Autism spectrum disorder (hereafter autism) is a neurodevelopmental condition characterized by differences in social-communication skills and the presence of repetitive or inflexible patterns of behavior or interests (American Psychiatric Association [APA], 2013). The differences in social-communication skills have been attributed to differences in social cognition including commonsense psychology (e.g., Baron-Cohen, 2000 for review), which is the general human tendency to make sense of other people’s actions by attributing psychological states to them (Moore, 2006). Moral psychology is a related aspect of social cognition that involves reflecting on norms for how to treat other people and non-human animals and coexist with them in society.¹ Moral psychology is often assumed to require typical commonsense psychology skills for its development. Thus, aspects of moral psychology have been theorized by some to be underdeveloped among autistic individuals (e.g., Shoemaker, 2015). In this study we aimed to describe autistic moral thinking from a strengths-based perspective rather than adopting a deficits lens.

Autistic² individuals often have difficulties interacting with others and with forming and maintaining friendships (Howlin et al., 2013). These difficulties are bidirectional (Milton, 2012) and relational (Morrison et al., 2020). To illustrate, just as autistic individuals may struggle to understand social nuances, neurotypical individuals may misinterpret the intentions of autistic individuals (Sheppard et al., 2016). In addition to communication challenges present in interactions between autistic and neurotypical individuals, difficult interpersonal situations faced by autistic individuals could arise in part from differences in moral development and subsequent moral reasoning. Li et al. (2014) found that autistic children aged 6–12 years did not modulate cooperative behavior in response to immoral acts of their partners whereas neurotypical children cooperated less with perceived moral wrongdoers. Autistic children’s failure to modulate behavior in response to others’ moral transgressions may lead neurotypical peers to like autistic children less, making it more difficult for the latter to develop and maintain friendships or possibly contributing to disproportionate rates of bullying victimization among autistic youth (Maïano et al., 2016).

Rationalist accounts of moral development (e.g., Piaget, 1932; Kohlberg, 1969, 1971) suggest that commonsense psychology skills, such as perspective-taking, are required for moral development. Piaget (1932) found that young neurotypical children use reasoning based on rules, whereas older children and adults rely more on others’ intentionality when reasoning about morality. The ability to consider intentions when judging moral culpability has been a key focus in research on morality in autism, wherein subtle differences have often been construed as deficits (for review see Garcia-Molina and Clemente-Estevan, 2019). However, such research has not shown robust connections between commonsense psychology and moral judgments among autistic individuals (for review see Dempsey et al., 2019). As such, autistic individuals appear to use heuristics for making moral judgments that rely less heavily on commonsense psychology than seen in neurotypical individuals, presenting a challenge to the single developmental pathway to moral maturity posited by rationalist accounts.

In contrast to rationalist theories, Haidt (2001) proposes an “intuitionist” model of moral reasoning wherein humans have evolved the capacity to develop moral intuitions in at least five domains or foundations (i.e., authority/respect, care/harm, fairness/reciprocity, in-group/loyalty, and purity/sanctity; Haidt and Joseph, 2008). The authority/respect foundation concerns virtues of obedience and deference; care/harm is related to kindness and care; fairness/reciprocity is linked to justice and trustworthiness, in-group/loyalty is defined by loyalty, patriotism, and self-sacrifice; and purity/sanctity concerns virtues of temperance, piety, and cleanliness (Haidt and Joseph, 2008).

To the authors’ knowledge, only one study has explicitly investigated Moral Foundations Theory holistically among neurotypical children (Peverill, 2020). In this study, 4-year-olds showed moral sensitivity to the authority/respect, care/harm, fairness/reciprocity, and purity/sanctity foundations, but not to the in-group loyalty foundation. Other researchers have provided support for children’s sensitivity to moral transgressions related to Haidt’s (2001) five moral foundations. Concerns for authority/respect have not been studied in infancy but there is evidence that elementary school children prefer peers who obey authority to peers who disobey authority (Martín-Antón et al., 2016). There is a rich literature supporting children’s sensitivity to care/harm violations emerging as early as at 3 months (e.g., Vaish et al., 2010; Hamlin and Wynn, 2011; Kanakogi et al., 2013, 2017). Fairness/reciprocity has also been extensively studied among young children, which has shown that children as young as 16 months anticipate and prefer equal resource allocation (e.g., Geraci and Surian, 2011; Schmidt and Sommerville, 2011). In terms of in-group/loyalty, Jin and Baillargeon (2017) found that 17-month-old infants anticipated that in-group members would help each other but that individuals who did not belong to the same group would not help each other. Finally, research investigating early emergence of the purity/sanctity foundation is limited but suggests that disgust and unnaturalness contribute to children’s moralization of novel acts (Rottman and Kelemen, 2012).

