In the context of AD, a few studies have explored the issue of perspective-taking with regard to metacognitive abilities by asking patients with AD to evaluate the performance of their relatives. The results indicated that AD patients were generally accurate in predicting their caregiver’s performance, despite difficulties in evaluating their own performance. These results were attributable to deficits in the patients’ ability to update their knowledge of their own cognitive functioning (i.e., a monitoring deficit in the metacognitive process that is restricted to the self) (McGlynn and Kaszniak, 1991; Kaszniak and Zak, 1996). However, in a review of the neuropsychological bases of metamemory, which is a subtype of metacognition looking specifically at the control and monitoring of memory performances, Kaszniak and Zak (1996) argued that such an interpretation of these findings may be problematic because cognitive abilities in the patients’ relatives did not decline as sharply. Therefore, the patients may be able to make accurate evaluations of their caregivers’ performance even when relying on out-of-date knowledge of their relatives’ cognitive capacities. To resolve this issue, Duke et al. (2002) used a combination of two methods to assess awareness: predictions and post-dictions regarding performance and discrepancies between patient and informant reports. AD patients and their spouses were asked to estimate each other’s performance and the performance of a fictional patient who suffered from moderate cognitive impairment. The estimations were made before and after performing a verbal memory task and verbal fluency task. Before making the predictions, the participants were given normative data based on their own age range, their spouse’s age range, and the fictional patient’s age range. The results showed that AD patients overestimated themselves but made accurate predictions and post-dictions of their spouse’s performance. With regard to evaluating the fictional patient, both AD patients and their caregivers overestimated performance. These results, however, did not allow the determination of whether AD patients’ difficulties in awareness were self-specific or global. However, by comparing the predictions and post-dictions, the authors suggested that some monitoring abilities, in an on-line fashion, might be preserved in AD patients. Indeed, the results showed that they reduced the extent of overestimating their abilities after completing the task. To date, this is the only study that has compared the ability of AD patients to evaluate the performance of a well-known person (e.g., spouse) and an unknown person (e.g., fictional patient).

Using a vignette technique as an indirect method to assess awareness, Clare et al. (2012a) explored the ability of PwD (including AD, vascular dementia and mixed Alzheimer’s and vascular dementia) to appreciate the deficits of a fictional person. PwD, their caregivers, and older adult controls were asked to identify and offer advice for the problems that were described in the vignettes. Vignettes were created and described a short scenario of a person who presented a typical case of advanced or early stage dementia or a healthy older adult. The results showed that, when presented with these vignettes, the participants were able to correctly identify and offer appropriate advice for the problem that was described in the vignette. Additionally, some of the participants in the dementia group, even some who presented a limited level of awareness, identified themselves spontaneously with the situation that was presented in the vignette. These results led the authors to suggest that the vignette method (i.e., a third-person perspective method) could be useful to help PwD acknowledge their own deficits.

Based on the evaluation of a third person, Mograbi et al. (2014) investigated the attribution of difficulty for the self and for others. Patients with AD were asked to judge how difficult a task was for them and would be for someone else their own age. Two experiments were conducted using computerized tasks. Experiment 1 used reaction time tasks, and Experiment 2 used memory tasks. The level of task difficulty was controlled, leading to one easy and one difficult task in each experiment. In both experiments, the authors showed that AD patients predicted that others would perform as well as themselves, despite their cognitive deficits. This study also explored correlations with premorbid personality and showed that higher neuroticism and agreeableness were associated with the attribution of more difficulty for the self and less difficulty for others. These findings were interpreted by the authors as the poor ability of people with AD to put themselves in someone else’s position. The authors explained that the different findings between their results and Clare et al. (2012a) were attributable to methodological differences. Indeed, in Mograbi et al. (2014), the cognitive demand of the task, where participants had to imagine themselves as someone else, was higher than in Clare et al. (2012a), in which patients did have access to written vignettes. Altogether, these results suggest that patients with AD are more accurate when evaluating the performance of someone else (e.g., caregiver or fictional person) than they are for themselves. Nevertheless, the methods that were employed in these studies had distinctly different features (e.g., known vs. unknown person, vignettes vs. actual observation, prediction of performance vs. judgment of difficulty level), thus limiting synthesis of the findings. Moreover, to date, no study has explored the impact of self-evaluation through a third-person perspective, using for example video self-observation, on awareness level in AD patients. Further research is needed to better understand the impact of perspective-taking on the awareness of deficits in dementia. Additionally, possible symptoms, such as a deficit in familiarity recognition for self and other faces, should also be considered in studies investigating the impact of perspective-taking in dementia (Van den Stock et al., 2012, 2013). Finally, it is important to acknowledge the differences in the underlying mechanisms of metacognitive deficits in different types of dementia, such as AD and FTD (Salmon et al., 2008; Rosen, 2011) and, therefore, to take those into account when exploring self vs. other-information processing in dementia.

