Abstract: <p>Eavesdropping involves the acquisition of information from third-party interactions, and can serve to indirectly attribute reputation to individuals. There is evidence on eavesdropping in dogs, indicating that they can develop a preference for people based on their cooperativeness towards others. In this study, we tested dogs’ eavesdropping abilities one step further. In a first experiment, dogs could choose between cooperative demonstrators (the donors) who always gave food to an approaching third person (the beggar); here, the only difference between donors was whether they received positive or negative reactions from the beggar (through verbal and gestural means). Results showed that dogs preferentially approached the donor who had received positive reactions from the beggar. By contrast, two different conditions showed that neither the beggar’s body gestures nor the verbal component of the interaction on their own were sufficient to affect the dogs’ preferences. We also ran two further experiments to test for the possibility of dogs’ choices being driven by local enhancement. When the donors switched places before the choice, dogs chose at random. Similarly, in a nonsocial condition in which donors were replaced by platforms, subjects chose at chance levels. We conclude that dogs’ nonrandom choices in the present protocol relied on the simultaneous presence of multiple cues, such as the place where donors stood and several features of the beggar’s behavior (gestural and verbal reactions, and eating behavior). Nonetheless, we did not find conclusive evidence that dogs discriminated the donors by their physical features, which is a prerequisite of reputation attribution.</p>