Since the 1970s, the cognitive revolution has led to an increasing interest in several functions in neuropsychology, particularly social cognition. The human being is studied through a social lens, as someone able to perceive and process social signals in order to behave appropriately during daily interpersonal relationships. More than 13,000 studies dealing with social cognition have been published during the last 5 years (PubMed). However, moral cognition and the knowledge of social conventions remain less studied than other well-known sociocognitive processes such as theory of mind (ToM) or empathy. Although morality can be understood as a unified concept from a functional point of view, it appears as a complex entity from a cognitive perspective (Greene, 2015), moral judgments being based on the integration of several inputs (e.g., emotions, others’ mental states, social knowledge, context, intentionality…) and underpinning adapted behavior (Mikhail, 2007). Thus, moral cognition might offer an interesting way to better understand how higher-order sociocognitive functions are integrated and/or interact with one another—another gap in the current literature on the neuropsychology of social cognition (Cassel et al., 2016; Etchepare and Prouteau, 2018).

Social cognition refers to a wide range of emotional and cognitive processes that enable humans to perceive and understand others’ emotions and mental states to adjust their own behavior (Shany-Ur and Rankin, 2011; Etchepare and Prouteau, 2018). A large distributed cerebral network involving prefrontal, temporal and insular structures is required for appropriate social interactions as it underpins accurate emotion recognition, ToM, empathy, and moral cognition (Kennedy and Adophs, 2012; Noonan et al., 2018). Emotion recognition involves perceptual capacities for decoding and making meaning out of emotional expressions using facial, prosodic or postural cues (McDonald, 2017). ToM refers to the ability to make inferences about others’ mental states in order to explain and predict their behavior (Premack and Woodruff, 1978; Conway and Bird, 2018). A distinction is usually made between cognitive and affective ToM (Brothers and Ring, 1992; Shamay-Tsoory and Aharon-Peretz, 2007). The former refers to the knowledge about others’ beliefs and/or intentions, whereas the latter corresponds to the understanding of others’ emotional states. Empathy, defined as the ability to share and understand another person’s feelings (Decety and Jackson, 2004), relies on two dissociable systems: the affective sharing of the pain and distress of others (i.e., emotional empathy) and the ability to adopt another’s psychological point of view (cognitive empathy; Shamay-Tsoory et al., 2009). Emotional empathy underlies empathic concern, defined as “other-oriented feelings of sympathy and concern for unfortunate others” (Davis, 1983, p. 114), which is particularly associated with a propensity to prosocial behavior. Cognitive empathy requires ToM processes but also the self-other distinction, top-down regulation processes and executive functions such as flexibility and inhibition (Preckel et al., 2018; Stietz et al., 2019). Finally, moral cognition corresponds to a set of capacities by which people learn, store and activate conventional and moral norms to make judgments and decisions about these norms (i.e., to decide whether an action is right or wrong; Haidt et al., 1993; Tasso et al., 2017; Voiklis and Malle, 2017). Conventional norms (e.g., eating with your hands) are mainly based on cultural rules, whereas moral norms (e.g., physically or emotionally hurting someone) appear more universal and involve emotional reactions (Hollos et al., 1986). Although both are judged as not permissible, moral transgressions (i.e., violation of a moral rule) are usually judged more severely than conventional transgressions (i.e., violation of a conventional rule; Blair, 1995; Nichols, 2002; Turiel, 2002). Moral judgment relies on the understanding of others’ mental states (i.e., judging the intentionality of action; inferring the victim’s emotional state), the semantic knowledge of norms but also on emotional abilities (Blair, 1995; Moll et al., 2005; Prinz, 2006; Young and Koenigs, 2007), for instance our level of empathic concern (Gleichgerrcht and Young, 2013). Thus, as predicted by Greene’s dual-process model (Greene, 2013), moral judgments are driven both by intuitive emotional responses and by conscious reasoning processes.