Crucially, Haidt’s theory does not rely on commonsense psychology for the development of moral judgment. This theory could help make sense of moral psychology among autistic individuals. For instance, autistic adults have described their condition as resulting in greater loyalty, honesty, and empathy for other autistic adults and non-human animals than for their neurotypical peers, which could affect their moral foundations profile (Russell et al., 2019). Zalla et al. (2011) found that autistic participants judged disgusting acts as equally morally wrong as harmful acts, in contrast with neurotypical participants who judged disgusting acts as less morally wrong than harmful ones. This difference could suggest greater prioritization of the purity/sanctity foundation among autistic compared with neurotypical participants. Additionally, some studies have shown that autistic youth, compared with neurotypical youth, place a greater emphasis on authority and rules than on abstract principles such as justice (Takeda et al., 2007; Senland and Higgins-D’Alessandro, 2016; Garon et al., 2018). This difference could signify greater emphasis on the authority/respect moral foundation among autistic compared with neurotypical individuals.

Moral reasoning, i.e., justification for one’s moral judgments, differs between autistic and neurotypical children despite similar moral judgments. When asked to justify moral judgments, autistic youth have been found to reiterate moral vignettes (Grant et al., 2005) or provide concrete, less elaborate justifications more often than their neurotypical peers (Shulman et al., 2012; Garcia-Molina et al., 2019). Autistic children’s justifications tend to be more rule-bound and focused on consequences rather than intentions, compared with those provided by neurotypical children (Takeda et al., 2007; Fadda et al., 2016; Garon et al., 2018; Garcia-Molina et al., 2019).

Differences in moral reasoning between autistic and neurotypical individuals despite relatively similar moral judgments could be explained in part by studies suggesting that moral and other forms of reasoning are post hoc rationalizations of intuitive judgments (Haidt, 2001; Mercier and Sperber, 2011). Rule- and consequence-oriented moral reasoning in autistic children may therefore indicate post hoc rationalizations for moral judgments that may be explained by the relatively concrete thought processes often observed in autism (Hobson, 2012), as opposed to underdeveloped moral intuitions.

Neurotypical children are sensitive to retributive justice from infancy. Hamlin et al. (2011) found that infants as young as 8 months old preferred agents who acted negatively toward antisocial individuals as opposed to agents who acted positively toward antisocial individuals. Geraci and Surian (2021) found that 21-month-old toddlers anticipated punishment for individuals who did not defend an innocent victim whereas they did not anticipate punishment for individuals who did help an innocent victim. Neurotypical toddlers’ expectations of punishment for non-defenders extends to anticipating corporal punishment but seems restricted to when the object being defended (or not) is agentic rather than inanimate (Geraci, 2021). Children aged 3–7 years have shown a tendency to endorse punishment for wrongdoers even at a cost to themselves (Yudkin et al., 2020; Marshall et al., 2021).

Some differences have been found in assigning blame and punishment to transgressors by autistic compared with neurotypical children. To illustrate, Li et al. (2018) found that autistic children aged 6–12 years were more likely to recommend a child be punished for hitting another child than were neurotypical children. Some researchers have found that blame was less tempered by intent among autistic compared with neurotypical children. For example, Salvano-Pardieu et al. (2016) found that autistic adolescents did not differ in their judgments of the level of seriousness of stabbing compared with punching, whereas neurotypical adolescents deemed stabbing as worse than punching. In contrast, Akechi et al. (2018) found that assigning blame did not differ between autistic and neurotypical youth. Rogé and Mullet (2011) found that use of intent in tempering blame judgments increased with age for both autistic and neurotypical participants; it could be that Akechi et al. (2018) did not find group differences due to their sample’s large age span (7–24 years).

To the authors’ knowledge, moral foundations theory has only been investigated in one study of autistic individuals—a qualitative investigation of moral foundations theory among autistic adults (Dempsey et al., 2020). The aim of the current mixed methods study was to investigate moral decision making and reasoning among autistic youth from the perspective of moral foundations theory. Our first hypothesis was that authority/respect and purity/sanctity moral foundations would be endorsed more strongly by autistic than by neurotypical children. Our second hypothesis was that autistic children would be more likely to indicate that children depicted in morality vignettes should be punished, particularly for social norms violations, compared with neurotypical children. We conducted this analysis while controlling for alexithymia traits, i.e., difficulty identifying and describing emotional states (Griffin et al., 2016), as these have been shown to influence moral decision making in autism (Patil et al., 2016). Our third hypothesis was that moral reasoning by autistic children would more often rely on rules and outcomes than among neurotypical children, as assessed in a subset of our participants using qualitative analysis.

