Introduction

The study of social cognition is concerned with the higher-order cognitive processes that allow individuals to interpret the behaviors of others [1]. These abilities allow us to process and understand social information in order to respond appropriately in everyday interactions [2–5]. Social cognition includes abilities such as theory of mind (ToM; i.e., the ability to recognise other people’s mental states to understand and predict their behavior), emotion recognition, empathy, moral judgments and the understanding of social norms [6, 7].

Healthy aging is associated with reliable improvement in emotional well-being [8] and social functioning [9]. Although social network size decreases with age, older adults’ social interactions with individuals who remain within their social networks are rated as being more satisfying [10]. Life experience is thought to influence how people process and respond to social information (e.g., [11]). For example, older adults are thought to be more receptive to emotional cues when making social judgements compared to younger adults (see [12]). Yet, some studies examining individuals’ ability to understand and evaluate relevant social information have reported poorer performance in healthy older adults compared to younger adults [13, 14].

One of the most extensively studied aspects of social cognition in healthy aging is ToM [14]. More recently, it has been argued that ToM is not a one-dimensional concept, but processes differ based on whether they refer to cognitive or affective judgements [15]. Cognitive ToM is defined as the ability to make inferences about the thoughts, intentions and beliefs of another individual. Affective ToM refers to the ability to make inferences about what another individual is feeling [15–17]. Age-related differences have been found where older adults perform more poorly compared to their younger counterparts on tests such as Reading the Mind in the Eyes (RME) [18], Faux Pas stories (a verbal story based test that requires participants to make ToM inferences from short interactions involving social norms violations and non-social norm violations) [19] and Happé’s Strange Stories, among others [20–29]. Yet, other studies have found age-related improvements in favour of older adults such as Happé et al. [29] or equivalent performance between younger and older adults [25, 30–35]. Potentially, age-related differences may be related to one aspect of ToM but not the other, for example cognitive ToM but not affective ToM. However, research into possible dissociations between cognitive and affective ToM has yielded mixed findings. In perspective taking tests which assess cognitive ToM, older adults perform more poorly than younger adults [20, 22, 23, 26, 30, 36–38]. Nonetheless, other authors have failed to find age-related differences in cognitive ToM [30, 31]. Affective ToM has been examined using tests such as the RME [18] where individuals are required to make inferences from the eye region of photographs. Older adults have been found to perform significantly more poorly than younger adults [20, 21, 36, 39, 40]. Video based ToM tests have also shown that older adults perform significantly more poorly than younger adults [28, 40]. However, Castelli et al. [30] and Li et al. [32] have both reported comparable performance between younger and older adults on the RME. Moreover, story-based affective ToM tests such as the Faux Pas test [19] have less consistently reported age-related differences with some studies reporting poorer performance with age [35] but others not reporting age-related differences [33]. Overall, it is unclear how social cognitive abilities, specifically cognitive and affective ToM abilities are affected by aging when the performance of older adults is compared to younger adults. A possible reason for the inconsistencies in the literature could be related to the way in which researchers assess ToM [13].

The aging literature has tended to assess the influence of age on the distinct components of ToM using different tests (e.g., [36]), and these paradigms vary in both their stimuli type and level of difficulty. For example, affective ToM has been examined using tests involving visual-static stimuli such as Tom’s taste test [23] and the RME [39]. In contrast, cognitive ToM has been examined using verbal vignettes [41] and visual-dynamic false belief story tests [20]. Existing tests of social cognition have been criticised as they require participants to read factual or fictional information regarding multiple characters and process mental state information. Poor performance could be a secondary consequence of broader cognitive difficulties [42]. Moreover, few aging studies have compared affective and cognitive ToM within the same test, making it difficult to contrast the influence of age on tests that are not directly comparable. Recently, Bottiroli et al. [22] attempted to measure cognitive and affective ToM using the Faux Pas test. They demonstrated that compared to younger adults, older adults performed poorer on cognitive ToM, but showed intact affective ToM abilities. Yet, some authors have argued that the Faux Pas imposes demands on both cognitive and affective ToM [13]. This test was designed before researchers explicitly regarded ToM as a multidimensional process and so there is no clear distinction between cognitive and affective ToM. Moreover, we would argue that the Faux Pas is a measure of affective ToM, as well as social norm understanding, since it primarily requires the participant to understand that a protagonist’s feelings have been hurt by a social norm violation.