Social cognition is important for life satisfaction, social interactions and lower degrees of loneliness, especially in old age. However, previous studies have suggested that some component processes of social understanding may decline in normal aging. First, emotion recognition is classically described as being lower in older adults, especially when participants are asked to identify negative emotions (e.g., Ruffman et al., 2008; Birmingham et al., 2018; Visser, 2020). Considering ToM, although findings are equivocal (e.g., Happé et al., 1998; MacPherson et al., 2002; Wang and Su, 2013), most of the previous studies showed that increasing age results in decreased performance on ToM tasks, especially when assessing cognitive ToM (Duval et al., 2011; Henry et al., 2013; Bottiroli et al., 2016; Laillier et al., 2019). Thus, older adults show reduced levels of cognitive empathy (see Beadle and de la Vega, 2019 for a recent review). On the contrary, there is little evidence that emotional empathy decreases in aging. Previous findings rather suggest that emotional empathy is similar to or even higher in older than in younger adults, especially for empathic concern (Sze et al., 2012; Ze et al., 2014; Reiter et al., 2017; Beadle and de la Vega, 2019). Consistently, previous findings suggested that older adults might be more motivated than younger adults to help others (Sze et al., 2012; Rosi et al., 2019; Mayr and Freund, 2020). Overall, the processes underlying moral judgments seem to be impacted by aging. However, surprisingly—the issue of whether and how moral judgments and social norms might change in normal aging remains poorly explored (Moran et al., 2012; Baksh et al., 2018; Margoni et al., 2018). Moran et al. (2012) investigated whether aging was associated with difficulties in making an intent-based moral judgment. This kind of judgment is based on the agent’s intentions, blaming intentional harmful acts more than accidental ones. Unlike their younger counterparts, older adults relied less on intentions than on outcomes in their moral judgments (Moran et al., 2012). Margoni et al. (2018) replicated these findings and extended them by showing that (i) this age-related change was associated with a decline in ToM and (ii) no age-related change was observed for the moral evaluation of helpful actions. Furthermore, using a new task to assess different components of social cognition within the same test, Baksh et al. (2018) showed that increasing age was predictive of poorer performance on the interpersonal understanding of social norms (Baksh et al., 2018).

However, methodological limitations of traditional sociocognitive tasks might be pointed out. First, most of previous studies assessed social cognitive processes separately from one another, preventing an integrative assessment of social cognition. Few studies in the literature have investigated the different aspects of social cognition (Dziobek et al., 2008) by including conventional and moral judgments within the same task (Baksh et al., 2018). Second, the way the age-related decline of social cognition influences older adults’ intention to act/react has been poorly assessed, except for the link between empathy and prosocial behavior in some studies (e.g., Beadle et al., 2015). This point seems crucial to better understand the impairments of social behavior in the pathological framework. Finally, previous authors have pointed out the lack of ecological validity of sociocognitive tasks usually proposed (Isaacowitz and Stanley, 2011; Dziobek, 2012; Dawson and Marcotte, 2017; Sohlberg et al., 2019). In daily life situations, people are able to interpret others’ emotions and mental states as they dynamically evolve. Consistently, it appears that when using a dynamic presentation, emotion recognition is significantly higher (Lambrecht et al., 2012; Grainger et al., 2015) and cerebral activations underlying social cognition are stronger than when using photographs (Kilts et al., 2003; Trautmann et al., 2009). Also, in daily life experience, we rely not only on several key cues from the communicator (such as their voice, posture and bodily expressions), but also from contextual information (Achim et al., 2013; Duclos et al., 2018; Allain et al., 2020). Taking these limits into account, with more naturalistic conditions the technological possibilities offered by virtual reality (VR), even with non-immersive settings (i.e., Zhang et al., 2003; Cushman et al., 2008) or by serious games (Parsons and Courtney, 2001; Baldwin and Dandeneau, 2009; Kato and de Klerk, 2017), might be an interesting alternative to assess sociocognitive abilities. Indeed, immersive as well as non-immersive VR offers the opportunity to assess social cognition within an integrative context because of richer visual and contextual cues. Its playful properties provide a better participant involvement while allowing for the precise presentation and control of dynamic multi-sensory 3D stimulus environments (Rizzo et al., 2004). Previous research in clinical settings showed that non-immersive VR-based measures may be more sensitive in the detection and treatment of cognitive impairments than traditional methods (e.g., Allain et al., 2014; Foloppe et al., 2018; see also Liu et al., 2019 for a recent review). The main aim of the present study was to assess moral cognition and the different related components of social cognition in a integrative manner, in order to investigate (i) the presence of an age-related decline in the way participants detect and assess conventional and moral transgressions; (ii) the presence of an age-related decline in ToM and empathy; (iii) the intention to react toward the social situations encountered during the task. For this purpose, an original task was developed using non-immersive VR to assess social cognition in more naturalistic conditions. A secondary objective was to investigate relationships between moral cognition and other sociocognitive processes. Based on the previous findings, we expected an age-related decline in detecting and assessing conventional and moral transgressions and poorer performance in ToM. However, empathic concern and participants’ emotions following transgressions might be higher in older adults, associated with more intentions to react (e.g., to help others in need).

The present study was designed to assess different sociocognitive processes within the same task, in order to investigate a potential decline in detecting and judging moral and conventional transgressions as well as potential age-related changes in ToM and/or empathy. We also wanted to assess participants’ intention to react toward social situations. To investigate these issues, a new sociocognitive task was developed and proposed to young and older adults. The originality of this computer-based virtual task (REALSoCog) is that it aimed to assess several sociocognitive processes within the same task, in more naturalistic conditions. A secondary purpose was to investigate relationships between moral cognition and other sociocognitive processes.