This was the first investigation of moral reasoning and judgment among autistic children from the perspective of moral foundations theory (Haidt, 2001). Our aim was to begin to develop an understanding of similarities and differences in moral foundation priorities between autistic and neurotypical children, to investigate recommendations for punishments for moral transgressions between the two groups, and to describe the children’s moral reasoning.

Our first hypothesis, that autistic children would be more disapproving of authority/respect and purity/sanctity foundations violations than neurotypical children, was not supported. Other studies have shown that older autistic individuals may be more sensitive to violations of these foundations than neurotypical adults (Zalla et al., 2011; Senland and Higgins-D’Alessandro, 2016). Perhaps differences in judgments regarding purity/sanctity among autistic individuals do not emerge until adolescence or adulthood. An alternative explanation for our divergent results is that our study employed more scenarios for each form of transgression (four versus two each in the Zalla et al., 2011, study). Perhaps this generated more representative responses leading to similar judgments of the foundations for both groups.

In terms of lack of differences in the authority/respect foundation judgments, past research has assessed moral values based on Kohlberg’s 1969, 1971 stage-based developmental moral hierarchy (Takeda et al., 2007; Senland and Higgins-D’Alessandro, 2016; Garon et al., 2018). This theoretical framework leads to interpretations based on the stage of moral maturity at which the child’s moral reasoning is deemed to lie. In contrast, we asked children to rate how bad each transgression was across all five moral foundations and without setting up a hierarchy. This methodological difference could have affected our results. Alternatively, our scale may have contributed to a ceiling effect. Indeed, 104 authority question responses reached the maximum “how bad is it” score. Autistic children may have indicated significantly more disapproval of authority transgressions had our scale permitted a broader range of responses.

Our second hypothesis that autistic children would be more likely than neurotypical children to recommend punishment for harmless social norms violations was partially supported, consistent with past findings (Rogé and Mullet, 2011; Salvano-Pardieu et al., 2016; Li et al., 2018). This could be due in part to the social communications differences observed among autistic children (American Psychiatric Association [APA], 2013). It is possible that they are more prone to making harmless social norms violations than their neurotypical peers and may experience correction or even punishment for these differences. Autistic children may in turn assume that other children who behave similarly will or should be punished for such behavior. It is important to note that the judgment of how bad the transgression was deemed by participants was a stronger predictor of endorsing punishment than was being autistic. Indeed, the effect of diagnostic group was only trending as significant in the general linear mixed model (i.e., only significant prior to correction for multiple testing).

Our third hypothesis, that autistic children’s moral reasoning would defer to rules and consequences more often than that of neurotypical children, was not supported—autistic children adduced these reasons less frequently than neurotypical children. This result, which is inconsistent with past research (Fadda et al., 2016; Garon et al., 2018; Garcia-Molina et al., 2019), could be due to differences in coding methodology. Most studies assessing autistic children’s moral reasoning have used pre-specified coding schemes designed to contrast consideration of intentionality with more concrete reasoning. Our inductive thematic analysis may have yielded more diverse themes without forcing children’s responses into pre-existing categories.

Consistent with past research (Grant et al., 2005; Shulman et al., 2012), autistic children sometimes offered responses that were less elaborate than those of neurotypical children; that is, they more often said that acts were “just okay” without elaboration. However, this difference was not apparent when children were asked to justify their judgments that actions in the vignettes were bad. Interestingly, neurotypical children’s responses to the vignettes were sometimes more elaborate, but more often made little sense. For instance, when asked why it would be bad to wear pajamas to school, one neurotypical child reasoned that one might need to swim in them, which would be uncomfortable. Coders judged these explanations to be post hoc rationalizations. Given the theorized social element of post hoc rationalizations (Haidt, 2001; Mercier and Sperber, 2011), perhaps the autistic children’s lesser use of such justifications reflects their social communication differences (American Psychiatric Association [APA], 2013). Alternatively, Grant et al. (2005) found that appropriate justifications for moral judgments were positively correlated with age among autistic and neurotypical children. The neurotypical children were younger than the autistic children in our subsample, although this difference did not reach statistical significance. Still, this age difference could have contributed to neurotypical children’s greater propensity to offer these types of justifications. A further possible explanation for the finding that autistic children provided less elaborate responses than the neurotypical children could be the lower verbal cognitive abilities among the former group.

A further finding of our qualitative analysis is that autistic children reported sadness (e.g., in response to a vignette depicting a child eating soup with a fork rather than a spoon), as well as uncertainty regarding their emotions more often than did their neurotypical peers. This could be due in part to higher rates of alexithymia among the autistic children in our sample, consistent with past research (Hill et al., 2004).