One important aspect of social cognition which has not typically been assessed in the aging literature is the ability to understand social norms from interpersonal and intrapersonal perspectives. While intrapersonal understanding of social norms has been explored in studies of dementia [43], adults with Autism Spectrum Disorders (ASD) [7] and patients with schizophrenia and bipolar disorder [6], few studies have examined this ability in healthy aging. In one of the only studies exploring interpersonal understanding of social norms in healthy aging, Halberstadt, Ruffman, Murray, Taumoepeau and Ryan [44] found that older adults were poorer at discriminating between socially appropriate and inappropriate behaviours from short videos of social interactions compared to younger adults.

Performance on social cognition tests have been shown to be influenced by variables such as personality traits and measures of IQ (e.g., verbal comprehension and perceptual reasoning). Charlton et al. [45] has argued that age-related difficulties in ToM are not independent of measures of intelligence. They found that the association between age and ToM abilities as measured by Happé’s Strange Stories test was fully mediated by perceptual reasoning and partially mediated by verbal comprehension. Further studies have found correlations between ToM and verbal abilities [24] and have shown that perceptual reasoning performance accounts for age-related differences, again on Happé’s Strange Stories test [28]. These findings suggest that some tests may not be simply assessing our social cognitive abilities and this has important implications for interpretations of age-related differences in performance.

A hallmark characteristic of ASD is pronounced impairments in social cognition [46]. Moreover, research has shown that difficulties in social cognition are responsible for social functioning impairment in ASD [47], suggesting that social cognitive abilities are important contributions to the quality of an individual’s social interactions. This finding is relevant for the present study since characteristics typically found in adults with ASD are continuously distributed within the general population [48–50]. Indeed, subclinical autistic-like traits referred to as the Broad Autism Phenotype (BAP) [51] within the general population are related to reductions in social cognitive ability [48]. Individuals who exhibit more BAP traits report experiencing more social and interpersonal problems [50, 52]. Additionally, recent evidence suggests that BAP traits in older adults are associated with lower levels of social support, and increased self-reported levels of depression and anxiety [53]. Given the findings discussed above, it would be of interest to examine the relationship between the ESCoT, measures of intelligence and the BAP, and compare these to the findings of established tests.

To our knowledge, no tests are currently available in the literature that allow clinicians and researchers to examine different aspects of social cognition such as cognitive and affective ToM and understanding of social norms within the same test. Yet, some authors have argued that reliable assessments of a given construct should have multiple measures and these should differ in modality [54]. This could possibly be the reason for contradictory findings in the aging literature [13]. Moreover, while tests like the Movie for the Assessment of Social Cognition (MASC) [55], the Awareness of Social Inference Test (TASIT) [56], the Awkward Moments Test [57] and the Empathic Accuracy Paradigm [58] already exist and are all useful indices of social cognitive functioning, they are not without their limitations. For instance, the TASIT [56] uses excerpts from short interactions so lacks important contextual information, the MASC [55] is dubbed in English, the Awkward Moments Test [57] uses television adverts of exaggerated interactions and the Empathic Accuracy Paradigm [58] uses scenes from hidden filming which limits the range of mental states to be inferred.

We attempted to address these issues using a novel test of social cognition called the Edinburgh Social Cognition Test (ESCoT). We devised the ESCoT to explicitly measure both cognitive and affective ToM in the same test. The ESCoT also provides a much-needed measure of interpersonal and intrapersonal social norm understanding. Few tests measure more than one social cognitive ability in a single test, but the ESCoT provides four distinct and potentially informative insights into social cognitive abilities.

The aims of this study were to investigate the relationship between the ESCoT and a) age, b) measures of intelligence and c) the BAP in comparison to established tests. Additionally, we sought to examine convergent validity between the ESCoT and other measures of social cognition. By closely examining different social cognitive abilities in a systemic manner using the ESCoT, this study sought to shed new light on the consequences of aging on social cognitive abilities in younger, middle-aged and older adults.

