Chippy, our Toller, is a terrible food thief. (Of course, the use of the word terrible is one of perspective. Given his impressive success rate, Chippy would argue that he is actually a very good food thief).

Chip has become so proficient at his food thievery that our dog friends all know to “keep eyes on Chippy” whenever we celebrate a birthday or have snacks after an evening of  training. We are often reminded of the now infamous “Birthday Cake Incident” during which Chip and Grace, an equally talented Aussie friend, succeeded in reducing a section of cake to mere crumbs, no evidence to be found. Suffice it to say, we watch food in our house.

Like many other expert food thieves, Chip is quite careful in his pilfering decisions. He will only steal when we are not in the room or when we are being inattentive. The parsimonious (simplest) explanation of this is a behavioristic one; Chip learned early in life that taking forbidden tidbits was successful when a human was not in the room and was unsuccessful if someone was present and attentive to him. In other words, like many dogs who excel at food thievery, Chip learned what “works”.

However, while a behavioristic explanation covers most aspects of selective stealing behavior in dogs, a set of research studies conducted by cognitive scientists suggest that there may be a bit more going on here.

Do Dogs Have a “Theory of Mind”? Dogs have demonstrated that they will alter their behavior in response to whether a person is actively gazing at them or is distracted. For example, in separate studies, dogs were more apt to steal a piece of food from an inattentive person and would preferentially beg from an attentive person (1,2). However, these differences can still be explained without a need for higher cognitive processing. A dog could learn over time that human gaze and attentiveness reliably predict certain outcomes, such as positive interactions and opportunities to beg for food. Similarly, inattentiveness might reliably predict opportunities to steal a tidbit (or two or five).

It is also possible that, just like humans, dogs use a person’s gaze to determine what that individual does or does not know. This type of learning is considered to be a higher level of cognitive process because it requires “perspective-taking”, meaning that the dog is able to view a situation through the perspective of the person and can then make decisions according to what that individual is aware of. The import of this type of thinking is that it reveals at least a rudimentary “theory of mind” – the ability to consider what another individual knows or may be thinking.

So, while it is established that dogs are sensitive to the cues that human eye contact and gaze provide, it has not been clear whether they can use this information to determine what the person may or may not know. Enter, the cognitive scientists.

The Toy Study: One approach to teasing out “theory of mind” evidence is to control what a dog observes about what a person may or may not be able to see. In 2009, Juliane Kaminski and her colleagues at the Max Planck institute for Evolutionary Anthropology set up a clever experiment in which they used two types of barriers; one transparent and one opaque (3). Dogs and the experimenter sat on opposite sides, and two identical toys were placed in front of each barrier, on the same side as the dog. The dog was then asked to “Fetch!”. They found that the dogs preferred to retrieve the toy that both the dog and the person could see, compared with a toy that only the dog could see.

These results suggest that the dogs were aware that their owners could not know that there was a toy located out of their view, and so retrieved the toy that they (presumably) assumed that their owner was requesting. An additional finding of this study was that the dogs were capable of this distinction only in the present, at the time that the owner’s view was blocked. When the researchers tested dogs’ ability to remember what the owner had been able to see in the past, such as a toy being placed in a certain location, the dogs failed at that task.

The Food Thievery Study: Recently, the same researchers provided additional evidence that dogs are able to consider what a human can or cannot see (4).  A group of 28 dogs was tested regarding their tendency to obey a command to not touch a piece of food while the commanding human’s ability to see the food was varied. The testing took place in a darkened room that included two lamps, one of which was used to illuminate the experimenter and the second to illuminate a spot on the floor where food was placed. During the test conditions, the experimenter showed a piece of food to the dog and asked the dog to “leave it” while placing the food on the ground. The experimenter alternated her gaze between the dog and the food as she gradually moved away and sat down. In two subsequent experiments using the same design, the experimenter left the room after placing the food and the degree of illumination were varied. For each experiment, four different conditions were tested: (1) Completely dark (both lamps turned off); (2) Food illuminated, experimenter dark; (3) Experimenter illuminated, food dark; (4) Both food and experimenter illuminated. In all of the conditions, the dog’s response with the food was recorded.

