What experimental experience affects dogs' comprehension of human communicative actions?

▶ Using a single-trial approach, we show that dogs correctly follow some communicative actions by a human experimenter to a target goal (hidden food), but not all. In particular, dogs located the hidden food following a single presentation of a distal point, a step toward the container, and a foot point with the experimenter's hands occupied. Under the same testing procedures, dogs failed in response to a foot point with the experimenter's hands free, and when the experimenter first pointed and then stepped, or the opposite. ▶ The fact that dogs succeeded on foot point with hands occupied, but not with hands free, suggests that they take into account the rationality of the actor, and in particular, consider the role of environmental constraints (experimenter's hands occupied) when evaluating the goal-directed nature of an action. ▶ Unlike many studies of dogs in which only the subject's response is videotape and coded, we coded the subject, the experimenter and the handler. On the basis of these coding procedures, we uncovered a significant number of trials that had to be discarded based on unintentional cueing or errors by both the experimenter and the handler. We urge caution in testing dogs in the absence of such coding procedures.