Anosognosia, autobiographical memory and self knowledge in Alzheimer's disease

Abstract

This article explores the relationship between lack of awareness of neuropsychological deficit, also termed anosognosia, and loss of self knowledge in Alzheimer's disease (AD). Specifically, it considers the hypothesis that anosognosia in AD can in part be explained by a loss of mnemonic ability in which knowledge about self-ability is degraded. To ground this hypothesis, we review evidence suggesting failure to update personal knowledge concerning task efficacy, loss of recollection with relative amplification of semanticization processes and loss of an updated representation of the self. We present a theoretical formulation as to how the features of memory impairment in AD may contribute to anosognosia, incorporating these notions in a reformulation of the Cognitive Awareness Model.

Introduction

Lack of awareness about cognitive deficits or illness is a conspicuous feature of Alzheimer's disease (AD) and can be considered to be a direct result of the disease process (Hannesdottir and Morris, 2007), although moderated by psychosocial factors (Clare et al., 2012a, 2012b). Hence, in the context of a neurobiologically determined disorder, loss of awareness in AD can reasonably be given the term ‘anosognosia’ following Babinski's (1914) original application in relation to lack of knowledge of neurological disease. Anosognosia in AD has been shown to affect multiple facets of cognition and behaviour, including not only perceptions of illness but awareness of deficit relating to cognitive functions, such as memory, executive functioning, language and social cognition, and also judgements about behavioural efficacy, including self care ability, social interaction, driving and instrumental activities of daily living (Clare et al., 2012a; Morris and Hannesdottir, 2004; Nelis et al., 2011). In other words, unlike some other forms of anosognosia, such as that found for example in cortical blindness (Anton, 1898) or hemiparesis (Babinski, 1914), the main feature in AD is the lack of specificity of anosognosia in relation to overall function. People with AD almost universally underestimate the presence of neuropsychological deficit, for example, minimising difficulties or attributing changes to normal ageing.

Techniques for measuring anosognosia in AD have enabled quantification of severity levels, which in turn have proved useful for experimental investigation. One approach is directly appraising levels of awareness of illness or cognitive deficit by interviewing the person with AD (e.g., Hannesdottir and Morris, 2007; McGlone et al., 1991; Reed et al., 1993). This approach can explore, for example, beliefs about illness status, which can be independent of awareness of neuropsychological impairment. A second approach is to use a questionnaire (or structured interview) in which the person with dementia rates their everyday abilities or symptoms and this is compared to a parallel informant rating. In this case, the questionnaire can be used to target particular aspects of awareness, for example, memory functioning (Morris and Hannesdottir, 2004). Although subjective memory complaints concerning cognitive dysfunction, including memory, are relatively common in older people (Jonker et al., 2000), a main finding in AD, even in the early stages, is the tendency to overestimate ability in comparison to informants; discrepancies between the two sources of information, termed the subjective rating discrepancy (SRD) (Hannesdottir and Morris, 2007), or lack of correlation quantifies the extent of anosognosia. Finally, experimental studies can be conducted in which the person with AD is provided with a task likely to result in failure and have to judge their performance, either by making a prediction or estimating performance on completion. The difference between performance and judgement gives rise to an objective-judgement discrepancy (OJD) measure. This approach measures the more immediate response to failure, and is potentially useful in determining the extent to which online error detection is compromised in AD.

Of note, whatever method is used, group studies of AD detect overall loss of awareness of illness or deficit, even if the results from the different techniques are not necessarily highly correlated (Hannesdottir and Morris, 2007). Such approaches have shown that anosognosia occurs in the early stages of AD (Clare et al., 2012b), and even in the prodromal stages, as in mild cognitive impairment (MCI), although there is evidence for variability in the latter group, with some patients showing overestimation of subjective memory complaints, and other evidence for underestimation of memory impairment (Jungwirth et al., 2004; Roberts et al., 2009; Tremont and Alosco, 2011; Vogel et al., 2004), such that reliance on subjective complaints of symptoms may be suspect with regards to diagnosis. Longitudinal studies suggest that in early AD levels of anosognosia are relatively stable (Clare et al., 2012b), with evidence that with further increases in the severity of dementia anosognosia becomes more prominent but in a non-linear fashion (e.g., Aalten et al., 2006; Mograbi et al., 2012a).

Theoretical accounts of anosognosia in AD are likely to be complex and take into account the different processes associated with awareness. These include, for example, monitoring the level of success or failure on everyday tasks, translating the outcome of monitoring into decisions about current function and then further abstraction relevant information into more permanent ‘self-belief.’ With AD, different accounts have been produced, including those suggesting that lack of awareness is due to a failure of monitoring and control processes (e.g., Cosentino et al., 2007; Souchay, 2007), the effects of emotional dysregulation when engaging in self-monitoring (Rosen, 2011), impaired decision-making processes (e.g., Agnew and Morris, 1998; Morris and Hannesdottir, 2004), or due to the inability to consolidate new personal information (Morris and Hannesdottir, 2004). Recently, the ‘Petrified Self’ hypothesis has been developed by Mograbi et al. (2009) in which it has been proposed that unawareness in AD may be differentially associated with specific aspects of memory function, tapping into episodic/semantic and remote/recent memory distinctions.

