The So-Moral-VR was developed as a variation of an existing computer-based tool extensively tested in a series of validation and clinical studies (Dooley et al., 2010; Chiasson et al., 2017; Vera-Estay et al., 2016; Beauchamp et al., 2019; Beauchamp et al., 2013; Garon et al., 2018a,b). The computer-based task includes between 10 and 19 socio-moral dilemmas, presented on a computer screen as a three static picture narrative of realistic, everyday social situations that require decision-making (see Figure 1) (Dooley et al., 2010). The scenarios portray child actors playing out various social scenarios representing conflicts centered on the moral domain (e.g., concerns with justice, welfare-harm, and rights) according to Social Domain Theory (Turiel, 1983) and the dilemmas were generated to include a range of moral values and social norms.

The dilemmas are presented in the first-person perspective (subjective point of view) (Dooley et al., 2010). The decision to use the first-person perspective was based on a number cognitive, neural, and design considerations. On one hand, the third-person perspective presents a form of indirect questioning that can be useful for avoiding social desirability (because thinking about what someone else should do is less confronting than admitting what one would really do) (Fisher, 1993). This distance from the social problem has also been associated with better reasoning because it encourages perspective taking and compromise (Grossmann and Kross, 2014). On the other hand, this standpoint introduces other issues that make it inadequate for measuring an individual’s own reasoning, decisions, and behavior. For example, functional neuroimaging evidence shows that different brain regions and networks are activated based on whether stimuli are presented in the first- versus the third-person perspective (Boccia et al., 2017; Avram et al., 2014). Relatedly, cognitive and socio-cognitive demands also differ, as shown by the reduced empathic response associated with third-person moral reasoning (Avram et al., 2014). Furthermore, the third-person perspective is associated with actor-observer bias. Defined as the tendency to explain our own actions by the situation, but other people’s actions by their personal traits, this bias can affect moral evaluation and does not accurately represent the processes used in real-life (Avram et al., 2014). Finally, the first-person perspective increases personal and emotional involvement (Orobio de Castro et al., 2002). Thus, to ensure that the task recruited cognitive and neural networks comparable to those engaged during real social interactions—and to mirror real-world experiences as closely as possible—we employed a first-person perspective to enhance ecological validity.

A pilot study by Dooley et al. (2010) provides details on the development of the original computer version of the task including data on familiarity, enjoyment, engagement and comprehension, which are also indicators of ecological validity. All dilemmas were designed to take place in everyday contexts that are meaningful to adolescents. The dilemma familiarity measure indicated that 80% of adolescents reported that they, or someone they knew, had encountered similar situations in real life. Comprehension was also measured to ensure the dilemmas could be understood with no accompanying text, thus precluding the necessity of high-level reading or attention skills (Dooley et al., 2010). Ninety-six percent of participant indicated that they understood the dilemmas, suggesting that they were accessible to most youth (Dooley et al., 2010).

One of the main objectives of developing the So-Moral task was to provide a quantitative SMR score, as this can lead to the creation of standardized norms, a practice at the core of psychological and neuropsychological assessment, and the comparison of SMR maturity levels across individuals (e.g., clinical populations, age groups). After viewing a dilemma, participants are asked whether they would engage in the proposed action and why they would or would not do so. For example, one dilemma shows a classmate asking for the answers to an exam. The participant (or user) is then asked, “Would you do this?” (i.e., give your answers to your friend) and “Why?.” The answers are recorded, transcribed, and coded to generate two scores. First, the socio-moral decision-making score (SMDM) is based on the dichotomic response (“yes” or “no”) to the first question. Second, the SMR maturity score is based on the answer to the second question (Why?). Verbatim justification responses are recorded and coded using a 5-level scoring system adapted from the cognitive-developmental approach (Gibbs, 2019; Dooley et al., 2010; Kohlberg, 1981). This distinction in scores aligns with theories arguing that moral judgment tends to be intuitive, whereas moral reasoning is constructed after the judgment has been made to justify it (Haidt, 2001). Developmental stages of moral reasoning have been adapted to fit the social nature of the dilemmas in the So-Moral task and consist of the following: (1) centration and authoritarian-based consequences; (2) egocentric/pragmatic exchanges; (3) interpersonal focus; (4) societal regulation; and (5) societal evaluation.