Results: There were several rather illuminating results in this study (sorry, bad pun):

• Dogs steal in the dark (when a person is present): When the experimenter stayed in the room, dogs were significantly more likely to steal the food when the entire room was in the dark. (They do have excellent noses, after all). If any part of the room was illuminated while the experimenter was present, the dogs were less likely to steal. Conversely, when the experimenter was not present, illumination made no difference at all and most of the dogs took the food. (Lights on or off; they did not care. It was time to party).
• What the Smart Dog Thieves Do: Within the set of dogs who always took the food, when the experimenter was present they grabbed the tidbit significantly faster when it was in the dark, compared to when the food was illuminated. This result suggests that the dogs were aware that the experimenter could not see the food and so changed up their game a bit. (I’ll just weasel on over to the food and snort it up…….heh heh…..she can’t see it and will never know…..I am such a clever dog….). Chippy would love these dogs.
• It’s not seeing the human…..it’s what the human sees: Collectively, the three experiments in the study showed that illumination around the human did not influence the dogs’ behavior, while illumination around the food did (when a person was present). This suggests that it is not just a person’s presence or attentiveness that becomes a cue whether or not to steal, but that dogs may also consider what they think we can or cannot see when making a decision about what to do.

Take Away for Dog Folks

Without a doubt, gaze and eye contact are highly important to dogs. They use eye contact in various forms to communicate with us and with other animals. We know that many dogs naturally follow our gaze to distant objects (i.e. as a form of pointing) and that dogs will seek our eye contact when looking for a bit of help (see Only Have Eyes for You and With a Little Help from My Friends). And now we know that dogs, like humans and several other social species, can be aware of what a person may or may not be able to see and, on some level, are capable of taking that person’s perspective into consideration.

As a trainer and dog lover, I say, pretty cool stuff indeed. Chip of course, knew all of this already.

Oh, and just one more thing……..

 

A Caution: I was really excited about this research because these results continue to push the peanut forward regarding what we understand about our dogs’ behavior, cognition, and social lives. Learning that dogs may be capable of taking the perspective of others, at least in the present, adds to the ever-growing pile of evidence showing us that our dogs’ social lives are complex, rich, and vital to their welfare and life quality.

That said, because these studies had to do with dogs “behaving badly”, (i.e. stealing food, Gasp! Oh No!),  I was a bit hesitant to write this essay. These studies provide evidence that dogs have a lot more going on upstairs than some folks may wish to give them credit for. And as can happen with these things, evidence for one thing (understanding that a person cannot see a bit of food and so deciding to gulp it on down), may be inappropriately interpreted as evidence for another (Oh! This must mean that dogs understand being “wrong”).

Well no. It does not mean that at all.

For those who reside in the (ever diminishing) camps of  “He knows he was wrong“;  “I trained him not to do that; He is just being willful” and “He must be guilty – He is showing a guilty look“: These studies show us that dogs understand what another individual may and may not know based upon what that person can see. This is not the same, or even close to being the same, as showing that dogs understand the moral import or the “wrongness” (whatever that means) of what they choose to do. Chippy knowing that I cannot see that piece of toast that he just pilfered is NOT the same as Chippy feeling badly that he took it. Remember, we put the last nail in the guilty dog coffin quite some time ago. (See “Death Throes of the Guilty Look“).

Bottom Line: These studies show us that dogs may be sneaky, but they don’t say anything at all about whether they’re feelin’ guilty.

Cited Studies:

1. Call J, Brauer J, Kaminski J, Tomasello M. Domestic dogs (Canis familiaris) are sensitive to the attentiaonal state of humans. Journal of Comparative Psychology 2003; 117:257-263.
2. Gacsi M, Miklosi A, Varga O, Topal J, Csanyi V. Are readers of our face readers of our minds? Dogs (Canis familiaris) show situation-dependent recognition of human’s attention. Animal Cognition 2004; 7:144-153.
3. Kaminski J, Brauer J, Call J, Tomasello M. Domestic dogs are sensitive to a human’s perspective. Behaviour 2009; 146:979-998.
4. Kaminski J, Pitsch A, Tomasello M. Dogs steal in the dark. Animal Cognition 2013; 16:385-394.