Imagine that a woman moves to a new town, and a neighbor stops by to introduce herself and deliver homemade cookies as a welcome present. How would we expect the woman to behave toward the neighbor in their next interaction? In what ways would considerations of reciprocity guide our expectations about this interaction? Our research asked 15-month-old infants similar questions.

In our experiments, infants watched live scenarios involving two women, agent1 and agent2, who appeared to be strangers to one another. To start, agent1 directed either positive or negative actions toward agent2: She gave her cookies or stole cookies from her. Later, agent1 worked toward completing a goal, and agent2 chose to help her, to ignore her need for help, or to harm her. Prior evidence indicated that without initial actions by agent1, infants would view all three actions by agent2 (helping, not helping, or harming agent1) as equally acceptable.^1-2 Of interest was whether these judgments would change, in accordance with reciprocity, if agent1first acted positively or negatively toward agent2. We obtained two main findings (see Figure 1).

First, by 15 months, infants already possess an expectation of reciprocity. After agent1 directed positive actions toward agent2, infants were surprised if agent2 later chose to harm agent1. Conversely, after agent1 directed negative actions toward agent2, infants were surprised if agent2 later chose to help agent1. These surprise responses were eliminated if agent2’s actions were not knowingly directed at agent1.

For infants, agent1’s initial actions thus set broad limits on agent2’s reciprocal actions. Simply put, infants expected agent2 not to harm someone who had been nice to her, and not to help someone who had been mean to her. Infants viewed these actions as inconsistent with reciprocity and therefore unacceptable.

Second, we found that in the types of scenarios we used, infants view a range of actions as consistent with reciprocity. In many contemporary studies on reciprocity³, investigators primarily focus on exact reciprocity or tit-for-tat exchanges (e.g., after agent1 acts positively toward agent2, agent2 should respond in kind and act positively toward agent1). But if we go back to older accounts of reciprocity, and in particular to the seminal writings of Robert Trivers⁴, we see a more nuanced approach. In this view, positive reciprocity contributes to the initiation, development, and maintenance of equitable reciprocal partnerships or friendships between unrelated individuals. Thus, if agent1 offers agent2 a gift or directs some other positive action toward her, if agent2 chooses to accept this overture of friendship and responds in kind, and if these positive interactions are repeated over time, a friendship may gradually develop that is advantageous to both agents. With respect to negative reciprocity, a similar logic applies except that the emphasis is on the return of injuries rather than benefits, and feelings of animosity rather than friendship are expected to arise across repeated negative interactions. One key feature of this approach is that it leaves open the possibility that agent2 may choose to remain neutral and not engage further with agent1: not accept her overture of friendship, and not match her act of aggression. Infants in our experiments appeared to acknowledge this possibility.

Following agent1’s positive actions, infants seemed to understand that how agent2 acted was likely to depend on whether or not she was interested in pursuing a relationship with agent1. Thus, infants were not surprised if agent2 chose to respond to agent1 in kind by helping her complete her goal, and they also were not surprised if agent2 chose to remain neutral and not help agent1 achieve her goal. In each case, infants could make sense of agent2’s behavior and hence found it acceptable.

It may seem surprising that infants understand that personal preferences are subjective and can vary from one person to the next, but there is much evidence to this effect. For example, infants understand that the fact that one person prefers playing with stuffed animals over balls provides no information at all about what another person will prefer.⁵ Likewise, infants in our experiments appreciated that although agent1 might be trying to make friends with agent2, agent2 might or might not want to become friends with her, and they viewed actions by agent2 that befit either possibility as acceptable.

Similarly, following agent1’s negative actions, infants viewed a range of actions by agent2 as acceptable, depending on whether or not she was open to actively pursuing an inimical relationship with agent1. Thus, infants were not surprised if agent2 chose to respond in kind and countered agent1’s harmful action with one of her own, and they also were not surprised if agent2 chose to not engage further with agent1 and simply ignored her need for help.

Together, our findings indicate that by 15 months, infants expect a modicum of reciprocity between strangers. Following positive actions by agent1, agent2 is expected to refrain from directing negative actions toward her. Similarly, following negative actions by agent1, agent2 is expected to refrain from directing positive actions toward her. Beyond this, however, infants hold no particular expectations about how far agent2 should go in matching the exact value of agent1’s initial actions.

In future research, we hope to examine whether infants would be more likely to expect in-kind responding or exact reciprocity if agent1’s positive actions involved greater costs to herself and greater benefits to agent2. In our experiments, agent1 offered small gifts that were neither needed nor requested by agent2. What if instead agent2 needed critical help and agent1 provided it at a substantial cost to herself? Evidence that infants expected in-kind responding when agent1 later needed help would suggest that from a young age, infants’ reasoning about reciprocity is sufficiently sophisticated to take into account meaningful contextual differences: A small, unnecessary gift will create a weaker obligation to respond in kind than will costly, critical assistance.

Recent investigations have revealed that early moral cognition includes expectations of fairness⁶, harm avoidance⁷, ingroup support¹, and authority⁸; our research extends this work by showing that an expectation of reciprocity is also present in infancy.

References

1. Jin, K. & Baillargeon, R. Infants possess an abstract expectation of ingroup support. Natl. Acad. Sci. USA 114, 8199−8204 (2017).
2. Ting, F., He, Z. & Baillargeon, R. Toddlers and infants expect individuals to refrain from helping an ingroup victim’s aggressor. Natl. Acad. Sci. USA 116, 6025–6034 (2019).
3. Chernyak, N., Leimgruber, K. L., Dunham, Y. C., Hu, J. & Blake, P. R. Paying back people who harmed us but not people who helped us: Direct negative reciprocity precedes direct positive reciprocity in early development. Sci. 30, 1273-1286 (2019).
4. Trivers, R. L. The evolution of reciprocal altruism. Rev.Biol. 46, 35–57 (1971).
5. Graham, S.A., Stock, H. & Henderson A. M. E. Nineteen-month-olds’ understanding of the conventionality of object labels versus desires. Infancy 9, 341–350 (2006).
6. Sloane, S., Baillargeon, R. & Premack, D. Do infants have a sense of fairness? Sci. 23, 196−204 (2012).
7. Ting, F. & Baillargeon, R. Toddlers draw broad negative inferences from wrongdoers’ moral violations. Natl. Acad. Sci. USA 118, e2109045118 (2021).
8. Stavans, M. & Baillargeon R. Infants expect leaders to right wrongs. Natl. Acad. Sci. USA 116, 16292–16301 (2019).