Theory of Mind Deficits and Awareness in Alzheimer’s Disease

Abstract

Introduction

Social cognition (SC) is often impaired by dementia, contributing to difficulties with relationships and social isolation. Additionally, people with dementia frequently have impaired awareness of their impairments, which may worsen distress for them, their family, and friends. We investigated SC in people with Alzheimer’s disease (PwAD), awareness of SC, and whether PwAD overestimate SC abilities compared to healthy controls (HC).

Methods

Participants completed the Interpersonal Reactivity Index perspective-taking subscale (IRI-PT), a self-report measure of SC, and The Awareness of Social Inference Test (short form) (TASIT-S), a task-based measure of SC. Insight was evaluated by comparing self-assessment with TASIT-S performance. The Addenbrooke’s Cognitive Examination (version 3; ACE-III) measured general cognition.

Results

We recruited 70 PwAD (mean age = 78.13; mean ACE-III = 71.80) and 17 HC (mean age = 71.53; mean ACE-III = 96.65). PwAD self-reported lower theory of mind (ToM) (IRI-PT = −2.16, 95% CI = −0.33 to 4.64, p = 0.09) and performed worse on TASIT-S emotion recognition (−2.04, 95% CI = −2.89 to −1.20, p < 0.001), sarcasm recognition (−0.91, 95% CI = −1.66 to −0.16, p = 0.02), and ToM assessments (−1.42, 95% CI = −2.27 to −0.57, p = 0.001). There was no significant relationship between IRI-PT and TASIT-S ToM in either group, nor in the participant group as a whole. We categorised “under-estimators” (low IRI-PT, high TASIT-S ToM), “over-estimators” (high IRI-PT, low TASIT-S ToM), and “congruent-estimators” and found that lower ACE-III was associated with ToM over-estimation.

Conclusion

Differences in self-reported and task-assessed SC were observed between PwAD and HC, but self-estimated and task-assessed ToM were not associated, suggesting that SC self-estimation is unreliable in PwAD and possibly in healthy older adults.

Introduction

Social cognition (SC) describes neurocognitive skills, which enable and preserve functioning in social groups [1], including emotion recognition (ER), the ability to identify others’ emotions through facial expressions, gestures, and voices, and theory of mind (ToM), the ability to understand others’ thoughts, beliefs, and intentions even when they differ from one’s own [2] and can be affected by dementia [3]. These skills enable interpretation of social cues, which can be used to make inferences about others’ thoughts, feelings, and intentions and to appropriately tailor one’s own behaviours. SC skills support people to build relationships and reduce social isolation, with beneficial effects on cognition [4], mental health [5], and carer burden [6]. Though most frequently associated with behavioural variant frontotemporal dementia [7], it is increasingly understood that SC changes also occur in other dementia variants, including Alzheimer’s disease (AD) [8–10]. There is some evidence that ToM deficits may occur in mild AD; however, other studies have found no evidence of impairment of ToM. This may reflect more complex, higher order ToM reasoning being vulnerable to impairment, whereas performance on basic ToM tasks may be preserved in mild AD. This raises the question of whether the more “higher order” process of awareness of changes to one’s ToM is impaired in people with mild AD [8, 11, 12].

People with dementia may lack awareness of their cognitive and functional decline, with impaired awareness sometimes presenting in the early stages of disease [13]. Lack of awareness is linked to increased risk of progression from mild cognitive impairment to dementia; lowered treatment adherence as it may affect a person’s ability to seek or accept help, comply with treatment, or plan for the future; and increased burden and distress for caregivers [14–17]. Understanding social cognitive impairment and related deficits of awareness may inform future approaches to support patients and their families to understand and adapt to changes. One approach to assessing awareness is to compare subjective beliefs about performance with objective task performance. The level of discrepancy between subjective and objective is interpreted as a measure of metacognition [18]. In this study, we used the perspective-taking subscale of the Interpersonal Reactivity Index (IRI-PT [19]; as a measure of self-reported ToM) and The Awareness of Social Inference Test (short form, TASIT-S [20]; as an objective ToM assessment).

