A new study published in Animal Cognition challenges the idea that dogs can judge a person’s character based on how that person behaves toward others. Researchers tested whether pet dogs could form “reputations” of humans after either watching them interact with another dog or interacting with them directly. Across all age groups, the dogs did not show a consistent preference for a generous food-giver over a selfish one, suggesting that reputation formation in dogs may be more limited—or harder to detect—than previously thought.
The research team set out to examine a long-standing question in animal cognition: can domesticated dogs evaluate people socially in ways that influence whom they choose to approach? While previous studies have reported mixed results, few have directly compared different age groups or tested both direct interaction and observational “eavesdropping” in the same experiment.
“I’ve always been fascinated by how animals think and behave. As a comparative psychologist, I study social cognition in group-living mammals to better understand the mental processes behind their interactions—and what that might reveal about our own behaviour. This study was part of my PhD project on social eavesdropping in dogs, wolves, and elephants,” said study author Hoi-Lam Jim, a postdoctoral researcher in the Yamamoto Lab at the Institute for the Future of Human Society at Kyoto University.
“We conducted this study as a follow-up to our earlier work (Jim et al., 2022), in which we tested pack-living dogs and wolves at the Wolf Science Center in Austria. These animals’ life experiences differ significantly from those of pet dogs. That study also found no evidence of reputation formation, but it remained unclear whether the negative results reflected cognitive limitations or differences in life experience.”
“With the current study, we aimed to disentangle the role of ontogeny in dogs’ social evaluation abilities by testing pet dogs of different ages,” Jim explained. “We predicted that older dogs would perform better due to greater experience with humans.”
Forty pet dogs of varying breeds and ages were recruited from a volunteer database. The researchers divided them into three age groups—young (1–3 years), adult (4–7 years), and senior (8–12 years)—to use age as a stand-in for lifetime experience with humans.
The experiment involved three conditions. In the eavesdropping condition, dogs watched two unfamiliar women interact with a “dog demonstrator.” One woman acted generously by offering food, while the other withheld it and turned away. In the control condition, the same actions were performed but without a dog present, to ensure that any response wasn’t just due to human gestures or tone. In the direct experience condition, the participating dog interacted with the generous and selfish partners directly.
Testing took place in an outdoor enclosure, with procedures carefully designed to avoid bias from experimenter position, clothing, or inadvertent cues from owners (who wore blindfolds). The main outcomes were which person the dog approached first and how much time the dog spent showing affiliative behaviors—such as staying close or looking at the person. Statistical models compared performance across conditions and age groups, while also accounting for trial order and possible side or color biases.
Contrary to expectations, the results showed no group-level preference for the generous partner in any condition. Only three individual dogs—two seniors and one young—consistently favored one partner, and one of these actually preferred the selfish person. Age did not predict success, and more exposure across trials did not make dogs more likely to choose the generous partner. Roughly a third of the dogs showed a consistent side preference, likely due to a shaded area on one side of the testing space, but removing these cases did not change the outcome.
“Our results did not provide clear evidence that dogs can form reputations of humans, either through direct interactions or by observing third-party interactions (eavesdropping),” Jim told PsyPost. “This suggests that pet dogs may not form quick judgements about people, and that reputation formation may be quite cognitively demanding for animals.”
“While it wasn’t surprising that dogs didn’t show signs of eavesdropping (since that ability is more cognitively complex), it was unexpected that we also didn’t find evidence for direct reputation formation. It seems intuitive that dogs would prefer someone who feeds them over someone who withholds food, so it was surprising that we couldn’t even demonstrate this basic preference in our study.”
The team suggests several possible explanations. One is that even direct reputation learning may require more repeated interactions than were provided here; studies with chimpanzees, for example, have shown that dozens of trials may be needed before a reliable preference emerges.
Another is that the visual cues distinguishing the two partners—different clothing colors—might not have been salient enough. It’s also possible that the dogs formed impressions but did not act on them in the low-stakes testing environment, especially since pet dogs are accustomed to generally positive treatment from strangers and were not food-deprived.
Environmental factors could also have played a role. Conducting the study outdoors introduced distractions and uncontrolled variables such as the sun–shade difference, which may have influenced choices independently of the partners’ behavior. The authors note that while lab settings remove such variables, they also strip away the ecological complexity of real-life decision-making.
“Although we didn’t find evidence that dogs form reputations of humans in our study, that doesn’t mean they lack the capacity,” Jim noted. “It’s possible that the experimental design, such as the two-choice paradigm, didn’t fully capture how dogs evaluate others. We also can’t rule out the possibility that dogs did form reputations internally but didn’t express their preferences in a way we could detect.”
The researchers suggest that future studies compare dogs from different environments, such as free-ranging populations, shelter residents, and working dogs, to see whether varying life experiences influence the ability to evaluate humans. They also recommend testing scenarios where negative behaviors—rather than generosity—are the focus, as some studies have found animals are more likely to avoid unhelpful or unfriendly individuals than to seek out helpful ones.
The study, “Do dogs form reputations of humans? No effect of age after indirect and direct experience in a food-giving situation,” was authored by Hoi-Lam Jim, Kadisha Belfiore, Eva B. Martinelli, Mayte Martínez, Friederike Range, and Sarah Marshall-Pescini.