Methods

Participants

A total of 91 healthy participants were recruited for this study: 30 aged between 18 and 35 years (15 male, 15 female), 30 aged between 45 and 60 years (15 male, 15 female) and 31 aged between 65 and 85 years (14 male, 17 female). The participants’ demographic information is reported in Table 1. None of the participants had any self-reported history of neurological or psychiatric disorders based on the Wechsler Adult Intelligence Scale (WAIS-III) exclusion criteria [59]. Participants were recruited from online advertisement, through a Psychology Department volunteer panel, and were reimbursed for their time. Written informed consent to participate in the study was obtained from each participant. The study was approved by the School of Philosophy, Psychology and Language Sciences (Psychology) Ethics committee at the University of Edinburgh.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Summary of demographic information.

https://doi.org/10.1371/journal.pone.0195818.t001

Results

Table 2 demonstrates the preliminary correlational analyses between cognitive ToM, affective ToM, inter- and intrapersonal understanding of social norms with VCI scores, PRI scores, age, years of education and gender. Variables with correlations that were significant at the p<0.20 level were included in the regression analysis. Table 3 shows a summary of the regression analyses for the subtests of the ESCoT.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Correlational analysis between the background predictors and measures of the ESCoT.

https://doi.org/10.1371/journal.pone.0195818.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Regression analyses for the subtests of the ESCoT with VCI and PRI scores.

https://doi.org/10.1371/journal.pone.0195818.t003

ESCoT total scores and IQ scores. The ESCoT total scores correlated with age (r = –0.42, p<0.20). Years of education (r = 0.13, p>0.20), gender (males = 1, females = 2, r = 0.09, p>0.20), VCI scores (r = 0.11, p>0.20) and PRI scores (r = 0.09, p>0.20) did not correlate with ESCoT total scores at p<0.20. Therefore, these variables did not meet criteria for inclusion in the model.

In the regression model, age was the only significant predictor of ESCoT performance (p<0.001, R² = 0.18). The inclusion of AQ, EQ and SQ scores produced a significant F-change (F-change = 5.44, p<0.05, ΔR² = 0.05). In the final model, only age (p<0.001) and AQ scores (p<0.05) were significant predictors of ESCoT performance, with older age and higher AQ scores predicting poorer performance on the ESCoT. EQ and SQ scores were excluded as predictors from the final regression model.

RME and IQ scores. For the RME scores, VCI scores (r = 0.28, p<0.20) met criteria for inclusion in the first stage of the analysis. Age (r = –0.10, p>0.20), years of education (r = –.07, p>0.20) and gender (r = –0.06, p>0.20) were not included. In the first model, VCI scores (p<0.05, R² = 0.06) predicted performance on the RME. Higher VCI scores predicted better RME performance. AQ, EQ and SQ scores were not retained in the final model.

RMF and IQ scores. Gender (r = 0.23, p<0.20), VCI scores (r = 0.38, p<0.20) and PRI scores (r = 0.27, p<0.20) met the criteria for inclusion in the regression models. Age (r = .02, p>.20) and years of education (r = –0.03, p>0.20) did not correlate with RMF scores at p<0.20. In the first model, VCI scores (p<0.01) and gender (p<0.05) were significant predictors of RMF performance (p<0.001, R² = 0.18). Female participants and participants who had higher VCI scores performed better on the RMF. No variables were entered into the model in the second stage as they did not meet criteria for inclusion.

In a regression model with PRI scores, both gender (p<0.05) and PRI scores (p<0.05) were predictors of RMF scores (p<0.01, R² = 0.13). Female participants and participants with higher PRI scores were associated with better performance on the RMF. AQ, EQ and SQ scores were not retained in the final model.

SNQ and IQ scores. Age (r = 0.14, p<0.20) and VCI scores (r = 0.23, p<0.20) met criteria for inclusion. Gender (r = 0.07, p>0.20), years of education (r = –0.12, p<0.20) and PRI scores (r = 0.06, p>0.20) did not meet criteria for inclusion in the regressions. However, all regression analyses were not significant (all p>0.10).

JoP and IQ scores. Age (r = 0.05, p>0.20), years of education (r = –0.05, p>0.20) gender (r = –0.13, p>0.20), VCI scores (r = –0.06, p>0.20) and PRI scores (r = –0.09, p>0.20) did not meet criteria for inclusion in the regression models. Moreover, all regression analyses for the JoP were not significant (all p>0.10).

