Anosognosia

More than a century of research has sought to uncover the neurobiology of anosognosia. Many theories are briefly reviewed herein; however, we refer readers to several seminal works for in depth discussion of such theories (Critchley, 1953; Heilman, 1991; McGlynn & Schacter, 1989; Prigatano & Schacter, 1991; Schacter, 1990; Vuilleumier, 2000, 2004). Von Monakow first described unawareness of deficit due to brain lesion (von Monakow, 1885), and Babinski later introduced the term anosognosia to describe unawareness of hemiplegia (Babinski, 1914). Anosognosia has since been recognized as a broad spectrum of disordered awareness associated with a variety of clinical syndromes (Bisiach et al., 1986), and the term is now used more generally to describe unawareness of impairment associated with any neurologic compromise (Loring, 1999). Herein, we apply the terms anosognosia and disordered awareness synonymously. Several other phrases are frequently used to describe this syndrome, such as “denial of deficit” and “poor insight,” but we refrain from using those, as they less clearly convey the syndrome's neurologic nature.

Although we focus here on neurologically based accounts of anosognosia, a variety of other theories exist. For example, several researchers have advocated motivational theories of anosognosia based on observations that patients who explicitly deny deficits generally had premorbid tendencies to regard illness as an imperfection, deny perceived inadequacies, and seek the esteem of others (Ullman et al., 1960; Weinstein & Kahn, 1955; Weinstein & Kahn, 1953). We also know that sociocultural factors such as minimization of distress contribute to individuals' perceptions of illness (Saravanan et al., 2004; Sussman, 2004; White et al., 2000). Although the role of these psychological and cultural variables cannot be dismissed, such factors cannot account for repeated neuroanatomic evidence linking anosognosia to specific neurologic syndromes and cortical regions such as the right prefrontal and parietal cortices (Heilman, 1991; Vuilleumier, 2004).

Neuroanatomic Perspectives

Neurologic accounts of anosognosia evolved largely in the context of unawareness for hemiplegia. Although severity of primary sensory loss was initially thought to be a primary factor driving unawareness of motor deficit, growing evidence suggested that awareness level and injury severity were frequently dissociated (Cutting, 1978). As such, researchers began to conceptualize anosognosia as a deficit in the higher order processing of a specific skill. For example, the discovery theory posits that direct personal experience with the deficit plus a specific type of mental flexibility may be needed to recognize impairment, and to adjust behavior based on this experience (Levine, 1990; Levine et al., 1991). That is, individuals must discover loss through observation, or infer loss through indirect means. There may be many means by which this discovery is disrupted; indeed, different neurologic injuries appear to alter phenomenological experience in unique ways, and consequently distort different aspects of self-awareness.

Centuries of work on brain injury and stroke have underscored the role of the right parietal cortex in supporting various aspects of self-awareness, particularly awareness for left hemiplegia (Bisiach et al., 1986; Heilman, 1991; Marcel et al., 2004; Meador et al., 2000; Rainville et al., 2003). Given its association with right parietal damage, anosognosia has been theorized to reflect a number of different abnormalities including a disordered body schema (Critchley, 1953; Head & Holmes, 1911), a severe disturbance in attention to the body (Mark et al., 1988), and a disconnection syndrome preventing information from reaching speech centers in the left hemisphere (Geschwind, 1965). While persuasive, such hypotheses were unable to explain why patients frequently remained unaware of hemiplegia even when the paralyzed left arm was brought into the right visual field and patients were asked specifically to attend to the left arm (Heilman, 1991; Heilman et al., 1993; Ramachandran, 1995).

