'Unwilling' vs. 'Unable': do domesticated dogs choose partners based on their intentional actions?

The capacity to understand the intentions of other agents is a useful capacity for all social organisms. Several studies to date have explored whether domesticated dogs (Canis familiaris) share this capacity. Although previous studies show that dogs behave differently toward humans who are unable versus unwilling to share food, it remains unclear whether dogs can use this information to guide broader social decisions such as partner selection. We therefore asked whether dogs would rely on an agent’s intentions—being unable versus unwilling to give food—when choosing whom to approach. In four within-subjects experiments (total N = 86), we gave dogs a choice between two experimenters: an unable experimenter who tried but failed to deliver a reward and an unwilling experimenter who intentionally withheld it. We found that dogs preferred to approach the unable over the unwilling experimenter but only in cases where they were familiarized to the two agents using an asymmetric but predictable reward schedule. Dogs showed no preference for an unable over an unwilling experimenter when given no rewards (Experiment 1), when receiving rewards from both experimenters’ sides (Experiment 2), or when rewards were delivered randomly between the experimenters’ sides (Experiment 4). Dogs showed a clear preference only when a single experimenter’s side was consistently paired with a reward in Experiment 3, regardless of that experimenter’s role. These results suggest that dogs can exploit intention information when choosing social partners, but only under specific motivational contexts.