Discussion

The current study presented a new within subjects’ measure of social cognition that assesses cognitive and affective ToM, as well as intra- and interpersonal social norm understanding, within the same test. We examined the effects of age, measures of intelligence and the BAP on the ESCoT and established tests of social cognition. Additionally, we investigated the relationship between the ESCoT and established measures of social cognition. Total ESCoT scores were predicted by the age of participants and their AQ scores, here increasing age and AQ scores resulted in poorer performance. Investigation of the subcomponents of the ESCoT revealed that performance on cognitive ToM was significantly predicted by age, with increasing age resulting in decreased performance on cognitive ToM. Affective ToM was also predicted by age but also gender; in this instance, better performance was associated with being younger and female. Moreover, performance on interpersonal understanding of social norms was predicted by age and AQ scores–increasing age and AQ scores were predictive of poorer performance. On the subtest of intrapersonal understanding of social norms, higher AQ scores predicted poorer performance.

Notably, the ESCoT total score and sub-test measures were not associated with the two measures of intelligence; verbal comprehension (VCI) and perceptual reasoning (PRI). This contrasts with performance on some of the more standard tests of social cognition. In the present study, we found that participants with higher verbal comprehension scores performed better on the RME, while RMF performance was significantly predicted by measures of verbal comprehension, perceptual reasoning and gender. Here, female participants and those with higher verbal comprehension and perceptual reasoning scores performed better on this measure of affective ToM. The correlation analysis demonstrated that ESCoT total scores significantly correlated with the RME and showed a trend towards significance with the SNQ, indicating convergent validity.

Similar to previous findings in the literature which have demonstrated aged-related difference in cognitive ToM [20, 22, 23, 26, 30, 36–38], age predicted poorer performance in cognitive ToM on the ESCoT. This provides further evidence that, as we get older, we experience difficulties in our ability to infer what another individual is thinking. Moreover, we found that increasing age predicted poorer performance in participants’ ability to infer what another is feeling, comparable to some [21, 28, 36, 40], but in contrast to other studies [22, 33, 35]. It could be argued that the findings here are more representative of the population, as we included adults aged 18–85 years while Bottiroli et al. [22] only included younger and older adults. Or, as Henry et al. [13] have argued, age-related differences can be the consequence of the type of task used. For example, Phillips et al. [34] examined how well older adults were able to assess the severity of contextual emotions of individuals in short stories. They found younger and older adults did not significantly differ in this ability. However, forced choice tests offer limited insights in understanding the relationship between age and social cognition. Primarily because there are few real-world social interactions where inferring what another person is feeling is forced-choice in nature. Overall, these results suggest that the process of healthy aging is associated with difficulties in both components of ToM.

To our knowledge, this is the first study to assess the ability to understand social rules in the same test as ToM and explicitly examine interpersonal (did X behave as other people should behave?) and intrapersonal (would you have acted the same as X?) understanding of social norms. Age was found to predict poorer performance on interpersonal understanding of social norms. These findings add to the preliminary findings of Halberstadt et al. [44] who showed poorer performance of older adults compared to younger adults on interpersonal understanding of social norms. We provide a novel finding in regards to intrapersonal understanding of social norms; we showed that the age of participants was not a predictive variable of performance. This suggests that age does not impact our own knowledge of how we should behave in social situations, and not all our social cognitive abilities are negatively affected by age. Both of these findings demonstrate that understanding of social norms warrants further investigation.

Although both cognitive and affective ToM were affected negatively by age, we do provide some evidence for a dissociation between the two processes in that performance is predicted by different demographic variables. This is analogous to the findings that cognitive and affective ToM correlate with different cognitive processes [22]. Cognitive ToM performance was negatively predicted by age while affective ToM was significantly predicted by age and gender. Like Duval et al. [23], we found that both cognitive and affective ToM show impairments with advancing age in the same study. The advantage in this study was that we were able to measure cognitive and affective ToM within the same test, unlike Duval et al. [23] who relied on different tests to measure these abilities and was therefore unable to control for task difficulty. Gender was only found to predict performance on affective ToM; this is similar to research found in the literature which has shown that women are significantly better at inferring what a character is feeling compared to men [18, 72, 73]. These results show that, as well as considering the consequences of aging on our social cognitive abilities, we should consider the gender of the sample population. Furthermore, they highlight the importance of adopting social cognitive tests that assess cognitive and affective ToM separately and suggest composite tests are not appropriate to accurately examine ToM in aging populations. Using within subjects tests are essential if we are to better understand whether aging does indeed affect cognitive and affective ToM in the same way.

