Thomas theorizes that infants have cognitive representations that categorize social relationships into models from early development as a possible form of core knowledge. This theory explains and fits well with much of the existing evidence on infant social cognition. In contrast, there exist findings that cannot be explained successfully. Several studies have shown that infants exhibit early signs of morality. Many of these studies are grounded in social evaluation through preferences for agents with good characteristics (e.g., helping agents). Thomas interprets these with greater parsimony, for example, as preferences for imitators, without making assumptions about morality. However, Thomas’ theory does not explain all the findings on morality in early development. Therefore, we explain the discrepancies and consider new directions for understanding the mechanisms underlying morality in early development.

First, as Hamlin (Reference Hamlin2024) notes, not all preferences for moral agents are necessarily explained by the preferences for imitative agents. Indeed, prosocial behaviors, such as helping behaviors, do not always stem from imitative behaviors (e.g., Kanakogi et al., Reference Kanakogi, Inoue, Matsuda, Butler, Hiraki and Myowa-Yamakoshi2017), and research has shown a preference for moral characters over imitators (Chae & Song, Reference Chae and Song2018). Second, infant studies on moral foundations, such as harm or care (Haidt, Reference Haidt2007), are also unexplained by relational models such as the authority ranking relationships (dominance) proposed by Thomas. For example, Kanakogi et al. (Reference Kanakogi, Okumura, Inoue, Kitazaki and Itakura2013) observe that infants avoid the aggressor and approach the victim even when the aggressor and victim are compared to neutral agents. This tendency to approach the victim cannot be explained by the avoidance of the dominant agent, as claimed in an experiment with similar content (Thomas & Sarnecka, Reference Thomas and Sarnecka2019), because there is no comparison with a neutral agent in the study. There is some concern, if not care, for the victim. In addition, Kanakogi et al. (Reference Kanakogi, Inoue, Matsuda, Butler, Hiraki and Myowa-Yamakoshi2017) show that an agent who intervenes in a harmful interaction is preferred, unlike one who intervenes in neutral interactions; this implies a negative evaluation of harm, which represents harm aversion (although this finding may also be interpreted in terms of the authority ranking relationships [legitimate authority] of Thomas’ theory). Third, and most importantly, a seminal study has demonstrated moral behaviors of infants attempting to punish agents who have harmed others: using a gaze contingency technique, Kanakogi et al. (Reference Kanakogi, Miyazaki, Takahashi, Yamamoto, Kobayashi and Hiraki2022) show that infants punish others who have attacked others selectively by their gaze. This behavior involves not only the categorization and evaluation of social relationships but also a decision-making process aimed at delivering appropriate punishment to the agent. Therefore, this behavior cannot be explained by Thomas’ theory; we must assume some underlying motivation for driving the behavior.

How would motivation for the behavior be generated, and how would moral decision-making take place? Primitive moral emotions, such as harm aversion, may influence moral behaviors such as punishment. Emotional contagion for others’ distress has long been reported in the developmental literature, with newborns crying when they hear others crying (Simner, Reference Simner1971). It has also been reported that infants show similar expressions of distress when they see others in distress, suggesting that empathy for distress is present from the age of three months (Davidov et al., Reference Davidov, Paz, Roth-Hanania, Uzefovsky, Orlitsky, Mankuta and Zahn-Waxler2021). These findings indicate that from birth, infants are sensitive to others’ state of harm. Through the experience of observing others’ suffering, the feeling of harm aversion may be acquired by pairing negative affect (via emotional contagion) with actions that harm others (Blair, Reference Blair2007). It is also possible that the principle of “not causing harm” exists as an innate core moral standard. Regardless, it is likely that the primitive moral emotion of harm aversion is present from an early stage of development. Although it is still an open question as to how harm aversion leads to punitive behavior, harm aversion would be the most appropriate emotion for punitive motivation.

One promising mechanism for decision-making is Damasio’s somatic marker. According to the somatic marker hypothesis, the feeling of bodily changes associated with emotions helps guide decision-making in situations of complexity and uncertainty (Damasio, Reference Damasio1994). The somatic marker hypothesis can be applied to moral judgment, during which strong emotional reactions arise when one considers sacrificing one person through one’s own direct action to save other people; these emotions have been reported to significantly influence moral decision-making (Greene et al., Reference Greene, Sommerville, Nystrom, Darley and Cohen2001). Interoception (the perception of internal bodily signals) also plays a critical role in moral decision-making processes. For instance, Cui and Nakano (Reference Cui and Nakano2024) showed that brain processing of heartbeats (heartbeat-evoked potentials: HEP) arising from the negative evaluations of harming others, regardless of the type of harm (direct or indirect), influences decisions not to harm others. From a developmental perspective, it has been shown that even three-month-old infants can discriminate between stimuli that move synchronously with their heartbeat and those that do not (Tünte et al., Reference Tünte, Hoehl, Wunderwald, Bullinger, Boyadziheva, Maister and Kayhan2025). Furthermore, in five-month-old infants, cardiac discrimination ability is correlated with HEP amplitudes when viewing emotional facial expressions (Maister, Tang, & Tsakiris, Reference Maister, Tang and Tsakiris2017). These findings suggest that infants have sensitivity to their interoception. Given that HEP has been reported to influence moral decision-making in adults (Cui & Nakano, Reference Cui and Nakano2024), they also suggest that interoceptive sensitivity may be relevant to decision-making even during early development. Considering these findings, it is plausible that infants make decisions to punish those who harm others based on the perception of bodily responses arising from intuitive negative evaluations of harmful actions.

To date, no study on infant morality has examined the relationship between morality and bodily responses or interoception. However, if bodily responses and interoception are involved in morality in early infancy, it sheds new light on the underlying mechanisms, from the categorization and evaluation of social relationships to moral decision-making. This complements both Thomas’ theory and Hamlin’s moral core theory (Hamlin, Reference Hamlin2013; Woo, Tan, & Hamlin, Reference Woo, Tan and Hamlin2022). A more comprehensive understanding of morality during early development would open new research directions in infant morality.