Social obligation begins far before people establish explicit cooperative relationships. Research on trust suggests that people feel obligated to trust other people even at zero acquaintance, thus trusting complete strangers even though they privately expect to be exploited. Such obligations promote mutually beneficial behavior among strangers and likely help people build goodwill needed for more long-lasting relationships.

Tomasello's account of the origins and nature of moral obligation rightly emphasises the key roles of social relations and a cooperative sense of “we.” However, we suggest that it overlooks the complexity of those social relations and the resulting prevalence of a divided “we” in moral social groups. We argue that the social identity dynamics that arise can lead to competing obligations in a single group, and this has implications for the evolution of obligation.

An adequate moral psychology of obligation must bear in mind that although the “sense of obligation” is psychological, what it is a sense of, moral obligation itself, is not. It is irreducibly normative. I argue, therefore, that the “we” whose demands the sense of obligation presupposes must be an ideal rather than an actual “we.”

For Tomasello's proposed ontology of the human sense of moral obligation, observations of early moral language may provide useful evidence complementary to that afforded by experimental research. Extant reports of children's everyday moral talk reveal patterns of participation and content that accord with the proposal and hint at extensions addressing individual differences.

We are very happy that someone has finally tried to make sense of rationalization. But we are worried about the representational structure assumed by Cushman, particularly the “boxology” belief-desire model depicting the rational planner, and it seems to us he fails to accommodate many of the interpersonal aspects of representational exchange.

Rationalization through reduction of cognitive dissonance does not have the function of representational exchange. Instead, cognitive dissonance is part of the “psychological immune system” (Gilbert 2006; Mandelbaum 2019) and functions to protect the self-concept against evidence of incompetence, immorality, and instability. The irrational forms of attitude change that protect the self-concept in dissonance reduction are useful primarily for maintaining motivation.

In making the case that “rationalization is rational,” Cushman downplays its signature liability: Rationalization exposes a person to the hazard of delusion and self-sabotage. In paradigm cases, rationalization undermines instrumental rationality by introducing inaccuracies into the representational map required for planning and effective agency.

Cushman raises the intriguing possibility that rationalization accesses/constructs intuitions that are not otherwise cognitively available. However, he substantially over-reaches in arguing that rationalization is mostly right on average, based on claims that the process must have emerged adaptively. The adaptiveness of “bounded rationalization” is domain specific and is unlikely to be adaptive in a large number of important applications.

In this commentary, we ask when rationalization is most likely to occur and to not occur, and about where to expect, and how to measure, its benefits.

Cushman's theory has implications for the philosophical debate about the nature of folk psychological states, for it entails realism about propositional attitudes. I point out a tension within his view and suggest a different view upon which rationalization emerges as a consequence of the adaptiveness of mentalizing. This alternative avoids the strong metaphysical implications of Cushman's theory.

Cushman claims that post hoc rationalization of habitual behavior can improve future reasoning. His characterization of habits includes two components: (1) habitual behavior is a non-rational process, and (2) habitual behavior is sometimes rationalized. We argue that Cushman fails to show any habits that are apt targets for rationalization. Thus, it's unclear when – if ever – rationalizing habits would improve reasoning.

We present a theoretical and an empirical challenge to Cushman's claim that rationalization is adaptive because it allows humans to extract more accurate beliefs from our non-rational motivations for behavior. Rationalization sometimes generates more adaptive decisions by making our beliefs about the world less accurate. We suggest that the most important adaptive advantage of rationalization is instead that it increases our predictability (and therefore attractiveness) as potential partners in cooperative social interactions.

While we agree in broad strokes with the characterisation of rationalization as a “useful fiction,” we think that Fiery Cushman's claim remains ambiguous in two crucial respects: (1) the reality of beliefs and desires, that is, the fictional status of folk-psychological entities and (2) the degree to which they should be understood as useful. Our aim is to clarify both points and explicate the rationale of rationalization.

This commentary extends the target article's useful concepts to consider collective instances of representational exchange. When groups collectively rationalize their actions, entire networks of beliefs and desires can be created and maintained in the form of shared moral narratives and system-justifying ideologies. These collective rationalization cases illustrate how adaptive advantages can come at the expense of the truth.

In this commentary, I highlight the relevance of Cushman's target article for the popular dual-process framework of thinking. I point to the problematic characterization of rationalization in traditional dual-process models and suggest that in line with recent advances, Cushman's rational rationalization account offers a way out of the rationalization paradox.

Cushman uses rationalization to refer to people's explanations for their own actions. In system justification theory, scholars use the same term to refer to people's efforts to cast their current status quo in an exaggeratedly positive light. We try to reconcile these two meanings, positing that system justification could result from people trying to explain their own failure to take action to combat inequality. We highlight two novel and contested predictions emerging from this interpretation.

Cushman argues that the function of rationalization is to attribute mental representations to ourselves, thereby making these representations available for future planning. I argue that such attribution is often not necessary and sometimes maladaptive. I suggest a different explanation of rationalization: making representations available to other agents, to facilitate cooperation, transmission, and the ratchet effect that underlies cumulative cultural evolution.

Cushman argues that “rationalization is rational.” We show that there is reasonable empirical clinical and forensic psychological evidence to support viewing rationalization as a quite suboptimal defense mechanism. Rationalization has been found to be associated not only with poorer emotional development, but also with a broad range of antisocial behavior, including not only shoplifting, but also pedophilia and murder.

Cushman's rationalization account can be extended to cover another part of his portrayal of representational exchange: thought experiments that lead to conclusions about the self. While Cushman's argument is compelling, a full account of rationalization as adaptive will need to account for the divergence in rationalizing one's actions compared to the actions of others.

Although rationalization about one's own beliefs and actions can improve an individual's future decisions, beliefs can provide other benefits unrelated to their epistemic truth value, such as group cohesion and identity. A model of resource-rational cognition that accounts for these benefits may explain unexpected and seemingly irrational thought patterns, such as belief polarization.