The only test of social cognition that was associated with the measures of the BAP was the ESCoT, suggesting that perhaps the ESCoT is more sensitive to difficulties in social abilities of individuals on the BAP compared to established tests. Here, lower scores in inter- and intrapersonal understanding of social norms were associated with higher scores on the AQ. Additionally, we found that the presence of more autistic traits predicted poorer overall performance on the ESCoT. These are novel findings but makes sense in the context of the BAP, as impaired social cognition is related to the milder social-behavioral phenotype described as part of the BAP [48, 50, 52]. Research on the understanding of social norms in healthy aging is limited but these findings are in line with research that show that adults with ASD perform poorer than controls on tests such as the Faux Pas which implicitly assess social norms understanding [74]. However, the relationship between ASD and intrapersonal understanding of social norms is less clear and requires further investigation For example, Baez et al. [7] found that adults with ASD do not significantly differ on this ability compared to controls. Nonetheless, the observed relationship between the ESCoT and the AQ demonstrates that this new test of social cognition may offer new insights into the relationship between the BAP and social cognition in healthy aging populations and may be valuable in ASD research.

An advantage of the ESCoT over other tests of social cognition, is that overall performance was not related to measures of IQ, namely verbal comprehension and perceptual reasoning performance. However, this is not the typical finding with social cognition measures. Charlton et al. [45] found that performance on Happé’s Strange Stories test (a composite ToM task) was fully mediated by performance IQ, executive function, and information processing speed and was partially mediated by verbal IQ. Moreover, again on Happé’s Strange Stories test, Sullivan and Ruffman [28] both found that ToM abilities were related to perceptual reasoning abilities. In both the current study and previous studies in the literature [64, 72, 75], performance on the RME and RMF was found to be predicted by verbal comprehension. In one study, the only significant predictor of performance on the RME test was verbal comprehension which accounted for 11.7% of the variance [76]. This has implications for studies using the RME and RMF to investigate affective ToM as they appear to be tests of verbal comprehension as well as affective ToM.

A limitation of the present study is that we did not examine the relationship between executive functions and the ESCoT. Given that social cognition has been associated with executive abilities in aging [22, 23, 27, 72], future work might explore potential associations between the ESCoT and processes such as inhibition, set-shifting and updating. Finally, it has been suggested that the clinical assessment of social cognition should emulate the way in which individuals process social situations in everyday life [4]. As argued by Henry et al. [13], dynamic-visual information such as images depicting a social interaction that lead to a protagonist in a particular mental state is more ecologically valid and information-rich compared to verbal narratives. Consequently, dynamic cartoons were chosen as the mode of presentation in the ESCoT. This allowed perceivers to use many more cues to make inferences [77], similar to real-life. Videos of real individuals interacting would be the ideal stimuli for assessing social cognitive abilities to maximise ecological validity. However, social interactions are highly complex [78] and social information can be difficult to control in real interactions. Therefore, it may be difficult to separate the specific social cognitive process that the test is intending to measure. With animated characters, specific social cognitive abilities can be more easily isolated and individual social differences can be controlled; essentially all of the parameters can be regulated. For these reasons, we chose to use animated interactions for the ESCoT.

This study is the first to assess cognitive ToM and affective ToM, as well as interpersonal and intrapersonal understanding of social norms within the same test in younger, middle-aged and older adults. We have provided further evidence for similar but distinct components of ToM and evidence for social norm understanding. These findings are useful in furthering our understanding of the consequences of aging on our social cognitive abilities. They also demonstrate specific advantages of the ESCoT over other tests of social cognition. The ESCoT is able to assess distinct aspects of social cognition within a single task and using a within subjects design, allowing for systematic comparisons of these abilities. This important feature of the ESCoT allows it to be a useful test for researchers examining age-related changes in social cognition, and perhaps as a test for clinicians in populations such as adults with ASD, traumatic brain injury patients or psychiatric and neurological patients with suspect social cognitive impairments. As a consequence of its design, the ESCoT can provide researchers and clinicians with an objective measurement of four important social cognitive abilities that are needed to interact with others. This is particularly useful in clinical settings since the results can be used to personalise interventions or educate caregivers about the difficulties the patient might be experiencing in processing social information and interacting with others. In conclusion, these findings show that the ESCoT is a valuable measure of social cognition and, unlike established and standard tests of social cognition, performance is not predicted by measures of verbal comprehension and perceptual reasoning. This is particularly valuable in order to get an accurate assessment of the influence of age on our social cognitive abilities.

Supporting information

Acknowledgments

We are grateful to all our participants for taking part in the study. We would like to thank Caroline McHutchison for her help with the inter-rater reliability analysis.