In general, our results suggest that judgments regarding transgressions against moral foundations made by neurotypical and autistic children were more similar than different, despite differences found in previous research (e.g., Garon et al., 2018), and despite markedly lower verbal reasoning skills among the autistic children. An explanation is that differences in social cognition in autism could present as differences in moral values when assessed based on rationalist theories, but that differences in valuing of the five moral foundations are not apparent when studied explicitly, as we did. To illustrate, autistic children have been found to focus more on rules and consequences than neurotypical children when judging the severity of moral transgressions (e.g., Fadda et al., 2016). In our Introduction, we interpreted such differences as possibly reflecting variations in valuing of the authority/respect foundation among autistic compared with neurotypical children. We argue that such differences may instead reflect relatively concrete (Minshew et al., 2002) or rigid (Poljac and Bekkering, 2012) thinking in autism without denoting higher prioritization of authority/respect over other moral foundations.

This interpretation, coupled with differences in verbal reasoning between groups, is consistent with our suggestion that moral psychology may develop through a pathway other than the commonsense psychology route suggested by rationalist theories. This interpretation also fits with results from Dempsey et al. (2020) that the autistic adults in their study, who were generally politically left-wing, emphasized the importance of care/harm and fairness/reciprocity above the other moral foundations posited by Haidt (2001). This profile of moral foundations has been observed to differentiate politically left- from right-wing ideologies (Graham et al., 2009). Given the homogeneity of political attitudes espoused by the parents of children in the current study, it is possible that broader cultural influences may be more responsible for the development of sensitivity to moral transgressions across foundations than are differences in social cognition in autism. However, given the novelty of the present study, and the fact that little research exists on the development of moral foundations theory even among neurotypical children (see Peverill, 2020), this interpretation is speculative. Resolving this issue would require testing with more politically heterogeneous groups of families of autistic and neurotypical children.

An interesting avenue for further exploration of moral foundations theory in autistic and neurotypical youth would be to investigate whether differing profiles of moral foundations predilections are present between the two groups. Researchers could add gender to these analyses in a larger sample in future to investigate the role of gender on moral decision making among autism and neurotypical individuals. This type of profile analysis has been conducted in studies with larger, more politically and culturally heterogeneous samples (e.g., Graham et al., 2009; Vecina and Chacón, 2019). Future research should also clarify whether autistic children are equally likely as neurotypical children to anticipate punishment for agents who have not defended an innocent victim in accordance with the care/harm foundation of moral foundations theory (Haidt, 2001; Geraci, 2021; Geraci and Surian, 2021).

Understanding the developmental course of moral foundations predilections, severity of judgments, and recommendations for punishment could offer guidance for interventions for autistic and neurotypical children to promote clarity and improve communication among children on the continuum from autistic to neurotypical. Such an intervention could take the form of a social narrative (Gray and Garand, 1993) for autistic and neurotypical children that illustrates the unique developmental challenges faced by some autistic children viz. punishment and morality. Another possibility would be a psychoeducational intervention for parents and educators illustrating the potential costs of applying negative consequences for harmless norms violations among autistic children, though additional research would be required first to validate this connection, then to test the intervention.

Overall, our quantitative and qualitative findings suggest that autistic and neurotypical children evaluate moral transgressions across Haidt’s (2001) five moral foundations similarly, despite previous research suggesting possible group differences in the relative salience of the authority/respect and purity/sanctity foundations. This was the case despite some differences in their moral reasoning and in the emotions elicited in response to the scenarios we presented. The most prominent difference that emerged from our study was autistic children’s greater likelihood of recommending punishment for relatively minor transgressions.

Insufficient understanding of autistic individuals has been cited by members of the autism community as a barrier to fitting into society (Pellicano et al., 2014). As such, autistic individuals and stakeholders (e.g., families of autistic people) have called for researchers to focus on topics that affect autistic individuals’ day-to-day lives (Pellicano et al., 2014). Our study was the first to investigate moral foundations theory in autistic and neurotypical children. Our findings that autistic children’s moral reasoning differs only subtly from that of neurotypical children contributes to our understanding of moral agency in autism and challenges assertions by some that autistic individuals have limited moral agency (e.g., Richman, 2018). Future studies could use longitudinal methods to track the development of moral foundations predilections and recommendations for punishment to further refine our understanding of moral development in typical development and in autism.

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 Research Ethics Board of the IWK Health Centre, Halifax, Nova Scotia, Canada. Written informed consent to participate in this study was provided by the participants’ legal guardian/next of kin.

ED conceptualized the project, collected and interpreted data, and wrote the manuscript. CM, SJ, SS, and IS contributed equally to collaborating with ED to conceptualize the research project, interpret data, and revise the manuscript. All authors contributed to the article 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.