Our results first showed a preserved ability to accurately detect moral and conventional transgressions with advancing age. This result is somewhat debatable, however, as OA tended to detect transgressions when there are none (i.e., in control situations). Cohort and generational factors may play a role, as YA were more permissive toward behaviors that were once more reprehensible (e.g., wearing a mini skirt). However, this age-related over-rating of transgressions existed only in lower educated OA and disappeared when the lower understanding of some social situations among OA was considered. Consistently, the transgression severity judgments seem to be similar in both groups with a distinction between moral and conventional norms, moral transgressions being more severely judged than conventional ones (Turiel, 1983; Ehrlé et al., 2011). Taken together, these results suggest—contrary to our expectations—the absence of age-related decline in the way participants detect and assess moral and conventional transgressions (see Moran et al., 2012; Margoni et al., 2018 for convergent results). Older adults’ judgments might be based on the agent’s intentionality rather than on outcomes, as in YA. In both groups, our results suggest a regulation of participants’ judgments (especially in terms of their assessment of the seriousness of transgressions). A thorough examination of why they judged behavior as a transgression by exploring participants’ oral justifications supporting their moral reasoning could help to specify age-related changes.

Second, the present results showed differential effects of normal aging on ToM depending on the component considered. Cognitive ToM appears to be preserved in OA while age-related decline concerns affective ToM here (see Henry et al., 2013 for a meta-analysis). REALSoCog proposes a first-order cognitive ToM assessment, which is well preserved in aging (McKinnon and Moscovitch, 2007; Duval et al., 2011) because less cognitively costly (Maylor et al., 2002) than the second-order cognitive TOM. Consistently with previous studies, OA performance decreases when it comes to inferring affective mental states (Slessor et al., 2007; Bailey and Henry, 2008; Duval et al., 2011; Fischer et al., 2017; Fernandes et al., 2019). The executive cost, especially inhibition capacities that are altered with age, is likely to contribute to these difficulties (Bailey and Henry, 2008; Duval et al., 2011; Wang and Su, 2013; Fischer et al., 2017). The relative preservation of cognitive ToM abilities in our cohort of OA could explain why they did not show any decline in moral/conventional judgments. Indeed, Moran et al. (2012) proposed that an age-related decline in cognitive ToM disturbs the way older participants infer the agent’s intentionality during moral reasoning. On the other hand, it is interesting that the difficulties in affective ToM disappeared when the victim was a senior. In other words, when faced with situations that favor taking the other character’s perspective and facilitate identification, the mentalization abilities of OA seem increased (see Sze et al., 2012 for convergent conclusions regarding empathy).

Third, our results showed that emotional reactivity and emotional empathy are preserved in normal aging (see also, Beadle et al., 2015; Reiter et al., 2017; Bailey et al., 2020). Although these results do not support the hypothesis of higher emotional empathy with advancing age, they might contribute to the current debate in the literature as to whether empathy levels increase in the elderly or not (Sze et al., 2012; Beadle et al., 2015; Wieck and Kunzmann, 2015; Bailey et al., 2020). REALSoCog offers a state assessment of emotional empathy, corresponding to a punctual and contextual emotional reaction provoked by observing the suffering of others. Unlike most tools used to assess this component in previous studies, it enables an objective measure that avoids the limitations associated with traditional self-evaluations, which may be biased due to social desirability (Beadle and de la Vega, 2019).

Finally, our results suggest that OA did not increase their intentions to act when looking at moral and social transgressions. However, in contrast with previous studies showing that OA are more likely to be involved in prosocial behaviors (Sze et al., 2012; Rosi et al., 2019; Mayr and Freund, 2020), this result is consistent with the observation of a similar emotional empathy and emotional reactivity to that of YA. Indeed, these components modulate social behavior: the personal involvement felt by a subject when judging a transgression predicts their propensity to share their disapproval (Brauer and Chekroun, 2005; Helweg-Larsen and LoMonaco, 2008). This is consistent with the finding of a greater action propensity for moral transgressions compared to conventional transgressions in both groups, the former being deemed more serious than the latter (Turiel, 1983), and therefore more likely to involve subjects emotionally. It should be highlighted that REALSoCog demonstrated a significant—but small—increase in inappropriate behavioral intentions in OA. These results might argue in favor of the sensitivity of the tool proposed, since inappropriate social behavior is not systematically detected by peers (Henry et al., 2009).

