Abstract
Reciprocity is one of the most prominent explanations for the evolution of stable cooperation. Although reciprocity has been studied for decades in numerous animal species and behavioural contexts, its underlying proximate mechanisms remain unclear. Domestic dogs provide a useful model species for the study of proximate mechanisms, though there are currently inconsistent findings regarding dogs' propensity to reciprocate. Here, we investigated whether, after minimal training, pet dogs would press a button, which remotely controlled a food dispenser, to deliver food to an enclosure occupied by a helpful conspecific that had provided them with food or an unhelpful conspecific that had not provided them with food. We included an asocial control condition in which the enclosure was unoccupied and a social facilitation control in which the food delivery mechanism was non-functional. Whether subjects were familiar with the helpful and unhelpful conspecifics was also varied. In addition, to investigate potential mechanisms underlying reciprocity, we measured subjects salivary oxytocin concentration before and after they experienced the helpful and unhelpful acts. There was no effect of the previous helpfulness or the familiarity of the partner on the number of times subjects pressed the button. However, there was also no effect of the presence of a partner or the operationality of the food delivery mechanism on the number of button presses, indicating that subjects were not pressing the button to provision the partner. Moreover, the experience of the helpful or unhelpful act did not influence subjects' salivary oxytocin concentration. Variation in findings of reciprocity across studies appears to correspond with differing training protocols. Subjects' understanding of the task in the current study may have been constrained by the limited training received. Additional tests to verify subjects' understanding of such tasks are warranted in future studies.