Transition stages (1.5, 2.5, etc.) are used to account for answers that provide elements of two consecutive reasoning stages. These transition stages are attributed when the response is not developed enough to belong to the higher level. For example, in the dilemma about whether a person would share their exam answers with a peer, a participant might say: “Yes (I would share my answers), because everyone has their own difficulties and the school system is trash” and obtain a score of 4.5 because they provide elements of social regulation (4) and a partial response for social evaluation (5), but without fully explaining their thinking regarding the latter (missing information regarding why it is “trash”). When an answer combines elements of two non-consecutive stages, the higher level is attributed. That is, a participant might provide a high-level component in their answer, but also include additional reasons that would score at a lower level if given in isolation. For example, for the same dilemma, a participant might answer “No (I would not share my answers), because the teacher is present and I think it goes against the goal, for example, I have studied and it is not good after all that to give my work to someone, while they have probably not done much. Studying is an accomplishment and it allows you to have good grades and the other person has not studied so it’s their problem. But for sure I will explain it to the person, even if they will not agree with my decision”. This participant would score 5 because their answer shows a societal evaluation that is developed enough to obtain this level, even though there are elements of the lower (1) centration and authoritarian-based level (reference to the teacher being present). The total moral maturity score is therefore the total of the SMR reasoning scores obtained for each dilemma.

In sum, the task design and scoring, grounded in a cognitive-developmental framework, assumes that more mature reasoning is characterized by greater cognitive flexibility and abstraction. It enables individuals to “decenter,” considering not only the personal consequences of a decision, but also its broader impact on societal functioning. For this reason, the system assigns the score based on the most mature reasoning expressed, even when the response also includes elements of simpler reasoning. Because the justification is articulated in the first person, it allows underlying socialization processes to emerge. While this qualitative richness may diminish when the tool is applied in quantitative research, its clinical and intervention potential lies precisely in this personalized dimension. Spontaneous responses enable young people to integrate situated social knowledge into their reasoning (e.g., “in my school or in my neighborhood X is not accepted”), and to articulate arguments drawing on deontological, utilitarian, care ethics, and other moral frameworks. Most importantly, it encourages them to reflect on and weigh these considerations in their decision-making. This process can help identify tendencies and biases in reasoning and could progressively guide individuals toward more socially adapted forms of moral reasoning and decisions.

In contrast to the computer-based version which presents the dilemmas in no particular logical order, the VR task was designed as a storyline with a sequential, logical presentation of the dilemmas to ensure that sense of presence is not disrupted after each dilemma is completed. Users begin in a virtual house, simulating the start of a typical day and in line with the goal of recreating everyday experiences in an adolescent’s life. The user picks up their mobile phone and leaves their house to complete the dilemmas, moving from one familiar setting to another (house to school, through the town, and into a restaurant, for example).

The original task required an examiner to elicit responses to the dilemmas by directly asking the user verbal questions. However, the use of an external individual in virtual reality can break the immersion phenomenon and sense of presence, thus reducing ecological validity (verisimilitude). In addition, this can introduce social desirability bias on the part of users, because they may feel pressured to answer in socially expected ways, potentially compromising the validity of the task (Oh et al., 2019; Palombi et al., 2023). To circumvent these problems, at the beginning of the task, the user picks up their mobile phone from a coffee table in their virtual house and keeps it with them throughout the whole assessment. The mobile phone is used to deliver the assessment questions using pre-recorded audio stimuli, as if the participants were speaking to someone on a call. This design feature was included with the goal of reducing social desirability bias by limiting the presence of the examiner and ensuring immersion is maintained, thus tending towards a self-administered task (Nederhof, 1985).