Conceptualizing anosognosia grows even more complex when taking into consideration the large body of research in patients with traumatic brain injury (Stuss et al., 1992), stroke (Feinberg, 2001; Feinberg et al., 1999), schizophrenia (Koren et al., 2004), and frontotemporal dementia (FTD; Grossman, 2000) that highlights the prefrontal cortex as a second neurologic area relevant to personal and social awareness. Interestingly, prominent behavioral disturbances in FTD (Miller et al., 1993) and “disorders of the self” such as delusional misidentification and reduplication syndromes (Feinberg, 2001; Feinberg et al., 1999) have been associated particularly with right rather than left prefrontal compromise. Further, while a thorough discussion of functional neuroimaging studies in healthy adults is beyond the scope of this review, we point out that such studies implicate the right prefrontal cortex as a critical region underlying aspects of self-awareness, such as recognizing one's own face (Platek et al., 2004) and making judgments about one's own memory abilities (Kikyo et al., 2002; Platek et al., 2004; Schnyer et al., 2004).

It is clear that both anterior and posterior portions of the right hemisphere support important aspects of self-awareness, and anosognosia is often seen with injuries spanning the right parietal and frontal lobes (Venneri & Shanks, 2004). In trying to tease apart the role of various cortical areas, Prigatano (Prigatano, 1991) articulately argued that self-awareness is unlikely to be a function of a specific cortical area. Extending Mesulam's (1985) suggestion that the cerebral cortex can be best understood in terms of the type of stimuli to which it responds, Prigatano proposed that various cortical areas likely support awareness for different types of information such that “if one is exploring self-awareness about social judgment, the ability to anticipate change, and so on, the prefrontal regions may be important. In contrast, self-awareness of body and body image may be more influenced by the inferior parietal lobe” (p. 121).

One possible explanation for the importance of the right hemisphere in both prefrontal and parietal syndromes of disordered awareness is that this hemisphere plays a critical role in affective functioning. Emotion has been hypothesized to activate the cognitive appraisal needed to adjust behavior and beliefs according to novel contingencies (Tiedens & Linton, 2001b). To the extent that emotional feedback is disturbed, an individual may not engage in normal verification procedures when confronted with a cognitive or behavioral deficit (Vuilleumier, 2000). Indeed, several studies cite apathy as the main correlate of anosognosia (Derouesne et al., 1999; Ott et al., 1996; Starkstein et al., 1996; Tiedens & Linton, 2001a; Vogel et al., 2004). This conceptualization is reminiscent of Babinski's (1914) and Critchley's (1953) descriptions of anosodiaphoria, lack of concern for hemiplegia, as a more mild manifestation of anosognosia.

It is not clear whether anosognosia in AD is associated with specific cortical changes in the right hemisphere, or whether disordered awareness results from global cortical deterioration. However, striking dissociations in awareness across patients with mild AD suggest that disordered awareness may indeed result from early compromise to a critical brain region or network. In fact, anosognosia in AD has been linked to right hemisphere abnormalities such as reduced perfusion in the right prefrontal cortex (Reed et al., 1993; Starkstein et al., 1995; Vogel et al., 2005) and increased plaque density in the right prosubiculum region of the hippocampus (Marshall et al., 2004). It will be important to further investigate the potential roles of the anterior and posterior regions of the right hemisphere in supporting processes of awareness as measurement of such processes becomes more objective.

The issue of measurement is highly critical to the discussion of anosognosia in AD. As will be detailed below, most of the research regarding the nature, etiology, and correlates of anosognosia in AD has been plagued by inconsistent findings. Discrepant findings certainly reflect the complex nature of anosognosia, but we suggest that methodological limitations are also to blame. Reconsideration of traditional assessment tools may facilitate investigation of anosognosia, and lead to more precise and consistent answers across studies.

Assessing Awareness in Alzheimer's Disease: Methodological Limitations

Nearly 15 years ago, emerging work suggested that some patients with dementia denied their deficit, although it was unclear to what extent such denial was motivationally based (McGlynn & Schacter, 1989). At that time, McGlynn and Schacter (1989) wrote the first integrative review of unawareness of deficit in dementia and other neuropsychological syndromes, criticizing studies for their “lack of conceptual clarity concerning anosognosia” (p. 154) and arguing that “much more attention needs to be paid to defining concepts operationally and devising systematic measurement techniques” (p. 145). In particular, the authors urged researchers to quantify the assessment of anosognosia and to consider this phenomenon as a spectrum rather than a unitary disorder.

