complained of an impoverishment of his entire autobiographical memory. On the autobiographical memory interview43 knowledge about autobiographical facts was good for childhood and for recent time (17.5 and 18 out of 21) but defective for early adulthood (11/21). Recall of autobiographical incidents was borderline to abnormal for all time periods and showed a similar time gradient with the most severe loss for early adulthood (childhood: 6/9; early adulthood: 1/9; recent time: 4/9). Recall of autobiographical episodes was further explored with a modified Crovitz technique.44 He was presented with 60 cue words (for example, train, friendship, hostility, swim, laugh, destroy). He was asked to produce an autobiographical episode related to each word and to say from which period of his life it stems. Episodes were rated for specifity and richness. He produced only 35 episodes, which is below average. In addition, he gave 12 general statements without episodic character, and refused to produce more specific information when requested to do so. In parallel to the famous faces test, production was poorest for the period from 1976 to 1985. We could not check the veracity of those episodes which he produced but they did seem to be plausibly true memories.


The patient had a severe anterograde memory deficit for verbal material affecting free recall and recognition. With both modes of testing a significant source of errors was constituted by false positive responses and intrusions. However, the patient confabulated neither on story recall nor in day to day memory. Retrograde memory was impoverished for both semantic and autobiographical information. On tests of retrograde memory he virtually never produced wrong information but stated that he did not know the response.

Primary memory for verbal material was within normal limits. Only when he had to maintain and simultaneously process verbal material his performance fell below normal values. He scored normally on tests of executive functions. Verbal fluency was lower than graphic fluency, and there were some word finding problems for unfamiliar items. These language difficulties were, however, far too mild to plausibly explain his severe amnesia for any verbal material.

The responsible lesion was unilaterally left sided. It had its centre in the nucleus accumbens but encroached on the inferior limb of the internal capsula and the ventral portion of the globus pallidus and nucleus caudatus. There was evidence of some atrophy of the head of the caudate nucleus and of frontal and temporal hypoperfusion.

Similar to patients with unilaterally left sided medial temporal or thalamic lesions4 6 45 and two other patients with strictly unilateral left sided basal forebrain lesions15 21 he had amnesia only for verbal material and no significant deficits of general intelligence, attention, or executive function. This constellation is unlikely to be a manifestation of diffuse brain damage.23 We therefore feel justified to interprete his amnesia as a sequel of the circumscribed basal forebrain lesion.

In addition to the anterograde memory deficit, he had difficulties recalling premorbidly acquired autobiographical and semantic information. As these were tested mainly verbally, it is not clear whether the deficit affected visual memory as well. Drawing of objects from memory was normal, but the husband and his wife reported some difficulties with retrieval of topographical memories. Recall of autobiographical episodes and of famous persons was poorest for the years from 1976 to 1985 when he was between 35 and 45 years old. Presumably, the poor result for “early adulthood” in the autobiographical memory interview included the same period. This gap may be an enhancement of the paucity of memories from this period of life which has been documented in normal subjects older than 50 years.46

If anterograde and retrograde amnesia had a common cause, this would be most likely to be a defect of retrieval rather than of memory consolidation or storage. There are, however, other patients with basal forebrain amnesia in whom premorbidly acquired autobiographical20 21 and semantic20memory were normal. Apparently the combination of anterograde and retrograde memory loss is not an invariable feature of basal forebrain amnesia. We think, however, that the published data on retrograde memory loss in basal forebrain amnesia are too scarce too permit any definite conclusions regarding their underlying mechanisms. Further discussion will concentrate on the anterograde verbal memory deficit.

The absence of a general weakness of executive function does not rule out the possiblity of a specific weakness of executive control of verbal memory. We discuss this possibility, and then the location of the critical lesion within the basal forebrain.


The patient produced numerous intrusions in recall of word lists, and the inaccuracy of his recognition memory for words was entirely due to the high number of false positive responses. At the same time, he confabulated neither on story recall, nor in recall from remote semantic and autobiographical memory, nor in daily life. False responses were thus restricted to memory for word lists. For these words, the decision whether a word coming to mind during recall was correct or not could be based on nothing else but on memory of this particular word’s occurrence in the word list. There was no way to infer the plausibility of the decision from partial recall of the memorised material or from preserved remote memory. By contrast, tests which did not provoke false responses asked either for memory of organised verbal information (story recall, autobiographical episodes), or for recall of names and information with a defining relation to pictures presented during recall (famous persons). In these tests, he could exploit partial recall of the memorised material or preserved remote memory for judging the plausibility of information coming to mind.

