Is dogs' tendency to follow human misleading communicative cues influenced by humans' auditory perspective?

Studying Theory of Mind in non-verbal populations requires designing tasks that address the distinction between responding based on directly perceivable cues and additionally inferring others' mental states. We designed and pre-registered an auditory version of a non-verbal change-of-location task, to investigate whether dogs are sensitive to a human communicator's mental states about the location of food. With control conditions we ruled out alternative cognitive processes such as associative learning. Dogs (N = 240) could witness that food was hidden first in one opaque bucket (A) and then relocated to a second opaque bucket (B) by an experimenter. Before being allowed to retrieve the food from one of the buckets, dogs received a misleading suggestion (A) from the communicator, who could not see the scene. In all conditions, the communicator could hear food being hidden in A, due to the presence of bells on the lid of this bucket. We manipulated whether she could also hear that food was removed from A and relocated to B (true belief) or not (silent bells on B, leading to her false belief). Importantly, in both conditions the communicator behaved identically (present in the room, suggesting A). Dogs' responses were not statistically different from those of a previous study using a similar change-of-location task in the visual domain (Lonardo et al. 288(1955), 2021, https://doi.org/10.1098/rspb.2021.0906). Dogs' choices in the present study, however, did not differ significantly across conditions, suggesting that any auditory perspective-taking ability they might possess did not have a sufficiently large effect to be detected in this study.