Assessment of anosognosia became increasingly quantitative throughout the 1990s with implementation of clinician rating scales (Derouesne et al., 1999; McDaniel et al., 1995; Reed et al., 1993; Zanetti et al., 1999) and discrepancy scores between patient and caregiver report of patient functioning (DeBettignies et al., 1990; Derouesne et al., 1999; Kotler-Cope & Camp, 1995; Mangone et al., 1991; Michon et al., 1994; Smith et al., 2000). The Anosognosia Rating Scale (Reed et al., 1993) is an example of a rating scale consisting of two initial interviews by a neurologist and psychologist who determine the patient's level of functioning. The patient's level of awareness is rated on a four-point ordinal scale covering: ready admission of memory loss, shallow awareness (inconsistent recognition, or recognition of trivial loss), no awareness (bland denial), and angry denial of impairment. Such categorical ratings are certainly valuable in that they provide a broad picture of the patient's awareness level. However, the spectrum of disordered awareness represented by such scales is significantly restricted in range preventing appreciation of more finely graded levels of awareness.

Calculation of discrepancy scores between patient and caregiver report of patient functioning have also been used to quantify anosognosia. Such scales have been informative in that they directly reflect patients' perceptions of their abilities. However, it is not clear that comparing this report with that of a caregiver accurately portrays patient awareness. Even though patients may endorse significant dysfunction, he or she would be characterized as having poor awareness if the caregiver reports a far higher level of dysfunction. Further, it is far from clear that caregiver report is an unbiased standard of accuracy.

Traditional methodologies such as rating scales and discrepancy scores are also less than ideal in that they generally assess awareness as a unitary phenomenon when in fact it may be multi-faceted. That is, patients may display varying degrees and forms of awareness for specific abilities and/or deficits. Starkstein and colleagues (1996) dissociated elements of cognitive and behavioral unawareness in AD, showing that cognitive unawareness was associated with more severe intellectual decline, higher frequency of delusions, more severe apathy, and less depression, while behavioral unawareness was related to higher disinhibition scores and pathological laughing. It is possible that anosognosia can be deconstructed into even more precise components; however, assessment tools must first be designed to capture these components.

Given the limitations of traditional methodologies, and the variable operational definitions of anosognosia used across studies, it is not surprising that findings regarding awareness in AD have been ambiguous (Derouesne et al., 1999; Duke et al., 2002). For example, although some investigations have demonstrated a positive correlation between dementia severity and anosognosia (Barrett et al., 2005; Feher et al., 1994; Lopez et al., 1994; Mangone et al., 1991; Migliorelli et al., 1995a, 1995b; Sevush & Leve, 1993; Starkstein et al., 1996), several studies fail to support this finding (DeBettignies et al., 1990; Michon et al., 1994; Reed et al., 1993; Smith et al., 2000; Zanetti et al., 1999). It appears most likely that anosognosia progresses in severity over the course of the disease in all patients; however, the time at which awareness becomes disrupted varies significantly across individuals. It has been hypothesized that disordered awareness is greater in patients with executive dysfunction (Fernandez-Duque et al., 2000; Lopez et al., 1994; Michon et al., 1994; Reed et al., 1993; Shimamura, 2000; Vilkki et al., 1998); however, this association has not been consistently observed (Dalla Barba et al., 1995; Starkstein et al., 1996; Vogel et al., 2004). Further, there has been no strong consensus for the hypothesized relationship between anosognosia and psychiatric symptoms such as depression and psychosis in AD (Reed et al., 1993; Sevush & Leve, 1993; Smith et al., 2000; Starkstein et al., 1997; Verhey et al., 1993).