We compared domains of SC between people with AD (PwAD) and healthy controls (HC). We compared participants’ self-reported and task-assessed ToM. Finally, we examined whether PwAD and HC over- or under-estimated their ToM and whether there were group differences in these estimations.

Methods

Study Design and Subjects

This cross-sectional study included 87 participants. PwAD were recruited through National Health Service (NHS) memory clinic referrals (n = 58) and Join Dementia Research (JDR; [21]; n = 12). HC were recruited exclusively from JDR (n = 17), a Website supported by the National Institute for Health and Care Research enabling people with dementia and healthy volunteers to register interest in participating in research. Study protocols were approved by NHS Wales Research Ethics Committees 6 (23/WA/0157) and 3 (17/WA/0416).

Inclusion/Exclusion Criteria

All participants had capacity to give informed consent to participate, were able to complete testing via a computer or tablet, and had no medical or psychiatric illness that could interfere with participation or compromise their safety. PwAD received a clinician-confirmed diagnosis of probable dementia caused by AD, as part of their standard UK memory clinic care, where diagnoses are made using the International Statistical Classification of Diseases and Related Health Problems, version 10 (ICD-10) [22]. HC had no diagnosis of cognitive impairment or dementia.

Procedure

Participants were given written information and at least 24 h to consider participation before asking questions and giving informed consent. Data were collected either remotely, due to COVID-19 restrictions, via an online research data collection Website (gorilla.sc) facilitated by a researcher, or face to face, with data entered into gorilla.sc by a researcher.

Measures

Social Cognition

The Awareness of Social Inference Test (short form) (TASIT-S [20]) is a clinically sensitive measure of SC using a series of video-recorded vignettes presented in three parts. Part 1 measures ER, asking participants to identify in ten short videos (15–60 s each) the dominant emotion of the actor in the scene (happy, sad, angry, anxious, surprised, revolted, or neutral). Scripts are ambiguous. In some scenes, there is one actor, talking directly to the camera. In others, there are two actors, and the participant is told to focus on one of them. We used participant’s number of correct part 1 responses as a measure of ER (minimum score = 0, maximum score = 10).

Parts 2 and 3 assess social inference, reflecting ToM. They probe for participants’ understanding of more complex emotions and what the actors know, believe, and intend. In parts 2 and 3, short video vignettes (15–60 s each) are conversational exchanges between two or more characters. Each video is accompanied by four questions probing inferences of what the characters are doing, trying to say, thinking, and feeling. There is a mixture of sincere and sarcastic exchanges, though the scripts and scenarios are ambiguous such that the sincerity or sarcasm must be interpreted from the speakers’ demeanour.

We used responses to the “do” questions from part 2, e.g., “Is he seriously suggesting they invite other family members?” as they most specifically assess participant understanding of sarcasm vs. sincerity [23]. These are considered a measure of second-order cognitive ToM, and in our study we refer to this as “sarcasm recognition” (SR). We included responses to questions for the sarcastic and sincere videos to assess whether participants believed they were recognising sarcasm, whether or not it was present. Part 2 is preceded by a “practice” item, but since this item was presented in the same manner as the subsequent nine vignettes, participant responses to that item were included in our analyses of SR, meaning there was a minimum possible score of 0 and a maximum possible score of 10.

We used the “think” question responses from the nine items in part 3, which investigate participants’ understanding of the thoughts and knowledge of others, e.g., “Does Ruth know that Rosie scribbled in the book?,” to measure first-order cognitive ToM [23]. The part 3 videos include additional cues to elucidate the perspectives of the characters, meaning that viewers are party to information that not all characters’ hold, with this information about the “true” state of affairs presented either before or after the key exchange of dialogue between characters; however, the practice item was not included in the analysis due to the lack of the “additional cues” in the practice vignette. The minimum possible score for this section was 0, and the maximum possible score was 9.

