Strategic retrieval
Confabulation affects remote memories—acquired before brain damage—as much as recent memories acquired subsequent to injury. Thus, an elderly patient who suffered a stroke 2 years ago may be as likely to confabulate about his army days during WW II as about his breakfast this morning. This observation has led to the hypothesis that confabulation is more the result of a deficit in retrieval than of a problem in encoding (registering) memories in the first place.38 (A problem with encoding would predict confabulation only with respect to memories acquired since the brain damage.)
Strategic retrieval refers to memory processes in which the individual uses a purposeful, “problem-solving” strategy to call up the desired memory. For example, someone trying to remember the name of a person he met at a party might begin by thinking, “Let’s see, I was standing next to the punch bowl. Then Mary said she wanted to introduce me to a friend of hers. Then the friend said . . . ,” etc. Thus, memory traces are deliberately organized by context, theme, and temporal order. According to the strategic retrieval hypothesis, a defect in these search processes ultimately leads to spontaneous and/or provoked confabulations.38,39
However, Nedjam and associates24 argue that if strategic retrieval deficits were responsible for confabulations, they should affect episodic and semantic memory equally. (Episodic memory involves autobiographical, contextual, and highly specific information, such as “I ate fish for dinner today”; semantic memory involves general or conceptual knowledge, such as “rectangles have 4 sides”.) In fact, however, confabulations have consistently been shown to affect episodic memory more than semantic memory—thus casting doubt on the strategic retrieval hypothesis.24
Temporal confusion
Finally, there is the temporal confusion hypothesis of confabulation, which is derived from the observation that spontaneous confabulations can typically be traced back to actual (rather than fictitious) events.3,13,19 This hypothesis holds that confabulations arise from the misattribution of aspects of past events to ongoing reality.3,4,27 For example, Schnider and coworkers40 described a 62-year-old woman who exhibited confabulations following a stroke that appeared to consist of mistakenly arranged elements of actual events: “She seemed to recognize the personnel on the ward but often confused their names or confabulated on the circumstances of their meeting. For example, when asked whether she recognized one of the examiners, she explained: ‘You are Dr S (correct). We did physiotherapy this morning (she had had physiotherapy, but not with Dr S), then we worked on the computer (she had regular computer training, but not that morning and never with Dr S) where you showed me a camel and other animals composed of dots (referring to a perceptual priming test taken 2 weeks previously)’.”40(p187)
Thus, consistent with the temporal confusion hypothesis, Schnider and coworkers40 demonstrated that a patient with spontaneous confabulations was capable of storing novel information normally but was unable to store the temporal order of the acquired information—what the authors aptly call, “memory without context.”
CLINICO-ANATOMICAL CORRELATIONS
Confabulation has been described in numerous neurological conditions, including dementia, traumatic brain injuries, anterior and posterior communicating artery aneurysm rupture or repair, subarachnoid hemorrhage, brain tumors, CNS infections, Wernicke-Korsakoff syndrome, and multiple sclerosis. Regardless of the inciting conditions from which they arise, spontaneous confabulations have been associated with lesions of the medial orbitofrontal cortex and its associated anterior limbic structures; namely, the basal forebrain, medial hypothalamus, right capsular genu, and dorsomedial thalamic nucleus.41 The medial orbitofrontal cortex and basal forebrain are supplied by the anterior communicating artery, which connects the left and right anterior cerebral arteries in the circle of Willis. As exemplified in our opening vignette, rupture or aneurysm of the anterior communicating artery is disproportionately associated with the development of spontaneous confabulations compared with other neurological conditions.19,31
Schnider and associates13 have previously observed that provoked confabulation has been associated with dorsolateral prefrontal as well as medial temporal (hippocampal) lesions. However, provoked confabulation has also been observed in neurologically healthy persons.12 Schnider28 therefore concluded that provoked confabulations have no anatomic specificity.
In contrast, a recent neuroimaging study by Turner and coworkers42 provided striking evidence that the critical deficit for provoked confabulation has its anatomical location in the inferior medial frontal lobe. Curiously, whereas Turner’s group examined only provoked confabulations, the lesions observed are consistent with those that are associated with spontaneous confabulations as described above. This suggests there may be substantial overlap in the neuroanatomical loci mediating these 2 types of confabulation.
CONFABULATION IN PSYCHIATRIC SETTINGS
Wernicke-Korsakoff syndrome
As noted, Korsakoff was the first clinician to describe confabulation formally; he observed the syndrome predominantly in chronic alcoholics but also in patients with other conditions.1 Today, Korsakoff syndrome is generally defined as the chronic amnesic syndrome that frequently follows acute Wernicke encephalopathy. The latter comprises the classic triad of confusion, ataxia, and ophthalmoplegia.43,44 So-called Wernicke-Korsakoff syndrome is believed to result primarily from thiamine(Drug information on thiamine) deficiency and can be seen in a myriad of medical conditions besides chronic alcoholism, including malnutrition, protracted vomiting, carbohydrate loading, chronic renal failure, and other chronic disease states.43,44
Patients with Korsakoff syndrome tend to confabulate most within the episodic/autobiographical memory domain.15 Although both spontaneous and provoked confabulations have been described in patients with Korsakoff syndrome, provoked confabulations are more common.14,15 Curiously, however, most neuroanatomical correlations have been established for the spontaneous confabulations less commonly seen in Korsakoff syndrome. Anatomical areas of the brain affected in patients with Korsakoff syndrome typically involve diencephalic regions such as the mammillary bodies and the region enclosed by the anterior and mediodorsal thalamic nuclei. Lesions in the medial thalamic nuclei are believed to give rise to spontaneous confabulations in patients with Korsakoff syndrome. These nuclei contain projections to and from the posterior orbitofrontal cortex.14,15
Alzheimer disease
In addition to the well-known memory problems seen in patients with AD, so-called intrusions are also frequently observed in these individuals. Intrusions are defined as unintentional productions of inappropriate responses in a memory task, and they have similarities to provoked confabulations.24,45,46 A recent study suggests that intrusions in patients with AD may be the result of interference of strongly represented, over-learned material in episodic memory.46 For example, if patients with AD are told to remember a deliberately altered version of a well-known fairy tale—such as “Goldilocks and the Four Bears”—they will often recount the story in their habitual, over-learned mode as “Goldilocks and the Three Bears.” Some neuropsychologists consider such intrusions as essentially provoked confabulations.
More recently—and perhaps more relevantly for psychiatrists—confabulation in AD has been associated with psychotic features and aggressive behavior. Lee and associates17 studied 32 AD patients and 10 healthy controls and found that those with AD confabulated in response to all types of questions tested, including personal episodic memory, orientation to time, and future planning. However, those patients who also demonstrated delusional and/or aggressive behavior showed significantly more provoked confabulations than nondelusional and nonaggressive AD patients.17 Whether these results demonstrate that provoked confabulations in AD patients differ from those seen in other conditions, or whether delusions and aggression simply aggravate mechanisms that underlie all provoked confabulations requires further investigation.