In this paper we propose an elaboration of the hypothesis in which anosognosia is caused primarily by a decline in specific mnemonic processes leading to a loss of personal knowledge. Central to this hypothesis is the notion that consolidation of personal information is a pre-requisite for the development of self-identity over the lifespan. Here, awareness of functional ability is considered to be a special case of personal trait judgement that concerns personal efficacy. Furthermore, it is proposed that awareness involves the shaping of individual knowledge and behaviour, with continual modification to maximise personal efficacy and minimise experience of failure (Morris and Hannesdottir, 2004). In other words, the mechanisms that facilitate awareness of functional ability should be considered developmentally, invoking memory and acquired representation, and teleonomically, invoking purposefulness.

As a cause of anosognosia, impairments in specific aspects of memory function may prevent the updating of the self-concept and result in the person not consolidating information concerning their reduction in neuropsychological function (Mograbi et al., 2009). In connection with this, Conway (e.g., Conway, 2005; Conway and Pleydell-Pearce, 2000) suggests the existence of a “working self”, which modulates encoding of new information based on goals and self-image. Here, memory provides continuity to the experience of selfhood by allowing storage of past information and the ability to project future scenarios (Addis and Tippett, 2004). It constraints what the self is, has been and can be (Conway, 2005). Evidence from the study of brain damaged patients has suggested that profound impairment in mnemonic function clearly result in alterations in identity and selfhood, but with relatively preserved personality (Corkin, 2002; Klein and Nicols, in press; Rosenbaum et al., 2005; Tulving, 1999).

Such research indicates that different forms of memory give specific contributions to the formation of a self-concept. For example, there is substantial evidence that trait judgements about self or others are based on semantic knowledge extracted from episodes or life experiences (Klein et al., 2002a, 2008). Furthermore, semanticized traits facilitate more rapid access than if judgements were to be made on the basis of episodic memory. Nevertheless, there is evidence that trait summaries are updated in an accumulative fashion, with updating associated with each new episode. The nature of trait judgement may depend on the amount of experience of a person, with cursory judgement based on recent behavioural examples and more extensive experience leading to more abstracted trait summaries (Sherman and Klein, 1994). In relation to the development of self knowledge, the long time span of experience suggests self trait summaries are not only strongly semanticized as personal semantic knowledge but through a lengthy neural consolidation process (Tulving, 1985a, 1985b; Tulving et al., 1988). Nevertheless, episodic memory also contributes to a sense of self, allowing autonoetic consciousness, time and travel and re-experiencing of details. The continuity and richness of self experience (i.e., the ability to integrate specific and detailed past and present selves) relies on episodic autobiographical memory, and recent evidence suggests this is one of the last cognitive abilities to appear during development and the first to be lost with ageing and amnesic conditions, such as in AD (Piolino et al., 2006).

In summary, the formation of self knowledge may start with recollective experiences and to be followed by semanticisation and this process can be considered in relation to self knowledge concerning personal efficacy, with semanticized trait information enabling a person to make predictions about task performance (Nelson and Narens, 1990) or facilitate evaluation of current performance in relation to past experience, including recalibrating future expectations (Moulin et al., 2000a).

To illustrate these processes as they might apply to awareness of failure, Table 1 provides schematic material relating to everyday activity, providing examples of individuals experiencing task failure. For the purposes of illustration, the protagonist fails in the activities, as might occur in AD, but the awareness processes are assumed to be working properly so as to indicate their mechanism. Take the car parking illustration as the example: the first stage is the experience of task failure in a familiar situation which is adequately monitored and gives rise to the realisation of failure. This is accompanied by specific memories that later on can be retrieved for further consideration in which further nuancing of the situation might occur by evaluating what happened in the particular context (the circumstances associated with failure to find the car park). The memory for the experience is then semanticised into a general understanding, in which with repeated negative experience personal knowledge about car parking experiences is coded more abstractly and then crystallised into a self-concept ‘I am now not so good at navigation.’ Note that these processes can occur in a reverse fashion with positive experiences or mastery of a situation, revising expectations upwards. In AD it is claimed in this paper that impairments in these consolidation processes in part give rise to loss of this recalibration and hence unawareness of neuropsychological deficit.

The mechanisms that support such processes have already been considered to some extent in an existing framework used to investigate anosognosia in AD, namely the Cognitive Awareness Model (CAM Fig. 1; Agnew and Morris, 1998; Hannesdottir and Morris, 2007; Morris and Hannesdottir, 2004; see also Ansell and Bucks, 2006). The essential features of this model are that incoming knowledge concerning task or activity performance is monitored by comparator mechanisms that contrast this information to that stored in a Personal Database (PDB), a specialised storage of information about current function. The output of the comparison leads to updating of the PDB if there is a mismatch between the two sources of data and the resulting information is released via a further mechanism, namely the Metacognitive Awareness System (MAS) to provide consciousness of decision making. In addition to sparing of error monitoring and detection, awareness can only be generated in the presence of multiple preserved memory abilities. For example, immediate error monitoring would depend on abilities such as iconic and working memory to hold information relating to failure for sufficient time to allow error evaluation. A core feature of this model is the PDB, which retains information about performance efficacy and is continually updated by experience; experiences of success or failure relating to personal efficacy should lead to a recalibration of information contained in the PDB (Hannesdottir and Morris, 2007). Furthermore the PDB is moderated by social and interpersonal influences, including aspects not considered in this article (see: Clare et al., 2012a). Previously we have argued that a range of neurocognitive factors relating to different facets of the model can produce anosognosia in AD, those concerned with memory or consolidation aspects giving rise to what we have termed mnemonic anosognosia (Agnew and Morris, 1998; Hannesdottir and Morris, 2007; Morris and Hannesdottir, 2004).