All participants completed the Interpersonal Reactivity Index (IRI; [19], a self-report measure that prompts participants to rate their social cognitive skills using a Likert scale scored from 1 to 5) (“A does not describe me very well” to “E describes me very well”). The IRI has high internal validity and test-retest reliability [24]. For this study, we used the seven-item perspective-taking (PT) subscale, which assesses the extent to which an individual reports seeing the world from other people’s perspectives, representing a self-report measure of ToM. The possible range of scores was 7–35.

General Cognition

All participants completed the Addenbrooke’s Cognitive Examination (version 3; ACE-III [25]) measuring general cognitive ability (attention, memory, fluency, language, and visuospatial abilities) with scores ranging from 0 (maximum impairment) to 100 (no impairment). In line with remote ACE-III guidance [26], we presented stimuli via video call and any required assistance was given locally.

Other Variables

Sociodemographic information collected from participants included age, sex, ethnicity (white; Asian or Asian British; black, African, Caribbean or black British; mixed or multiple ethnicities; other), and education (years).

Statistical Analyses

We described sociodemographic characteristics, IRI-PT score, and TASIT-S performance in PwAD and health controls and compared group differences using Hedge’s g for continuous variables and Cramér’s V for categorical variables. We then used linear regression to compare IRI-PT and TASIT-S domains between PwAD and HC. We then used linear regression to investigate relationships between IRI-PT and TASIT-S domains and ACE-III scores, first for the whole participant cohort and then separately in PwAD and HC. These regression models were first unadjusted, then adjusted sequentially to account for age and sex and education. Coefficients derived from these models indicate the number of IRI-PT or TASIT-S points higher for HC compared to PwAD and for each 10 points higher ACE-III score. We also present standardised coefficients for fully adjusted regression models.

We described characteristics of participants who had scores deviating from both IRI-PT and TASIT-S ToM means by more than 0.33 of a standard deviation (informed by previous estimates of minimal clinically important differences in PwAD) [27]. High TASIT-S ToM performance and low IRI-PT self-estimation constituted “under-estimators,” low TASIT-S ToM and high IRI-PT constituted “over-estimators,” and non-outlying participants were categorised as “congruent-estimators.” Statistical analyses were run using STATA version 18.0.

Results

Sample Characteristics

We recruited 87 participants (44 females, 43 males), aged 57–98 (mean = 76.84, SD = 8.23). Of these, 70 were PwAD (34 females, 36 males), mean age 78.13 years (SD = 7.45), and 17 were HC, mean age 71.53 years (SD = 9.35). Mean ACE-III scores for PwAD were 71.80 (SD = 12.81) and for HC participants 96.65 (SD = 3.26). Participants were predominantly from white ethnic backgrounds (94%) (see Table 1 for participant demographic information).

Table 1.

Sociodemographic information for all participants and descriptive data for all participants performance on IRI-PT and TASIT-S domains

​	All participants (N = 87)			PwAD (N = 70)			HC (N = 17)			Between-Group differences and effect sizesa
	mean	SD	range	mean	SD	range	mean	SD	range
Age, years	76.84	8.23	57–98	78.13	7.45	57–98	71.53	9.35	57–92	t(85) = −3.11, p = 0.00–0.83 [−1.37, −0.29]
Education, years	15.00	3.09	6–20	14.33	2.93	6–20	17.76	2.05	13–20	t(85) = 4.57, p = 0.00, 1.22 [0.66, 1.78]
Gender, n (%)
Females	​	44 (51)	​	​	34 (49)	​	​	10 (59)	​	χ2(1, N = 87) = 0.58, p = 0.45
Males	​	43 (49)	​	​	36 (51)	​	​	7 (41)	​	0.08
Ethnicity, n (%)
White	​	82 (94)	​	​	65 (93)	​	​	17 (100)	​	χ2(3, N = 87) = 1.29, p = 0.73
Asian or Asian British	​	2 (2)	​	​	2 (3)	​	​	0	​
Black, African, Caribbean, or black British	​	2 (2)	​	​	2 (3)	​	​	0	​	0.12
Mixed or multiple ethnicity	​	1 (1)	​	​	1 (1)	​	​	0	​
ACE-III scores (total)	76.66	15.23	35–100	71.80	12.81	35–96	96.65	3.26	89–100	t(85) = 7.90, p = 0.00, 2.12 [1.50, 2.73]
IRI – PT	25.21	4.67	8–33	24.79	4.71	8–33	26.94	4.21	17–33	t(85) = 1.72, p = 0.09, 0.46 [–0.07, 0.99]
TASIT-S – emotion recognition (sum score)	4.89	1.76	1–8	4.49	1.68	1–8	6.53	0.94	5–8	t(85) = 4.81, p = 0.00, 1.29 [0.73, 1.85]
TASIT-S – sarcasm recognition (“do” questions sum)	7.21	1.43	2–10	7.03	1.40	2–10	7.94	1.34	6–10	t(85) = 2.42, p = 0.02, 0.65 [0.11, 1.18]
TASIT-S – ToM (“think” questions sum)	6.15	1.67	2–9	5.87	1.61	2–9	7.29	1.45	3–9	t(85) = 3.34, p = 0.001, 0.89 [0.35, 1.43]

