Dogs have a unique capacity to follow human pointing, and thus it is often assumed that they can comprehend the referential meaning of such signals. However, it is still unclear whether dogs perceive human directional gestures as signals referring to a target object (indicating what to manipulate) or a spatial cue (indicating where to do something). In the present study, we investigated which of these alternative interpretations may explain dogs' responses to human pointing gestures in ostensive communicative and nonostensive cuing contexts. To test whether dogs select the cued object or the cued location, subjects were presented with 2 alternative object-choice trials. An experimenter first attracted the attention of the dog either by calling the dog's name and looking at it (ostensive condition, n ϭ 24) or by clapping the hands (nonostensive condition, n ϭ 24) then pointed at 1 of 2 different toy objects. Subsequently, the experimenter switched the location of the 2 target objects in full view of the dogs by grasping the objects and making a 180°turn. Dogs were then allowed to choose between the 2 objects. In the ostensive condition, dogs showed a significant bias toward the cued location compared with the nonostensive condition in which they performed at chance. These results suggest that pointing refers to a direction or location for dogs, but only if they are addressed with ostensive cues that indicate the communicative intention of the signaler.

Ostensive signals preceding referential cues are crucial in communication-based human knowledge acquisition processes. Since dogs are sensitive to both human ostensive and referential signals, here we investigate whether they also take into account the order of these signals and, in an object-choice task, respond to human pointing more readily when it is preceded by an ostensive cue indicating communicative intent. Adult pet dogs (n = 75) of different breeds were presented with different sequences of a three-step human action. In the relevant sequence (RS) condition, subjects were presented with an ostensive attention getter (verbal addressing and eye contact), followed by referential pointing at one of two identical targets and then a non-ostensive attention getter (clapping of hands). In the irrelevant sequence (IS) condition, the order of attention getters was swapped. We found that dogs chose the target indicated by pointing more frequently in the RS as compared to the IS condition. While dogs selected randomly between the target locations in the IS condition, they performed significantly better than chance in the RS condition. Based on a further control experiment (n = 22), it seems that this effect is not driven by the aversive or irrelevant nature of the non-ostensive cue. This suggests that dogs are sensitive to the order of signal sequences, and the exploitation of human referential pointing depends on the behaviour pattern in which the informing cue is embedded.

Both human infants and nonhuman primates can recognize unfamiliar entities as instrumental agents ascribing to them goals and efficiency of goal-pursuit. This competence relies on movement cues indicating distal sensitivity to the environment and choice of efficient goal-approach. Although dogs' evolved sensitivity to social cues allow them to recognize humans as communicative agents, it remains unclear whether they have also evolved a basic concept of instrumental agency. We used a preferential object-choice procedure to test whether adult pet dogs and human toddlers can identify unfamiliar entities as agents based on different types of movement cues that specify different levels of agency. In the navigational agency condition, dogs preferentially chose an object that modified its pathway to avoid collision with obstacles over another object showing no evidence of distal sensitivity (regularly bumping into obstacles). However, in the goal-efficiency condition where neither object collided with obstacles as it navigated toward a distal target, but only 1 of them exhibited efficient goal-approach as well, toddlers, but not dogs, showed a preference toward the efficient goal-directed agent. These findings indicate that dogs possess a limited concept of environmentally sensitive navigational agency that they attribute to self-propelled entities capable of modifying their movement to avoid colliding with obstacles. Toddlers, in contrast, demonstrated clear sensitivity to cues of efficient variability of goal-approach as the basis for differentiating, attributing, and showing preference for goal-directed instrumental agency. (PsycINFO Database Record

15 16Evidence suggests that infants and adults attribute different importance to certain object 17 properties when performing object-directed actions. Namely, infants tend to rely on 18 information about an object's location, whereas adults are more likely to base their actions on 19 its features. In this study, we tested whether the strategic choices of infants (aged 13 months) 20 and adults would be modified by the context of the demonstration. Participants watched as an 21 experimenter hid a ball under one of two different coloured containers, using either a 22 communicative or a non-communicative manner. Then, the locations of the two containers 23were changed out of sight of the participant. During the test, participants were encouraged to 24 look for the ball under one of the containers. We found that adults were more likely to follow 25 a feature-based strategy than infants. However, there was no effect of the context of the 26 demonstration, suggesting that communication may play different roles in encoding object 27 properties and directing overt behaviour. 28 29

No abstract