To induce the question coming from the phone, a “Wizard of Oz” effect (i.e., a behavior that seems to come from a virtual character or environment in the VR assessment, but is actually driven by a human controlling it) was used by programming specific keyboard keys for the examiner to trigger the questions within the virtual environment at appropriate moments without needing to directly interact with the user, allowing for more control and flexibility, and better human-computer interaction and user experience (Pan and Hamilton, 2018). The initial version of So-Moral-VR (Morasse et al., 2021) was developed using Virtools®, but the task was subsequently reprogrammed and upgraded to Unity®. Platform choices were based on the developers’ recommendations, available resources at the time, and technological compatibility (e.g., Virtools® was not compatible with Oculus® devices, which emerged over the course of development and initial validation). The transition from Virtools® to Unity® also enabled enhancements in visual stimulus quality.

The VR task also enhances ecological value by incorporating interaction, navigation, and audio elements within an immersive virtual environment. These additions make the stimuli more dynamic and realistic (Paiva-Silva et al., 2016; Rymarczyk et al., 2011; Sato and Yoshikawa, 2007). The VR task was developed to be experienced using six degrees of freedom (6DoF) headsets to maximize immersion and interactivity. A 6DoF navigation allows users to move freely in three-dimensional space—forward/backward, up/down, left/right—as well as rotate around three axes of head movements (pitch, yaw, and roll). To move within the virtual environment users must press a button on the joystick to move forward and look in the direction they wish to go. This navigation mode contributes to ecological validity, as people typically look in the direction they intend to move. This approach also mirrors real-world cognitive demands, requiring users to actively produce movement aligned with their intended direction. The task navigation mode was designed to be as immersive as possible while maximizing user control, to minimize the risk of cybersickness (Davis et al., 2014). Teleportation, a navigation mode used often in VR games, is not the default mode in So-Moral-VR, as continuous movement facilitates control during the assessment (though it is possible to activate the teleportation navigation mode using the joystick). Further, users can remain seated throughout the task to facilitate transferability to clinical settings—where physical space is often a limitation—and to further reduce the risk of cybersickness (Kim et al., 2023).

To prevent users from becoming disoriented in the complex VR environment and to keep the task duration as short as possible, wayfinding markers (small indicator lights) were added to guide them through the virtual city, and certain paths were intentionally blocked. These lights were designed to be subtle enough to preserve the sense of presence. A common feature in many video games, wayfinding markers are familiar to many users, unlikely to disrupt immersion and flow, and can help to reduce cognitive loads, negative affect and feelings of pressure (Irshad et al., 2021). Although these markers may seem to affect the ecological validity of the task—since participants are not entirely free to explore—they actually reduce the cognitive demands of deciding where to go and reflect a real-world parallel: in daily life, individuals usually have a goal or destination in mind and do not wander aimlessly. Hence, the use of wayfinding markers enables users to focus on the primary task, which is to respond to the moral dilemmas, rather than on navigation or exploration. Further, such features can reduce cognitive load and avoid attentional distractors which may support task validity in individuals with attentional difficulties.

Sound elements were included to ensure users direct their attention to the appropriate locations in the virtual environment, thereby supporting spatial attention (the ability to selectively pay attention to relevant locations in the visual field) (e.g., dog barking). These audio cues also enhance engagement, realism, sense of presence and task validity (Cooper et al., 2018; Rogers et al., 2018; Moore and Fallah, 2001). The audio clips included in So-Moral-VR are intentionally short in order to closely match the original version of the task (which does not include sound or words) and ensure comparability, a design consideration supported by Morasse et al. (2021).

The locations depicted in the computer version were reproduced in VR to be as detailed as possible (e.g., the store is fully stocked and includes product details) to ensure familiarity and closely resemble real-life scenarios and enhance presence (Abramson et al., 2021; Cerda et al., 2021; Pouke et al., 2019). Some complex movements were created using motion capture technology to allow more precise and fluid avatar movements (e.g., throwing a rock), further increasing immersion and realism.