Social cognition differences between PwAD and HC and by overall cognitive impairment.

ACE-III, Addenbrooke’s Cognitive Examination (version 3); SD, standard deviation; IRI-PT, Interpersonal Reactivity Index (perspective-taking subscale); TASIT-S, The Awareness of Social Inference (short version).

^aEffect sizes are reported as Hedge’s g for continuous variables and Cramér’s V for categorical variables.

Group Differences

IRI-PT and TASIT-S scores were compared for each group (Table 1). Results of linear regression analyses for all participants are shown in Table 2. For each model, unstandardised regression coefficients with 95% confidence intervals are reported. These coefficients represent the effect sizes, quantifying the magnitude of association between predictors and outcomes.

Table 2.

Relationship between dementia status and overall cognition and self-reported and task-assessed social cognitive domains

​	​	​	Model 1 unadjusted		Model 2 adjusted for age and sex		Model 3 + adjusted for education
			coefficient	95% CI, p value	coefficient	95% CI, p value	coefficient	95% CI, p value	standardised coefficient	95% CI, p value
IRI-PT (sum)	HC vs. PwAD	​	2.16	−0.33 to 4.64, p = 0.09	2.03	−0.54 to 4.60, p = 0.12	1.69	−1.12 to 4.50, p = 0.23	0.14	−0.08 to 0.37, p = 0.21
TASIT-S ER (sum)		​	2.04	1.20 to 2.89, p < 0.001	1.73	0.85 to 2.61, p < 0.001	1.73	0.77 to 2.70, p = 0.001	0.39	0.20 to 0.58, p < 0.000
TASIT-S SR (“do” question sum)		​	0.91	0.16 to 1.66, p = 0.02	0.48	−0.27 to 1.23, p = 0.20	0.20	−0.60 to 1.00, p = 0.62	0.06	−0.16 to 0.27, p = 0.60
TASIT-S ToM (“think” question sum)		​	1.42	0.57 to 2.27, p = 0.001	0.89	0.06 to 1.72, p = 0.04	0.46	−0.42 to 1.34, p = 0.30	0.11	−0.09 to 0.31, p = 0.28
IRI-PT (sum)	Per 10-point higher ACE-III score (total participants, then in PwAD and HC)	Total	0.31	−0.34 to 0.97, p = 0.34	0.49	−0.4 to 1.19, p = 0.16	0.39	−0.37 to 1.16, p = 0.31	0.13	−0.11 to 0.36, p = 0.29
		PwAD	−0.13	−1.01 to 0.76, p = 0.78	0.19	−0.77 to 1.14, p = 0.70	0.13	−0.86 to 1.11, p = 0.79	0.04	−0.22 to 0.29, p = 0.79
		HC	2.22	−4.78 to 9.21, p = 0.51	0.52	−6.80 to 7.84, p = 0.88	0.54	−7.16 to 8.24, p = 0.88	0.04	−0.41 to 0.49, p = 0.86
TASIT-S ER (sum)		Total	0.62	0.43 to 0.82, p < 0.001	0.58	0.37 to 0.79, p < 0.001	0.64	0.40 to 0.88, p < 0.001	0.57	0.40 to 0.73, p < 0.000
		PwAD	0.51	0.23 to 0.78, p = 0.001	0.53	0.25 to 0.81, p < 0.001	0.57	0.28 to 0.87, p < 0.001	0.47	0.27 to 0.66, p < 0.000
		HC	−0.11	−1.70 to 1.48, p = 0.89	0.16	−1.66 to 1.99, p = 0.85	0.18	−1.71 to 2.07, p = 0.84	0.06	−0.44 to 0.56, p = 0.82
TASIT-S SR (“do” question sum)		Total	0.25	0.05 to 0.44, p = 0.01	0.15	−0.05 to 0.36, p = 0.14	0.09	−0.13 to 0.30, p = 0.44	0.09	−0.13 to 0.31, p = 0.44
