Prosociality has received increasing interest by non-human animal researchers since the initial discoveries that suggested it is not a uniquely human trait. However, thus far studies, even within the same species, have not garnered conclusive results. A prominent suggestion for this disparity is the effect methodology can have on prosocial responses in animals. We recently found evidence of prosociality in domestic dogs towards familiar conspecifics using a bar-pulling paradigm, in which a subject could pull a rope to deliver food to its partner. Therefore, the current study aimed to assess whether dogs would show a similar response in a different paradigm, based on the token exchange task paradigm frequently used with primates. In this task, dogs had the option to touch a token with their nose that delivered a reward to an adjacent receiver enclosure, which contained a familiar conspecific, a stranger or no dog at all. Crucially, we also included a social facilitation control condition, whereby the partner (stranger/familiar) was present but unable to access the food. We found that the familiarity effect remained consistent across tasks, with dogs of both the bar-pulling and token choice experiments providing more food to familiar partners than in a non-social control and providing less food to stranger partners than this same control. However, in contrast to our previous bar-pulling experiment, we could not exclude social facilitation as an underlying motive in the current task. We propose this difference in results between tasks may be related to increased task complexity in the token choice paradigm, making it harder for dogs to discriminate between the test and social facilitation conditions. Overall our findings suggest that subtle methodological changes can have an impact on prosocial behaviours in dogs and highlights the importance of controlling for social facilitation effects in such experiments.