It is very possible that studies have produced inconsistent results given the wide variety of subjective methodologies implemented to study anosognosia in AD. We suggest that implementing objective measures of awareness may overcome the limitations of traditional methodologies, and offer consistent metrics to be used across studies. Once this has been accomplished, researchers may begin to uncover the relationship of specific components of anosognosia to disease variables such as prognosis, neuropsychological profile, disease duration, patterns of neuropathology, and the presence of psychiatric symptoms.

Judgment of Learning and Feeling of Knowing Tasks

Nearly 10 years ago, Kaszniak and Zak (1996) encouraged dementia researchers to employ established metacognitive assessments such as feeling of knowing and judgment of learning tasks used extensively in healthy adults. Metacognitive assessment focuses on uncovering the mechanisms by which individuals “know what they know,” and is characterized by multiple objective techniques that have the potential to inform the study of awareness in AD (Hart, 1965; Leonesio & Nelson, 1990; Metcalfe & Shimamura, 1994; Nelson & Narens, 1990; Nelson & Narens, 1980, 1984; Schacter, 1983; Shimamura & Squire, 1986). While the term metamemory is generally used to refer specifically to people's knowledge about their memory abilities and may be especially appropriate for discussing deficits in AD (Metcalfe & Shimamura, 1994), we use metacognition as a more general term throughout this paper to cover knowledge regarding a wider range of cognitive skills.

To a large extent, metacognitive studies have investigated the nature of two complementary processes, monitoring and control, each of which serves a specific regulatory function (Nelson & Narens, 1990). Monitoring refers to the mechanism by which individuals evaluate the correctness of potential responses. This process is based on a collection of information about one's own knowledge and performance, and involves the flow of information from the object level to the metalevel (Shimamura, 2000). Control refers to the self-regulation of behavior, such as the allocation of study time, the initiation of certain cognitive strategies, and the decision to volunteer the best available candidate answer (Klatzky & Erdelyi, 1985).

In advancing the study of anosognosia in AD, we find monitoring processes of particular interest to the extent that they reflect patients' abilities to judge their own cognitive abilities, or more specifically, to recognize their memory deficit. Two commonly used methods of evaluating monitoring processes are Judgment of Learning (JOL) and Feeling of Knowing (FOK) tasks (Metcalfe & Shimamura, 1994). These tasks are presented in conjunction with memory tests in an attempt to assess individuals' beliefs about how readily they will learn, remember, or recognize information. JOL tasks require subjects to predict, at the time information is being studied, the likelihood of subsequently recalling recently studied items (Nelson & Narens, 1990). Alternatively, FOK tasks involve making judgments, after the recall phase of a memory task, about the likelihood of subsequently recognizing nonrecalled semantic information (Hart, 1965) or episodic information (Schacter, 1983). Monitoring accuracy is then determined on an item-by-item basis, by correlating individual judgments and memory scores. Significant evidence suggests that both JOL and FOK ratings reliably predict memory performance in healthy individuals (Hart, 1965; Leonesio & Nelson, 1990; Nelson et al., 1984). Notably, however, these ratings pertain to two qualitatively different memory functions (free recall and recognition), and are only weakly correlated with one other (Leonesio & Nelson, 1990; Souchay et al., 2004).

Modifying JOL and FOK Tasks for Use in Alzheimer's Disease

A handful of innovative studies over the past two decades have applied JOL, FOK, and other objective metacognitive methodologies to studies in AD. In the following section, we review several of such studies, highlighting potential pitfalls and advising researchers to consider a number of issues in future investigations including: (1) How does the timing of the metacognitive judgment affect results?; (2) Is the metacognitive task clearly and consistently defined, and dissociated from the primary cognitive task?; (3) Is the metacognitive rating scale feasible for patients with cognitive impairment?; and (4) Does the primary cognitive task allow for a range of metacognitive judgments?