Cushman suggests that his individualistic account of rationalization is complementary to a social account of rationalization based on reputation management. However, the two accounts make conflicting predictions, so that data about rationalizations should help tell which account better explains features of rationalizations.

If our ability to rationalize evolved by improving practical decision making, its expression (when we feel the need to rationalize) and content (how we rationalize) should not be strongly modulated by the presence of others, and it should not (as a rule) lead to the acceptance of inaccurate beliefs (since those are unlikely to improve practical decision making). By contrast, if our ability to rationalize evolved by serving reputation management ends, its expression and content should be strongly modulated by the presence of others, and it could lead to the acceptance of inaccurate or practical maladaptive beliefs, in a trade-off between the practical and the social value of the beliefs (see Kurzban & Aktipis Reference Kurzban and Aktipis2007; Mercier Reference Mercier2012).

On the first point – when people feel the need to rationalize – we rely on results from the cognitive dissonance literature, given that cognitive dissonance is a driver of the need to rationalize. Experiments have shown that the feeling of dissonance is highly modulated by social context: Participants are less likely to experience dissonance and to rationalize their actions when these actions are private rather than public (e.g., Tice Reference Tice1992; see also Leary Reference Leary1995; Tedeschi & Rosenfeld Reference Tedeschi and Rosenfeld1981). More generally, people seem more concerned with appearing consistent than with being consistent (Tedeschi et al. Reference Tedeschi, Schlenker and Bonoma1971).

Second, the content of rationalizations depends more on what we think others will deem acceptable than on the likelihood of improving practical decision making, as if the rationalizations were generated by an internal “press secretary” (Kurzban & Aktipis Reference Kurzban and Aktipis2007; see also Haidt Reference Haidt2001; Nisbett & Wilson Reference Nisbett and Wilson1977). This would explain, for instance, why cognitive dissonance manipulations only affect explicit beliefs (Gawronski & Strack Reference Gawronski and Strack2004). If the rationalizations that cognitive dissonance gives rise to aimed at improving our practical decision making, they should affect implicit and explicit beliefs, since both matter for practical decision making. By contrast, if these rationalizations serve social goals, only explicit beliefs – the ones we can share – matter. Moreover, the influence rationalizations have on our explicit beliefs can be practically detrimental. In all the standard cognitive dissonance paradigms, dissonance reduction leads to less accurate beliefs – for example, that a truly boring task isn't really that boring (Festinger & Carlsmith Reference Festinger and Carlsmith1959).

When rationalizations affect our actions, they often do not do so in a way that is compatible with Cushman's account. Many experiments on reason-based choice (Simonson Reference Simonson1989; for reviews, see Mercier & Sperber Reference Mercier and Sperber2011; Shafir et al. Reference Shafir, Simonson and Tversky1993) have shown (1) that often participants make worse decisions, from a purely practical point of view, when they engage in rationalizations, and (2) that these decisions fit with the expectations participants have of what decision will look best. In the classic reason-based choice study (Simonson Reference Simonson1989), asking participants to justify their decisions (and thus to provide rationalizations) pushed them to make suboptimal decisions (as they were influenced by logically irrelevant factors). However, the participants thought these arguably less practical rational decisions were easier to justify and less likely to be criticized. In another study, Briley et al. (Reference Briley, Morris and Simonson2000) have shown that asking participants in Asia and the United States to justify their decisions affected these decisions, but in opposite ways as a function of the culture to which they belonged. In Hong Kong, participants who provided rationalizations were more likely to choose a compromise option, while participants in the United States were less likely to do so. In each case, the deviations were in line with the society's cultural values, and thus with which rationalizations would be more likely to be accepted. In both studies, rationalizations influenced participants’ decisions in a way that made them arguably less practically rational, but more socially acceptable (see also Baumeister Reference Baumeister1982; Baumeister & Cairns Reference Baumeister and Cairns1992).

One of the strengths of Cushman's account is that it explains why people's behavior would be guided by past rationalizations: If rationalizations were a purely social tool, why let our future behavior be guided by them? Here, we suggest that Cushman neglects one of the main reasons why rationalizations are socially effective: They commit the speaker to the rationalizations that the speaker offered. By providing rationalizations, we convey the following information: (1) I share the same values as you; (2) I made this decision because of these values; and (3) in the future, I will keep following these values. If (3) were absent, that is, if speakers didn't commit to their rationalizations, the rationalizations would be largely worthless. By being committed to what they say, senders implicitly acknowledge that if their message is found to be unreliable, they will pay reputational costs (e.g., see Mazzarella et al. Reference Mazzarella, Reinecke, Noveck and Mercier2018; Vullioud et al. Reference Vullioud, Clément, Scott-Phillips and Mercier2017).

The fact that we are committed (to some extent) to our rationalizations explains why we stick to them: If we didn't, we'd suffer social costs. Rationalizations are thus no different from any other form of commitment. For example, when we morally condemn a behavior, we commit to not engaging in this behavior ourselves, and we dislike people who engage in behaviors they have previously condemned (Jordan et al. Reference Jordan, Sommers, Bloom and Rand2017). Indeed, our need to appear consistent has long been associated with reputation management (Leary Reference Leary1995; Sperber Reference Sperber and Sperber2000; Tetlock Reference Tetlock1992).

Last but not least, if our ability to rationalize had evolved to improve practical decision making, the human exception would be quite puzzling from an evolutionary perspective. Being influenced by multiple “non-rational processes” (target article, sect. 1.1, para. 7) and having to make complex decisions are not exclusive to humans. Our social account of rationalization avoids this pitfall: Rationalization is human-specific because the human social niche has exerted selective pressures on our ability to manage our reputation so as to compete in biological markets (Mercier & Sperber Reference Mercier and Sperber2017).