		PwAD	0.17	−0.09 to 0.43, p = 0.20	0.10	−0.17 to 0.38, p = 0.46	0.03	−0.25 to 0.31, p = 0.83	0.05	−0.19 to 0.29, p = 0.68
		HC	1.22	−0.96 to 3.39, p = 0.25	0.23	−1.97 to 2.43, p = 0.83	0.25	−2.02 to 2.52, p = 0.82	0.05	−0.37 to 0.47, p = 0.81
TASIT-S ToM (“think” question sum)		Total	0.38	0.17 to 0.59, p < 0.001	0.28	0.06 to 0.49, p = 0.01	0.20	−0.04 to 0.43, p = 0.10	0.19	−0.01 to 0.40, p = 0.06
		PwAD	0.25	0.03 to 0.53, p = 0.07	0.22	−0.08 to 0.51, p = 0.15	0.17	−0.14 to 0.48, p = 0.28	0.17	−0.07 to 0.40, p = 0.16
		HC	1.46	−0.85 to 3.77, p = 0.20	0.03	−1.90 to 1.97, p = 0.97	0.00	−1.87 to 1.87, p = 1.00	−0.01	−0.34 to 0.32, p = 0.96

General cognition coefficients represent mean TASIT-S domain score (ER, SR, or ToM difference per 10 points higher ACE-III score.

No association between general cognition or dementia status and self-reported perspective taking (IRI-PT) was found in fully adjusted models. For task-assessed ER, dementia status and higher ACE-III score were associated with better TASIT-S ER performance for all participants and PwAD. Adjusting for age, sex, and education, scores on TASIT-ER were 0.57 higher (95% CI: 0.28 to 0.87, p < 0.001) in PwAD per 10-point higher ACE-III score. HC scored 0.18 points higher (95% CI: −1.71 to 2.07, p = 0.84) on the TASIT-ER per 10-point higher ACE-III score.

There was no association between dementia status or general cognition and TASIT-S SR performance in adjusted models. Differences in TASIT-ToM performance were associated with general cognition in model 2, adjusted for age and sex, for all participants (b = 0.28 (95% CI: 0.06 to 0.49), p = 0.01), but these associations were no longer significant when also adjusting for education (b = 0.20 (95% CI: −0.04 to 0.43), p = 0.10).

Awareness of ToM in PwAD and HC

Trend lines for participant’s IRI-PT and TASIT-S ToM show modest linear relationships between the tasks for each group (Fig. 1), with correlation coefficients of 0.10 for PwAD and 0.05 for HC. We categorised participants into three groups: over-estimators (n = 5), who scored more than 0.33 SD higher than average on the self-evaluated ToM and 0.33 SD lower than average on task-evaluated ToM; under-estimators (n = 7), who had a reverse pattern to the over-estimators; and congruent estimators (n = 75), whose self- and/or task-evaluated ToM scores were within the average SDs observed for the group (for demographic characteristics of these groups, see Table 3). The mean ACE-III score of over-estimators was 62.60 (SD = 19.02), whereas for under-estimators it was 74.29 (SD = 21.20), and for congruent estimators 77.81 (SD = 14.08).