Efforts were made to include avatar diversity within the environment such as representations of different ethnicities, gender, and age groups. This design feature was included to enhance inclusion, cross-cultural validity and realism. However, it is important to acknowledge that use of the task in certain groups, regions, or cultures may require further adaptations to accurately reflect regional realities (Reyes et al., 2024).

In sum, a number of design and development choices were made to ensure the ecological validity of the So-Moral-VR including storyline, scripts, visual, dynamic and auditory stimuli. These choices support the verisimilitude of the task by qualitatively demonstrating how the So-Moral-VR task closely reproduces adolescent daily life and social situations. An empirical study was then conducted to quantitatively assess ecological validity in the So-Moral-VR.

Two cross-sectional study recruitment phases were conducted with the same participant criteria. Some of the assessment measures differed between the two phases, but a core set of measures allowed pooling of the data for certain variables. Exact samples for each measure are reported in the analyses.

Sixty-two participants aged between 11 and 25 years (41.9% male, Mage = 16.72, SD = 2.87) were recruited through convenience sampling via flyers posted in recreation and community centers, social media, and high schools, colleges, and universities. Inclusion criteria were: (1) aged between 11 and 25 years and (2) fluent in French or English. Exclusion criteria were: (1) a diagnosed (as reported by the primary caregiver or the participants themselves) neurodevelopmental, acquired, or psychiatric condition such as attention deficit hyperactivity disorder, learning disabilities, schizophrenia, generalized anxiety disorder, autism spectrum disorder, or epilepsy; (2) prior acquired brain injury leading to a visit to the emergency department (e.g., concussion, traumatic brain injury, stroke); and (3) any cardiac condition. The study was approved by the local Research Ethics Committee. All participants provided written informed consent prior to participation (for participants under the age of 14 years, consent was provided by parents or legal guardian). Participants received a modest monetary compensation for participating. Participants completed an in-house demographic questionnaire to document medical history, educational level, and ethnicity.

Statistical analyses were conducted using SPSS software version 29.0. Data were examined for any violation of test assumptions (normality, linearity, and homoscedasticity). Two participants did not complete the VR task due to significant cybersickness symptoms, and one participant completed some questionnaires but did not attend the in-person assessment. Exact sample sizes are presented given slight differences in the administration protocols and missing data. To report on verisimilitude, descriptive analyses were conducted to assess immersive tendencies (Involvement, Focus and Game subscales, N = 58; Emotion subscale N = 32), cybersickness symptoms (N = 31) and realism (N = 31). To identify the strengths and weaknesses of the task in terms of verisimilitude, descriptive analyses of specific items from the SSQ (N = 31) and PQ realism (N = 31) subscales were conducted. Since literature has shown that cybersickness can negatively impact the sense of presence (Riches et al., 2019), a Pearson correlation analysis between cybersickness symptoms and realism was performed (N = 31). To support veridicality, Spearman correlation analyses (N = 59) were performed to examine the relations between SMR maturity and conduct problems, peer relationship problems, and prosocial behavior. Descriptive analyses of social desirability (N = 62) were also conducted to interpret the results in light of potential response bias (Schonfeld et al., 2005). Further Spearman correlations were conducted between social desirability and So-Moral-VR scores (SMR N = 59; SMDM N = 57), as well as with conduct problems (N = 62), peer relationship problems (N = 62), and prosocial behavior (N = 62), to better understand how social desirability may affect participant responses. Results with p < 0.05 were considered statistically significant. The strength of correlations and effect sizes were determined according to Cohen’s criteria (Cohen, 1988).

The overall mean immersive tendencies (ITQ) score was 4.511 (SD = 0.803), which corresponds to a rating between “occasionally” and “often.” Among the subscales, the highest mean score was observed for Involvement, followed by Focus and Emotion, all within the “occasionally to often” range (see Figure 3). The Game subscale showed the lowest score, corresponding to “occasionally” (see Figure 3).