On this account, supervisory control of recall was intact. He used available knowledge for controlling the plausibility of information coming to mind. The source of false responses might be sought in a general enhancement of “feeling of familiarity”24which levelled the signal to noise ratio between memorised items and items which were similar but had not been presented. Feelings of familiarity were assigned indistinctly to any word that came to mind during recall and misled his responses when he had no external evidence for checking the credibility of this feeling. False responses may also have resulted from a strategic decision at the level of supervisory control. He may have opted for a liberal response bias whenever he had no resources for estimating the plausiblity of responses. In any case, intrusions and false positive responses do not indicate a failure of supervisory control of memory.


The basal forebrain consists of three functionally distinct compartments47: the corticopetal, mainly cholinergic, system, the extended amygdalae, and the ventral striatopallidal system. Cholinergic cells are concentrated in the septum, the diagonal band, and the nucleus basalis.48 The extended amygdalae stretch from the centromedial nuclei of the amygdalae medially and rostrally through the substantia innominata into the medial portion of the nucleus accumbens where they join the ventral striatopallidal system. The ventral striatopallidal system is constituted rostrally by the nucleus accumbens and caudally by subcommissural continuations of putamen and globus pallidus. The nucleus accumbens occupies a central position in the functional interplay between these components of the basal forebrain and between the basal forebrain and other brain regions. It constitutes a junction between extended amygdalae and the ventral striatopallidal system and it has efferent connections to the corticopetal system. It is therefore able to modulate cholinergic output.47 48 The nucleus accumbens sends efferents to the hypothalamus, substantia nigra, autonomic brainstam nuclei, and the pallidum, which provides a connection to the dorsomedial nucleus of the thalamus.49 It receives afferent projections from dopaminergic midbrain nuclei and intralaminar thalamic nuclei, from the hippocampus and basolateral amygdalae, and from the prefrontal, insular, and temporal association cortex.47 50-52 It is thus in a position to integrate inputs from multiple cortical and subcortical areas including the hippocampus and the amygdalae, and to exert modulatory influences on widespread cortical function.

Animal studies have suggested a role of the nucleus accumbens in memory. Rats with selective lesions of the nucleus accumbens have been found to fail on tests such as the Morris water maze or delayed win-shift foraging, which depend on mnemonic function and which are known to be sensitive to hippocampal lesions.51 53-55 It has not, however, been settled whether these deficits are reflections of mnemonic dysfunction or of other behavioural disturbances.50 56

Published studies of basal forebrain amnesia lend credibility to a role of the nucleus accumbens in human memory. In a group study of patients with operated anterior communicating artery aneurysms, Irleet al 17 found memory disturbances only in patients with “combined basal forebrain-striatum” lesions. The nucleus accumbens forms part of the ventral striatum. Nucleus accumbens lesions were visible on MRI or CT in several published cases of basal forebrain amnesia.12-14 18 20 In a further patient15the lesion affected the ventral pallidum and in two other patients MRI showed atrophy of the caudate nucleus.21 22 It is notable that in some of these cases,13 15 21 22 as well as in our patient, the nucleus basalis, septum, and diagonal band, which have the greatest concentration of corticopetal cholingergic cells, seem to be unaffected.

In summary, there seems to be rather convincing evidence that affection of parts of the striatopallidal system makes an important contribution to basal forebrain amnesia.17 The evidence for the nucleus accumbens as critical structure is less unequivocal, but does seem to deserve further studies in animals and humans.51

Because the patient’s lesion encroached on the anterior limb of the internal capsula our finding would be compatible with the hypothesis that lesions of fibre tracts rather than cell bodies are responsible for amnesia from extrahippocampal lesions.15 27 In particular, the capsular lesion may have damaged the anterior thalamic peduncle which connects the dorsomedial nucleus of the thalamus to the prefrontal cortex. It should be noted, however, that the inferior thalamic peduncle which connects the basolateral amygdalae with the dorsomedial nucleus, runs caudally from the anterior commissure and seemed not to be affected. Interruption of fibres between the prefrontal cortex and thalamus may disconnect strategic processes supported by the frontal lobes from memory functions depending on the thalamus.57 However, as we have argued in the previous section, strategic control of memory processes seemed to be well preserved in our patient.


Linked Articles

  • Editorial commentary
  • Correction
    BMJ Publishing Group Ltd