As suggested by Morris and Mograbi (2013), the behavioral evidence may reflect the existence of different networks that are involved in self/other appraisal. The authors included this notion in a reformulated version of the CAM (Agnew and Morris, 1998; Morris and Hannesdottir, 2004; Hannesdottir and Morris, 2007; Morris and Mograbi, 2013), which is a neurocognitive model of anosognosia, accounting for the heterogeneous presentation of disordered awareness, through domain-specific modules (see Figure 1). In fact, they proposed different memory records for self- and other-information. First, the Personal Database, which contains semanticized representations about the self, and the Autobiographical Conceptual Memory System, which contains lifetime knowledge concerning experienced events, allow the appraisal of self-evaluation. Both are relying on personal semantics, which is acquired through personal, social, and cultural experiences. Second, these memory systems are separated from the Generic Memory System, which stores other material that permits evaluations of others and is based on general semantic knowledge.

Recent neuroimaging evidence supports the idea of different neural bases for processing self vs. non-self information. For example, Ruby and Decety (2001, 2003, 2004) explored the neuroanatomical correlates of perspective-taking regarding different domains: motor, conceptual, and emotional. The authors showed that, for these three domains, the frontopolar, somatosensory, and inferior parietal cortices are implicated in the process of self vs. other distinction. As underscored by Ruby et al. (2009), some of these brain areas, such as the orbitofrontal cortex and temporoparietal junction, have also been shown to be related to measures of anosognosia in AD. Additionally, these brain regions are partially overlapping with those that comprise the default mode network (DMN). Indeed, the DMN, activated when participants are not engaged in a specific cognitive task, includes the frontal, parietal and temporal regions (for review see Buckner et al., 2008). Moreover, this network has been shown to be affected early by AD neurodegenerative process (for review see Mevel et al., 2011) and, as suggested by Bond et al. (2016), the impairment of the DMN by AD neurodegeneration could explain the differences in metacognitive abilities depending on the perspective through which the information is presented. However, to date, no neuroimaging study has specifically explored the issue of perspective-taking (first- vs. third-person perspectives) regarding metacognitive processes in AD.

The findings regarding the relationship between metacognition and perspective-taking may have important clinical implications. In fact, studies have shown that the lack of awareness is related to higher engagement in risk situations, earlier institutionalization, more difficulties with treatment adherence, and greater caregiver burden, all of which increase the financial impact on society (Starkstein et al., 2007; Arlt et al., 2008; Bertrand et al., 2013; Horning et al., 2014). Interventions that increase awareness may eventually bring important clinical benefits. Throughout this review, we highlighted the possibility of using third-person perspective paradigms to do so. This type of intervention has already been successfully used with anosognosic patients with neurological or psychiatric disorders, such as hemiplegia after stroke or schizophrenia (Vikram et al., 2008; Fotopoulou et al., 2009; David et al., 2012; Besharati et al., 2015). However, when using interventions to improve awareness, caution needs to be taken with the patient’s emotional state. Studies have shown that awareness regarding deficits in AD may lead to negative mood changes, including depressive symptoms (Harwood et al., 2000; Clare et al., 2012b). Thus, further efforts need to be made by clinicians to acknowledge changes in their patients’ mood states during interventions that attempt to decrease the level of anosognosia.