Considering our secondary goal of investigating relationships between moral cognition and other sociocognitive processes, we found slightly different patterns in YA and OA. Our regression analyses showed that, consistent with previous studies (Voiklis and Malle, 2017), YA’s moral/conventional judgments seem to rely on the agent’s intentions and on an analysis of mental states (especially in terms of their assessment of the intensity of others’ emotions), but also on the analysis of their own mental state (in terms of emotions they feel). This result suggests a double influence of the “social brain” and the “emotional brain” (Young and Dungan, 2012) in moral/conventional judgments (Greene et al., 2001; Nichols, 2002; Moll et al., 2008; Tasso et al., 2017; Bretz and Sun, 2018). In contrast, OA’s moral/conventional judgments did not rely on their own emotions but only on the analysis of others’ mental states (in terms of the others’ intentionality and the intensity of others’ emotions). Overall, this exploratory study suggests that REALSoCog could be an interesting task to assess sociocognitive processes in normal aging. This task seems to provide an objective assessment of some social cognitive aspects (especially emotional empathy and social behaviors), which are often only assessed based on subjective tools. Although the sociocognitive functioning of OA is marked by considerable inter-individual heterogeneity, requiring caution in generalizing the results of the present study, it has also to be underlined that the nature of the task itself may explain some discrepancies with results reported in the literature. Offering more naturalistic conditions, this computer-based virtual task may have reduced the involvement of other non-social cognitive functions that can interfere with the social cognitive functioning in OA, thus contributing to the finding that fewer age-related changes were revealed using the REALSoCog task in comparison with the classical tasks used in the literature. Although the acceptance or relevance of non-immersive VR could be questioned in the elderly due to a reduced exposure to this kind of technology in comparison with YA, it has to be underlined that our results cannot be explained by group differences in their feeling of immersion in the virtual environment, since comparable judgments were observed. Finally, the integrative measure of sociocognitive functioning may offer possibilities to investigate and better understand relationships between moral cognition and other sociocognitive processes such as theory of mind and empathy.

Nevertheless, methodological limitations need to be discussed. First, our experimental design did not allow to directly test the added value of the use of a virtual environment in comparison with classical tasks. Even whether the REALSoCog task is hypothesized to be more naturalistic offering a more ecologically valid assessment, we did not compare OAs’ performances on this task vs. on traditional tasks. Future studies are required to further investigate this issue to confirm the validity of the REALSoCog task. Second, although slightly lower than in the YA, the OA’s educational level was quite high in the present study. This may have compensated for age-related decline. However, when contrasting OAs’ performances according to their educational level, it appears that the pattern of lower educated OA was quite similar to the one observed in higher educated ones on moral cognition, ToM, empathy and declarative intentions of social behaviors. It should also be noted that an intergenerational comparison implies demographic and ideological pluralism, which can influence participants’ normative theory with changes in conventional norms over time (Turiel, 1983). Furthermore, the test is long and can lead to fatigue and/or disinvestment by the participants. Some situations were also associated with ambiguous interpretations as suggested by the corresponding understanding score. In a future study, item analysis will contribute to selecting the most reliable situations for a shorter version of the task. Basic perceptual differences between YA and OA might have contributed to explain some of the age-related decline observed. Such abilities should have been controlled. This is also the case for potential cybersickness symptoms (Kourtesis et al., 2020) or differences in the familiarity with using computer devices (Zygouris and Tsolaki, 2015), where both may have affected performances. Considering the former, it has to be underlined that cybersickness symptoms are mainly associated with immersive VR (Plechatá et al., 2019). Although we cannot totally rule out the presence of such symptoms in our results, it should be noted that in case of non-immersive VR, these symptoms appear after long time exposure (Tazawa and Okada, 2001) while this is not the case with the REALSoCog task. For the latter point, although OA might be less familiar with using computers, all our participants were trained with the simulated city environment for as long as they needed before the experimental task.

This integrative social cognition task will offer both theoretical and clinical perspectives. From a theoretical point of view, it may provide an interesting way to better understand how higher-order sociocognitive functions are integrated and/or interact with each other, which is currently a gap in the literature on the neuropsychology of social cognition (Cassel et al., 2016; Achim et al., 2020). From a clinical point of view, despite its interest for both diagnostic and rehabilitation purposes, the neuropsychological assessment of social cognition remains underexamined in clinical practice mainly due to a lack of reliable tools (Kelly et al., 2017). Thus, future studies should investigate the validity and the potential value of the REALSoCog task in clinical neuropsychology, as well as clarify the sociocognitive processes interdependency. A more immersive and interactive version of this task may also be of interest.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by the Comité d’Ethique de la Recherche de l’Université Paris Cité, N°IRB: 00012020-115. The patients/participants provided their written informed consent to participate in this study.

PN, NE, and PP contributed to conception of the study. AG-B and EO contributed to design of the virtual environment and technical assistance. E-FM organized data collection, performed the statistical analysis, and wrote the first draft of the manuscript. PN wrote sections of the manuscript. All authors contributed to manuscript revision, read and approved and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.