Testing dogs in ape-like conditions: the effect of a barrier on dogs' performance on the object-choice task

Recent reviews have found marked procedural and methodological differences in the testing of different taxonomic groups on the object-choice task. One such difference is the imposition of a barrier in the testing environment of nonhuman primates in the form of a cage, necessitated to ensure the experimenter's safety. Here, we conducted two studies with domestic dogs (Canis lupus familiaris) in which we compared the performance of dogs tested from within a child's playpen and dogs tested without this barrier present. In Study 1, in a within-subjects design, we found no effect of the barrier on dogs' ability to use a pointing cue, but there was an increase in instances in which dogs failed to choose a cup. In Study 2, in a between-subjects design, dogs tested with a barrier failed to perform above chance, and were also more likely to fail to make a choice. When dogs tested without a barrier made an incorrect response, these were more likely to be incorrect choices than no choice errors. We discuss the implications of these differences in behavioural responses in function of the presence of a barrier and the necessity of ensuring matched conditions when comparing across species.