A new MIT study finds people follow precedent rather than turn taking in boss employee, mentor mentee, and other hierarchical ties, complicating the standard behavioral economics picture of universal reciprocity.
If your boss hands you a coffee, you probably don't open a mental tab. A friend doing the same thing is a different story. A new study from MIT, published this week in the journal Open Mind, helps explain why that asymmetry is built into how people read relationships, not a quirk of etiquette.
The paper, by graduate student Alicia Chen and senior author Rebecca Saxe, runs a series of short-story experiments in which participants read vignettes about everyday generosity: a coworker buying someone a coffee, an aunt preparing a meal for a niece, a manager covering a report for a direct report. The researchers vary whether the relationship is symmetric, like two friends or cousins, or asymmetric, like an aunt and a niece, a manager and an employee, or older and younger siblings. Then they ask what the reader expects to happen next, whether the gesture is returned, escalated, or simply continued in the same direction.
The headline result, reported in the MIT News release and the Open Mind article, is that the answer depends entirely on the shape of the relationship. In equal-status pairings, people expect generous acts to be paid back turn by turn, the way the standard economic model of reciprocity predicts. In hierarchical or role-based pairings, that turn-tracking disappears. The same direction of generosity is expected to continue, and the question of who owes whom effectively drops out. "When people know the relationship is asymmetric, they don't expect reciprocity; they expect the same action to keep on going," Chen told MIT News.
That finding puts pressure on a foundational assumption in behavioral economics and game theory: that reciprocity is a universal default, an instinct that fires whenever someone does something nice and then waits to be paid back. Saxe's lab is arguing, on the basis of these story-based predictions, that universality is the wrong level of description. "We just follow precedent because following a precedent is easier. We all know what to expect, and we don't have to keep track of what happened last time," Saxe said. Keeping score, in other words, is the expensive option, and it is reserved for the social situations where equality is the thing being maintained. In a hierarchy, the hierarchy is what is being maintained, and the lower-cost strategy is to keep doing what was already done.
The everyday architecture this maps onto is wide. A mentor who writes a recommendation letter for a former student does not appear to expect the student to write one back. A senior employee who covers a shift for a junior one is not running a tab. A parent who pays for a wedding is not, in any normal accounting, awaiting reciprocal nuptial funding. The Open Mind paper's claim is that people navigating these situations are not failing to apply a universal rule. They are applying a different rule, one calibrated to the asymmetry. The vignette paradigm lets the researchers test that prediction without asking people to introspect, by tapping expectations about a third party in a story.
The honest limits matter. The experiments measure what people expect to happen in made-up situations, not what they or anyone else would actually do in a real office or kitchen. The press release frames the work as the first experimental demonstration of the specific, relationship-conditional shape of reciprocal expectations, but the underlying Open Mind paper is where the sample sizes, statistical reporting, and cross-condition comparisons actually live, and those details have not been independently verified here. The participants are US-based and English-speaking, and the authors themselves treat generalization across cultures as future work rather than as a finding. Anyone tempted to read this as a claim about human nature writ large should treat it, for now, as a claim about how Americans parse the relationships in the kinds of stories they were shown.
What the lab says it wants to do next, per the release, is build computational models that put social-relationship factors alongside the payoff-and-benefit factors that have dominated formal models of generosity, so the two can be quantitatively compared. The work is funded by the Simons Foundation Autism Research Initiative and the Patrick J. McGovern Foundation, both of which have a longer interest in how people read other minds and infer rules. Whether the relationship-conditional pattern holds up under those formal models, and outside the English-language vignette paradigm, is the next test. The structural claim, though, is already clear enough to redraw a familiar question: the next time you wonder whether you owe someone a favor back, the answer may be hiding in plain sight in the shape of the relationship, not in the ledger of who went first.