Many dog owners would agree that there are few things more tragic to a playing dog than the ball rolling under the couch. Whenever it happened to my dog when I was in the room, she would look up at me, then back to the ball in what seemed like a silent plea for assistance. Of course, appearances can be deceiving. Some scientists argue that this human-directed gaze behavior, rather than being a form of intentional communication, is a learned response to certain situations: Based on previous experience, the dog has learned that when a desired object is out of reach, looking at a human leads to that object being moved within reach.
Marshall-Pescini et al. (2013) investigated this issue by giving dogs a task that, though first possible, became impossible in a later trial. They wanted to see if and how the dogs’ human-directed gazing behavior changed when the task became impossible. The task was to obtain food that was placed under an overturned, transparent Tupperware container. The food and container were placed on a board, so the food could be obtained using various strategies (e.g. turning the container over, pushing the container off the board). The first three trials were solvable, as the container was merely placed over the food. However, in the fourth trial, the container was attached to the board, making the task unsolvable. During all the trials, the dog’s owner (the “caregiver”) stood about a foot away from the board, facing it, while an experimenter also stood about a foot away from the board, adjacent to the owner.
Merely comparing the dogs’ behavior in the solvable and unsolvable trials would not rule out the theory that their human-directed gaze behavior is a learned response, so Marshall-Pescini et al. added a second condition. In the first condition (Attentive), both the owner and experimenter faced the board. In the second condition (Back), the owner faced the board, but the experimenter faced away from the board. If dogs’ human-directed gazes are a form of intentional communication, then presumably the dogs would gaze more at the person actually looking at them and the container. (This, of course, requires the dogs to understand that communication requires attention, or, at the very least, eye contact.)
Marshall-Pescini et al. also conducted this study with very young, preverbal children (15-27 months old), so they could directly compare dogs to humans. In this part of the study, the caregiver was the children’s nursery school teacher.
Marshall-Pescini et al. found that, in the unsolvable trials, both the dogs and the toddlers increased alternating their gaze between the caregiver and the container, compared to the solvable trials. Additionally, in the Back condition, they gazed more at the caregiver (who was facing the board) than at the experimenter (who was facing away from the board).
These results suggest that the human-directed gaze behavior of dogs (and toddlers) really is a form of intentional communication, rather than a learned response. We’ve already seen that dogs can understand human communication, but it seems that dogs can intentionally communicate with us, too!
(Although the overall results were the same, there were a few interesting differences in gazing behavior between the dogs and toddlers. For example, in the Attentive condition, dogs looked at both the experimenter and the caregiver for help, but toddlers looked at the experimenter much more than the caregiver. Marshall-Pescini et al. attribute this difference to some important discrepancies in environmental factors when testing the toddlers and dogs. First, the dogs were tested in a completely unfamiliar room, whereas the toddlers were tested in a room at their nursery school. Second, the “caregiver” in the dog experiment was the dog’s owner, while it was a nursery school teacher in the toddler experiment. How might these factors have caused the differences in gazing behaviors of the dogs and toddlers?)
Source:
Marshall-Pescini, S., et al. “Gaze alternation in dogs and toddlers in an unsolvable task: evidence of an audience effect.” Animal cognition (2013): 1-11.
Bird Brains and Monkey Minds 
wouldn’t it have been more significant if they had the owner of the dog turn away and the experimenter face the board? cuz then could familiarity have played a factor in who the dog turned to?
so if the owner was facing away that might test eye contact with the experimenter rather than familiarity with the owner? if that makes any sense…
Yes, definitely! I don’t think the researchers took that into account when they ran the study (nor the possibility that the dogs might be more bonded to their owners than the toddlers are to their nursery school teacher). The researchers actually did a pilot study where both the experimenter and the caregiver turned around, but it was too distressing for the toddlers, so maybe that’s why they chose to always have the familiar caregiver facing the board.
It would be nice if the researchers did the study again with the caregiver turned away, like you suggest, to see how this affects the dogs’ and toddlers’ behavior. They do mention some previous research that suggests that dogs prefer to communicate/obtain information from their owners, rather than a stranger, but it would be interesting to see if this is true even when the owner is clearly not attending to the dog.
Also, I just found a paper where the researchers did the same experiment with dogs that have an anxiety disorder — I haven’t had the chance to read it fully, but it seems really cool: http://www.sciencedirect.com/science/article/pii/S0168159113001354