Fig. 1.

Relationship between self-reported ToM (IRI-PT) and task-assessed ToM (TASIT-S), with test means and 0.33 of standard deviations represented. Participants in upper-left quadrant represent under-estimators, with low self-reported ToM (IRI-PT) but high task performance ToM (TASIT-S ToM). Participants in the lower-right quadrant represent over-estimators, with high self-reported ToM (IRI-PT) but low task performance ToM (TASIT-S ToM). All other participant data points represent congruent estimators.

Table 3.

Demographic information for over-estimators, under-estimators, and congruent estimators

​	Over-estimators (N = 5)	Under-estimators (N = 7)	Congruent estimators (N = 75)
Dementia status
PwAD	4	5	61
HC	1	2	14

​	Mean	SD	Range	Mean	SD	Range	Mean	SD	Range
Age, years
Total group	87.60	9.10	78–98	75.71	7.20	60–81	76.23	7.85	57–90
Education, years
Total group	12.00	3.94	6–17	15.71	2.81	12–20	15.13	2.99	11–20
ACE-III scores (total)
Total group	62.60	19.02	50–96	74.29	21.20	35–100	77.81	14.08	36–100
Gender, n (%)
Females	​	3 (60)	​	​	3 (43)	​	​	38 (51)	​
Males	​	2 (40)	​	​	4 (57)	​	​	37 (49)	​

Discussion

Social cognitive decline can be symptomatic of AD dementia [3, 28] and affect social functioning and social relationships, contributing to isolation and worsened cognition, mental, and physical health [4, 29]. We sought to understand social cognitive differences between PwAD dementia and HC and investigate to what extent PwAD are aware of their social cognitive abilities. Impaired awareness of cognition or function in PwAD can be a feature of disease progression and risks exacerbating strain on social relationships. Limited awareness of decline may contribute to PwAD being resistive to help: they may not see why support is necessary; may misinterpret social contexts, leading to potentially inappropriate behaviours, which may compound issues of caregiver burden, isolation, and stress [4, 6, 14].

We found that PwAD were impaired in ER and ToM compared to HC and that worse overall cognitive impairment was associated with impaired ER in models adjusted for age, sex, and education and that association was found for worse general cognition and impaired ToM in models adjusted for age and sex, although not when additionally adjusted for education. More severe dementia was associated with more severe impairment of some facets of SC. This contrasts with findings from a longitudinal study of people with AD [30], which did not find that cognitive decline was associated with SC impairment, although that study used a carer-report measure of SC, upon which our study improves with a task-based assessment of SC, which also clarifies affected domains. Our finding of impairment in specific SC domains in PwAD compared to HC is supported by some studies [3, 31] and studies comparing PwAD with people with mild cognitive impairment [2]. It is not possible to determine the extent to which these differences are related to general cognitive impairment or a specific social cognitive impairment, although the tasks used to assess SC in this study are designed to be simple to complete, not requiring complex language or memory [32].

We found weak linear relationships between self-reported and task-assessed ToM for PwAD; this relationship was less pronounced in the HC group. The HC sample size was small, with limited distribution of scores in both IRI-PT and TASIT-S ToM. This ceiling effect and lack of heterogeneity may contribute to our not being able to establish a relationship between the tasks for HC. A larger, more variable sample of HC may afford the opportunity to investigate this further. The modest relationships between IRI-PT and TASIT-S ToM scores may reflect the tasks measuring dissimilar constructs. While the IRI-PT subscale measures the extent to which participants feel they have the tendency to try to see the world from the perspective of others, framing of the questions reflects to what extent participants try to “put themselves in other people’s shoes,” which does not necessarily capture how well they feel they understand others’ perspectives. Being a self-report measure, the IRI-PT scale may also be vulnerable to social desirability bias as participants completed the scale with a researcher.