Two participants did not complete the task due to cybersickness. The mean level of cybersickness symptoms was 0.583 (SD = 0.059), falling between “not at all” and “a little.” The nausea symptom subscale had a mean of 0.484 (SD = 0.599), while the oculo-motor difficulties subscale had a mean of 0.710 (SD = 0.629), both falling between “not at all” and “a little” and reflecting only mild symptom levels. The three highest mean scores on individual SSQ were: “Fullness of head”, “Eye strain” and “General discomfort” (see Figure 4), corresponding to “a little.” The lowest mean scores were reported for “Burping”, “Stomach awareness” and “Increased salivation” (see Figure 4), corresponding to “not at all.”

The mean realism score on the PQ was 3.80 (SD = 1.184), corresponding to “somewhat” and indicating a moderate sense of presence in relation to the realism of the virtual environment. The highest-scoring items were: “How involved were you in the virtual environment experience?”, “How much did the visual aspects of the environment involve you?” and “How much did your experiences in the virtual environment seem consistent with your real-world experiences?” (see Figure 5), falling, respectively, between “mildly involved” and “completely involved”/“somewhat”, and “completely”/“moderately consistent” and “very consistent.” The lowest-scoring items were: “How compelling was your sense of objects moving through space?”, “How compelling was your sense of moving around inside the virtual environment?” and “How natural did your interactions with the environment seem?” (see Figure 5), corresponding to “moderately compelling”/“borderline.” A moderate and significant association was found between cybersickness symptoms and sense of presence (R² = −0.376, p = 0.037).

Results showed a statistically significant correlation (R² = 0.283, p = 0.030) between SMR (M = 27.831; SD = 5.847) and prosocial behavior (M = 9.113; SD = 1.189), with a small effect size. The associations between SMR and conduct problems (M = 2.161; SD = 1.58; p = 0.938) and peer relationship problems (M = 1.790; SD = 1.621; p = 0.896) were not statistically significant. The mean score on social desirability was 16.774 (SD = 5.026), indicating a moderate influence of social desirability on participants’ answers. Social desirability was not significantly correlated with either SMR (p = 0.177) or SMDM (M = 6.842; SD = 1.031; p = 0.689). Further, total SMDM had a mean of 6.842 (of 9 possible points), suggesting that participants do not always provide the most socially desirable answer; 24% of the time they made a decision that is less morally or socially adapted. A small, but statistically significant, correlation was observed between social desirability and conduct problems (R² = −0.299, p = 0.018). No significant associations were found between social desirability and peer relationship problems (p = 0.481), or prosocial behavior (R² = 0.246, p = 0.054). Given the latter neared the significance threshold with a small effect size, it is possible that the association may emerge in a larger sample.

Numerous design choices were made during the development of So-Moral-VR to ensure its ecological validity, as described in the Design and Development section, and to ensure that the task closely mirrors the cognitive demands of everyday situations, thus enhancing verisimilitude. The selected dilemmas present a storyline and virtual environments (e.g., shops, school, streets) that are likely to be commonly encountered by adolescents, as demonstrated by Dooley et al. (2010). However, since the dilemmas and virtual environment were created and validated in Canada and Australia, they may be more representative of western, educated, industrialized, rich, and democratic (WEIRD) societies. The relevance of the content may be different in other regional, cultural or socio-economic settings (Reyes et al., 2024), which could affect familiarity and ecological validity and may require trans-cultural adaptations. For example, families with fewer resources may not go to restaurants. One dilemma represents a billiard game, which may not be popular in certain regions. As such, the task is likely to require trans-cultural adaptation to maintain an adequate level of verisimilitude in different socio-cultural contexts (Pinto et al., 2025).

