Baumard et al. ask, “What makes humans moral beings?” (target article, Abstract). They propose an answer that involves people adapting to their social environment by sharing the costs and benefits of cooperation fairly and suggest a moral sense evolved to guide the distribution of gains resulting from cooperation. But their so-called moral sense is actually what most people would refer to instead as a fairness sense (a point the authors acknowledge at the end of the article). By redefining the term “morality” to mean fairness – and, indeed, fairness narrowly conceived as involving balancing inputs into and outcomes from interactions amongst strangers – they turn a blind eye to the rich and complex relational contexts in which we normally interact with other people and in which most of our moral concerns naturally arise. (They also sidestep using the term moral to refer to our relationally dependent normative obligations to benefit and not to harm our fellow humans, which is more in concert with the lay use of the term and with our own sense of it as well.)

We challenge the assumption Baumard et al. make that humans have just one general moral strategy they follow in interacting with other people. Psychological research in the relationships field (including years' worth of research by one of us, but that of many others as well) indicates clearly that people do not follow the same type of cooperative norm in all their relationships. Instead, they utilize different norms in different relationships and at different relationship stages (Clark & Beck Reference Clark, Beck, Fincham and Cui2011; Clark & Mills Reference Clark and Mills1979; Reference Clark and Mills1993; Reference Clark, Mills, Van Lange, Kruglanski and Higgins2012; Clark et al. Reference Clark, Mills and Powell1986; Reference Clark, Mills and Corcoran1989). In some (“communal”) relationships people keep track of partner needs and benefit partners non-contingently in response to their needs, desires, and goals. In other (“exchange”) relationships, people follow a tit-for-tat strategy. Adhering to either norm can be considered moral, depending on the type of relationship. For instance, it is wrong for a parent to neglect to feed his child but not equivalently bad to neglect to feed a stranger. It is also wrong for a person not to pay a grocer for an orange but not wrong for a child not to repay his parent for the orange. Relational context clearly matters.

Moral theories based solely on empirical research involving interactions between strangers (e.g., most of the economic game research on which Baumard et al. rely) oversimplify the answers to “how” and “why” people behave morally, because they fail to include observations of people who have or desire a variety of types of relationships with others. Since most moral behavior occurs in the context of our relationships with people whom we know personally and/or about whom we care deeply, psychological accounts of morality relying on studies of interactions between strangers fail to account for the majority of contexts in which moral issues actually arise and matter. We recognize that economic games are a current favorite among economists, researchers in business schools, and some psychologists. They are neat, straightforward, and easily modeled mathematically. However, the literature on human cooperation extends far beyond the literature on behavior in these games (see, e.g. for instance, Clark & Lemay Reference Clark, Lemay, Fiske, Gilbert and Lindzey2010; Tyler Reference Tyler2010; and the now burgeoning literature in relationship science generally).

Berscheid and others have argued, eloquently and convincingly, that we cannot truly understand human behavior if we remove it from relational context (Berscheid Reference Berscheid1999; Reis et al. Reference Reis, Collins and Berscheid2000). Bugental has argued that social algorithms vary by social function (Bugental Reference Bugental2000). We agree with both. The prospect of future interactions, the varying nature and function of those interactions, and individuals' personal relationship histories make a huge difference to people's behavior, impacting people's motivations, emotions, and decisions and, most relevant to this commentary, the very nature of their cooperation with one another.

Baumard et al. do briefly review how performance on economic games is influenced by relationship histories (see their discussion of the impact of participants' prior interactions and of findings reported by Cronk [Reference Cronk2007] and Cronk & Wasielewski [Reference Cronk and Wasielewski2008]). Yet they fail to embrace the implications of these observations and to incorporate them into their theorizing.

The vast extant social-psychological literature relevant to cooperation could inform Baumard et al.'s theory. Here are just four of many lessons to be garnered from that literature. First, as already noted, there is more than one flavor of mutualism supporting human cooperation. Sometimes people set forth explicit contracts with one another that specify their roles and duties. Sometimes (as Baumard et al. suggest) people form more implicit cooperative relationships in which they expect to benefit in direct proportion to their contributions. Sometimes people implicitly agree to assume a degree of responsibility for one another's welfare and they non-contingently provide benefits in response to needs and desires as those needs and desires arise (and, to make matters more complicated, the degree of assumed responsibility varies by relationship and can be symmetrical or asymmetrical; Mills et al. Reference Mills, Clark, Ford and Johnson2004). Second, the actions we take to form, strengthen, or repair mutualistic relationships differ by relationship stage (Clark & Beck Reference Clark, Beck, Fincham and Cui2011). To win people over to a cooperative communal relationship, for instance, we begin by offering more benefits than we request, but, if a commitment to a relationship is made, offers and requests even out if needs are even (Beck & Clark Reference Beck and Clark2009). Third, individual differences matter – a lot. We enter new relationships with the baggage of our relationship histories in tow. These histories impact our confidence that forming mutualistic relationships with others will work out and, consequently, our willingness to enter various forms of such relationships in the first place. If trust in others is very low, we may prefer very explicit rather than implicit agreements to exchange this for that. If trust is high, we can risk adding some truly need-based communal relationships to our mix of mutualistic relationships (cf. Murray et al. Reference Murray, Holmes and Collins2006). Fourth, it appears that the procedures or relational processes involved in determining how benefits are allocated in relationships often matter more to our judgments of whether behavior is moral and our willingness to remain in interactions than the does the nature of actual allocations (Tyler Reference Tyler2010).

In sum, any complete account of morality must take into consideration all of our varied relationships and the literature on cooperation broadly conceived. Studying only strangers and postulating just one type of mutualism will result in an incomplete theory of mutualisms.