We categorised participants as outliers if they deviated from the mean scores of the IRI-PT and TASIT-S ToM by more than 0.33 of a standard deviation [27], with over- or under-estimators described based upon the directionality of their deviations. We were interested in whether the characteristics of outlying participants could tell us more about what role general cognition may play in mis-estimation of SC. Both PwAD and HC could be over- or under-estimators, although the mean general cognitive score of over-estimators was lower, fitting with the hypothesis that PwAD may over-estimate their ToM more so than HC due to impaired insight. Our over- and under-estimator groups sample sizes were small; thus, conclusions are tentative and require further investigation. However, our study suggests that both cognitively impaired and healthy people can have impaired awareness into their social cognitive abilities, which indicates the potential limitation of self-report measures of SC. Considering that SC research often relies upon such measures, studies with self-report measures should be interpreted cautiously and more research is required to consider factors associated with impaired SC awareness and improve measurement of these domains.

Study findings should be considered in light of its limitations. The study sample lacked diversity, with participants predominantly from white ethnic backgrounds. The TASIT-S is culturally biased toward Australia, whereas this study was conducted in the UK. We had a small sample size of HC, and the HC group was significantly different to the PwAD group with regard to age (t(85) = −3.11, p = 0.00) and education (t(85) = 4.57, p = 0.00). We lacked longitudinal data, precluding examination of changes in SC and awareness over time. Alternative assessments of awareness, such as carer report, may have provided more information. However, this study is novel in directly comparing self-assessment and task-assessment of ToM in this population.

Conclusion

Our findings contribute to understanding the impact of AD on SC, the deterioration of which can be associated with impaired social relationships, increased risk of isolation, and increased caregiver burden. This report highlights differences in PwAD and HC in ER, which may contribute to relational difficulties and caregiver burden [33]. Differences between PwAD and HC were also seen for ToM abilities, though to a lesser degree. This points toward specific areas, which should be targeted through development of information resources and social cognitive training to lessen the impact of social cognitive decline on PwAD and their family and friends. Future research should examine the temporal course of SC and awareness and whether this can be modified through psychosocial intervention.

Statement of Ethics

The studies’ protocols were reviewed and approved by NHS Wales Research Ethics Committee 6 (approval No. [23/WA/0157] and NHS Wales Research Ethics Committee 3 [17/WA/0416]). Written informed consent was obtained from all participants.

Conflict of Interest Statement

The authors have no conflicts of interest to declare.

Funding Sources

This research was funded by the Wellcome Trust (222932/Z/21/Z). A.M. was funded by the Wellcome Trust (203376/2/16/Z) and the Wellcome Centre for Human Neuroimaging (203147/Z/16/Z). For the purpose of Open Access, the author has applied a CC BY public copyright licence to any Author Accepted Manuscript version arising from this submission. A.S. is supported by the University College London Hospitals National Institute for Health Research Biomedical Research Centre. The funding sources had no role in the study design, execution and analysis, and manuscript conception, planning, writing, and decision to publish.

Author Contributions

F.E.R., A.M., J.H., and A.S. conceived and designed the study. F.E.R., A.M., and A.S. collected data. F.R. conducted analyses and drafted the manuscript. All authors reviewed, edited, and approved it.

Funding Statement

This research was funded by the Wellcome Trust (222932/Z/21/Z). A.M. was funded by the Wellcome Trust (203376/2/16/Z) and the Wellcome Centre for Human Neuroimaging (203147/Z/16/Z). For the purpose of Open Access, the author has applied a CC BY public copyright licence to any Author Accepted Manuscript version arising from this submission. A.S. is supported by the University College London Hospitals National Institute for Health Research Biomedical Research Centre. The funding sources had no role in the study design, execution and analysis, and manuscript conception, planning, writing, and decision to publish.

Data Availability Statement

The data that support the findings of this study are not publicly available due to sensitivity of health-related information obtained from patients that could compromise the privacy of research participants but are available from the corresponding author upon reasonable request.