The So-Moral-VR allows limited customization of the environment and avatars, and enhancing these aspects could increase sense of presence, body ownership, emotional presence, engagement and intrinsic motivation, all of which can contribute to stronger ecological validity (Waltemate et al., 2018; Koulouris et al., 2020). The immersive, interactive, and dynamic aspects of VR, along with the use of sound cues likely contribute to reproducing the contextual and cognitive demands of everyday situations, further enhancing verisimilitude (Sato and Yoshikawa, 2007; Paiva-Silva et al., 2016; Rymarczyk et al., 2011; Sato et al., 2008; Cooper et al., 2018; Rogers et al., 2018; Moore and Fallah, 2001). Nonetheless, technological advancements that have occurred since initial task development could offer further technical features that enhance verisimilitude. For instance, using 360-degree videos with human actors could increase the realism of both avatars and virtual environment, potentially offering more nuanced emotional expressions and more natural movement (Barreda-Ángeles et al., 2020). On the other hand, using this technology could also limit avatar and environment customization.

The So-Moral-VR task shows verisimilitude in typically developing adolescents (i.e., how closely a task mirrors the cognitive demands of everyday situations) as reflected by four indicators: the participants’ moderate to high ability to feel immersed in the virtual environment, the low incidence of cybersickness symptoms, a moderate level of realism, and the association between cybersickness and realism.

The positive associations between the SMR score and prosocial behavior supports the veridicality of the task in typically developing youth (i.e., the extent to which test scores correlate with other measures that are predictive of real-world task performance).

Importantly, no association was found between the two So-Moral-VR scores (SMR, SMDM) and social desirability, suggesting that design choices – such as using the mobile phone interface external control of verbal interventions—were probably effective in limiting social desirability bias. However, the impact of this design change was not measured directly. Nevertheless, this innovative feature may not eliminate the social desirability effect entirely, as the user is still aware that an adult is nearby. Future research could quantitatively compare the use of a human examiner versus the cell phone method, as well as study the effect of other design choices and features on social desirability. For example, a self-administered and fully anonymous remote assessment could be considered, as well as using fully automated transcription and coding, which could encourage users to self-disclose or respond more honestly.

Many neurodevelopmental conditions, acquired brain injuries, mental health problems, and social disadvantage are associated with social difficulties (Anderson et al., 2023; Beauchamp et al., 2013; Kenny and Lennings, 2007; Beauchamp et al., 2019; Fong and Iarocci, 2020) that can in part be explained by underlying deficits in social cognition. As a first step to documenting the ecological validity of the So-Moral-VR tool, only typically developing children and adolescents were recruited for the current study. Future work should determine whether the results apply to clinical populations of interest that may be at-risk for social difficulties and to evaluate other psychometric constructs, such as sensitivity, by comparing performance on the VR task between typically developing and clinical populations. It is possible that the task may require adaptations to ensure comprehension or other accessibility features in such populations.

Few reports that introduce novel psychological and neuropsychological assessment tools provide detailed background information on design and development decisions that lead to the final content and structure of the tasks. Further, there is a lack of formal psychometric validation and comprehensive evaluation of ecological validity within VR tasks (Neveu et al., 2025). This study aimed to address these gaps in relation to an innovation VR measure of SMR. However, the study also has some limitations that need to be considered. Questionnaires were used to measure presence, yet this approach is sometimes criticized because participants are required to recall how they were feeling, which can be prone to recall errors (Parsons et al., 2017). To limit this, the questionnaire was completed directly after the So-Moral-VR task. Future work should use objective or neurophysiological measures to track sense of presence (Parsons et al., 2017). Similarly, measures of veridicality also relied on self-report, which may not be the most accurate representation of everyday behaviors and subject to desirability bias. Observational behavioral measures could be used to provide a more accurate measure of naturalistic behavior. Nevertheless, social cognition rating scales are reported to better predict adaptative and daily functioning than some neuropsychological tasks (Silberstein and Harvey, 2019). Finally, the empirical analyses combined two samples of participants and therefore some measures were only used in a subset of the study sample.