Dogs may not build “reputations” of generous or selfish people, even with age, study finds
A new study in Animal Cognition challenges a popular idea about canine social smarts: that dogs form lasting “reputations” of people based on how they treat others. Testing 40 pet dogs across three age groups, researchers found no evidence that dogs preferred a generous human over a selfish one after either watching that person interact with another dog or directly experiencing generosity or selfishness themselves. Age made no difference: young, adult, and senior dogs all performed at chance level, suggesting reputation-building—as tested here—may be rarer or harder to detect than many assume.
The question matters because dogs live closely with humans and often rely on them for food, safety, and guidance. Previous studies have produced mixed results, with some finding that dogs favor “nice” or helpful people and others finding no such bias. The new work set out to tighten the design by using human–dog demonstrations (rather than human–human), making the social information more relevant to canine observers, and by comparing indirect “eavesdropping” with direct experience in the same experiment.
In the eavesdropping condition, each dog watched two unfamiliar people interact with a dog demonstrator: one person was generous (feeding and speaking kindly), the other withheld food and turned away. A control group saw the same human actions without a dog present, to rule out simple cues or theatrics. In a separate session, the same subjects experienced the roles firsthand: one partner fed them, the other did not, across multiple brief trials. Dogs’ choices—who they approached first—and their affiliative behaviors were recorded. The sample included 40 pet dogs (balanced by sex), classified as young (1–3 years), adult (4–7), or senior (8–12) to test whether life experience shapes reputation formation.
Across conditions, the pattern was consistent: dogs did not choose the generous partner more often than chance, whether they had merely watched the interactions or had personally been fed or refused. Time spent near the partners told the same story. Crucially, age did not shift the results; seniors were no better than youngsters at favoring the generous person, contradicting the idea that years of human experience automatically sharpen this skill. The authors conclude that their data do not support reputation formation under these circumstances and underscore the methodological challenges of studying such a subtle social ability.
The findings sit within a broader, sometimes contradictory literature. Early “positive” results in eavesdropping studies were later explained by confounds like “local enhancement,” where animals simply go to the place where action happened rather than evaluating the actor. Eavesdropping itself is cognitively demanding: it requires attention to third-party interactions, remembering who did what, and then using that memory in a new context. Even in direct interactions, dogs may prioritize immediate cues—tone of voice, posture, food visibility—over building an abstract, longer-term “reputation” of a person. The authors note that such complexities, along with design differences across studies, help explain why results in this field often diverge.
For owners and trainers, the takeaway is not that dogs “can’t tell” who treats them well, but that translating brief demonstrations or short encounters into a stable preference may require stronger, more consistent signals than those used here. Future work could test longer learning periods, clearer contrasts between partners, or more ecologically familiar contexts. Until then, the study adds a careful note of skepticism to a compelling story about canine social evaluation—and a reminder that what looks obvious in everyday life can be surprisingly hard to prove in the lab.
Source: Do dogs form reputations of humans? No effect of age after indirect and direct experience in a food-giving situation. Animal Cognition (2025). https://link.springer.com/article/10.1007/s10071-025-01967-w
