Inference in wolves and dogs: the ‘cups task’, revisited

Abstract

Inferential reasoning, which refers to the process of arriving at a conclusion from a series of premises, has been studied in a multitude of animal species by using the ‘cups task’ paradigm. In one of the versions of this set-up, two opaque cups (one baited and one empty) are shaken in front of the animal. As only the baited cup makes a noise when shaken, the animals can locate the reward by inferring that only a baited cup would make noise that an empty cup would make no noise or both. In a previous iteration of this paradigm in wolves, Canis lupus, and dogs, Canis familiaris, wolves seemed to outperform dogs. However, assessing neither each species' inference capabilities, nor their relationship with each other was not feasible because of the lack of control conditions. In this study, several conditions in which the baited cup, the empty cup or no cups are shaken are added to tackle this issue. Results indicate that wolves and dogs made their choices based on the saliency and order of the stimuli presented and not based on inference, which is consistent with the previous study. In addition, the potential causes behind the animals' performance are discussed, and alternative paradigms that may be more apt to measure inference abilities in wolves and dogs are proposed in this study.