We are learning that in many ways dogs are much more like people then we thought they were. Consider the following very human situation. A woman is told "If you are holding any mutual funds you should liquidate them soon because a crash is coming in the mutual fund market." How likely is she to follow this advice if the person speaking to her is her hairdresser? Would she be more likely to take the suggested action if the person was a professional financial advisor?


Research is now accumulating which shows that dogs, like people, tend to evaluate just how much knowledge they think that a person has before accepting their guidance and instructions. Dogs are not simply four-footed robots that can be programmed to respond to instructions regardless of the state of affairs. If they think that a person is knowledgeable, at least when it comes to information about things which are important to the dog, they are more likely to accept commands from that individual. This was elegantly demonstrated in a series of experiments conducted by Michelle Maginnity and Randolph Grace of the Department of Psychology at the University of Canterbury in Christchurch, New Zealand.


They started off by capitalizing on the fact that dogs respond to human gestures such as pointing. They trained a set of 16 dogs so that they knew that if they went to a container that the researcher pointed to, the researcher would open that container and give them the treat inside of it. After a while the dog would reliably go to any one of four containers that was pointed to.
Next they erected a low screen which would hide the food containers but not the upper body of anyone behind it. The dog and its owner sat facing the screen and watched while either of two researchers would obviously bend down and fuss around with each of the containers (although the dog could not see them actually manipulating the containers since their view was blocked by the screen). The researchers would then drop the screen and point to the container with the food.


Now here is where things began to get interesting. In the actual experimental test there are two researchers in the room. One of them is sent out of the room so that they can't see where the food is being placed, while the other bends down behind the screen and puts the food treat in one of the containers. The first experimenter is then called back, the screen is dropped and each of the two experimenters points to a different one of four food containers. Remember that the dog has seen one of these two women leave the room during the time when the food was hidden, and that, of course, means that that woman actually doesn't have any knowledge as to where the treat is. So when the woman who was absent points to a container it is likely that she is guessing, 

 

while the other experimenter should obviously have the information as to where the food is. If the dog is sensible it should respond to the instruction from the woman who knows. And, it turns out the dogs are sensible. Even though the woman playing the role of the "Knower" and the "Guesser" are randomly changed around from trial to trial, the dog chooses to go to the container indicated by the person who knows in most instances

 

These investigators followed up by making the situation much more subtle. Now they set up a situation so that there were three experimenters behind the screen, two women and a man in the middle. While the dog watched he saw the man bend over and fiddle around with the food containers behind the screen, ultimately placing the food treat in one of them. While he did this one of the women sat with her hands over her eyes, so that she obviously could not see which container had the bait. The second woman, although she also had her hands on her face, they did not cover her eyes, which meant that she could look down and see where the man had placed the food. So the setup looked much like this.
Once again the screen was lowered and each of the women pointed to a different container. Remember one of these people could not see the placement of the food and the other could. So whose pointing instruction does the dog respond to? Once again dogs act in a reasonable manner and evaluate the advice of the woman who has the information as being more valuable and so they end up choosing the correct container most of the time.


Finally, to see whether dogs will pick up really subtle cues as to who has the information that they need the researchers ran a third experiment. This was set up like the one before with the two women and the man who placed the food between them. Only now both of the women kept their hands in their laps while one container was being baited. There was a difference in the behavior of each of them, since the "Knower" attentively watched while the food was being placed, while the "Guesser" looked up at the ceiling, away from the food containers and the dog. As before, when the screen was dropped both the woman who actually knows and the one who is guessing each point to different containers. Once again the dogs evaluated the state of knowledge of both of these women and chose to follow the advice of the one who knows significantly more often.


Psychologists believe that these experimental results are really important. They show that dogs have what is called a "Theory of Mind". You shouldn't confuse this with the idea that dogs have an idea as to how the brain works, but rather Theory of Mind refers to the ability that an individual has to interpret what another individual might be seeing, feeling, and knowing. It involves understanding that others might have a different perspective, a different amount of knowledge, and even different motives and emotional states. Think of it as a sort of "mind reading ability" since it allows us to interpret what is going on in another person's mind.


In human beings this Theory of Mind develops slowly. Although we have some signs that two-year-olds are beginning to develop this kind of perspective on behavior, it will not be until the child is nearly 4 years of age before he can reliably perform the same task that we see dogs doing. Most other animals are not very good at this. Although chimpanzees and Capuchin monkeys can eventually learn to trust an individual who knows the answer more than one who does not, it requires many learning trials and is not very stable.


It appears that there is something special about dogs. It seems that they have evolved, or perhaps we could better say co-evolved, to live cooperatively with human beings. If a species depends so much upon interactions with another species a little bit of "mind reading" might help a lot. So it was really adaptive for dogs to evolve in a way in which they could "take our perspective" and also learn which information we might reliably have and which we might not.

Based on this set of data the bottom line seems to be that if you don't know, and your dog knows that you don't know, you probably shouldn't be giving him instructions or advice since he is likely not to respond to you in that situation.