Does your dog understand you're looking for your keys?

Scientists at the University of Portsmouth, UK, wanted to find out exactly that. More precisely to find out if domestic dogs understand human actions as goal-oriented.

This is an interesting question for scientists trying to assess the intelligence of animal species. Human infants start developing awareness of other people's goal-orientedness after about 5 months of age. Previously scientists have seen some signs of this awareness in primates. The ability to understand goal-oriented behavior in others is a good metric for judging the intelligence levels of animals.

The experiment consists of several trials implementing a habituation-dishabituation paradigm. Now, to explain what that is. The subject is provided stimulus by an object or activity repeatedly until it provides no further interest to the subject. Then the stimulus is changed to something else and the subject starts displaying interest again.

In one experiment done on human infants, an infant would be taken to a room near a table. An object would be placed on the table. The infant would be repeatedly shown this table until they got used to it (habituated). Next either the object would be moved to the opposite side of the table, or the infant moved to the opposite side. Either way the distance to the object would remain same. In the case of the infant being moved, the infant showed the same level of habituation as previously, but in the case of the object being moved, the infant showed new interest in the object (dishabituation).

In a similar exercise, infants were allowed to observe a person interact with an object repeatedly. Then the person would suddenly switch to a new object in the same location. The infants would then show surprise after witnessing this change of the person's actions but if the person starts interacting with the same old object but in a new location, the infant shows no new interest. This means the infant understands the person's actions being goal-oriented towards a specific object.

Does this perceptive ability also exist in animals? Similar experiments to the previous examples have also been done on primates and the ability to distinguish goal-directed actions has been documented. But what about dogs?

In this current study, the same kind of experiment was done on dogs. Because dogs have co-evolved with humans for about 30 000 years, they might be more mindful of human behavior. Researchers have found that dogs can distinguish friendly and threatening behaviors in humans, plus they follow pointing gestures and understand the targeting nature of pointing. Overall 52 domestic dogs were used in the study.

The experiment

The experiment. a) First the experimenter interacts with one of the objects.
b) The objects switch side and the experimenter interacts with the
same object on a different side.
c) The experimenter interacts with a different object on the first side.
Two objects were used. Neither object had been shown any initial interest by the dogs. Dogs were allowed to explore the testing room, then the owner sat in a chair behind the dog. 

In the habituation phase a person enters the room, crouches down and interacts with one of the two objects, by touching it and looking at it intently. The trial lasts until the dog looks away or for a minimum/maximum of 2 and 12 seconds. After that the experimenter left the room. To see if the dogs reacted differently if instead of a person there was an object doing the experiment, a black box was used, which was manipulated by a long stick. Again the box was removed if the dog looked away or 12 seconds had passed. The looking time of the dogs was recorded. 

At the next phase, the dog and the owner left the room for 1 minute after which the objects in the room were swapped. The dogs were given another familiarization trial, to make sure the objects didn't influence the results by the fact that they were in new positions. The dog then saw three consecutive trials where the agent approached the same side but interacted with the new object and three consecutive trials where the agent approached a different side but interacted with the same object. The order of these trials was switched to balance it across dogs in all groups.


The dogs looked at the person much longer than the object when interacting
with a new object.
The dogs looked significantly longer at the new-object side when the actor was a live person, than compared to when the actor was an inanimate object. This suggests that dogs may view the actions of humans as goal-directed, but not the actions of similarly behaving inanimate objects. This is the first time that this kind of phenomenon is documented in a non-primate species.


Share on Google Plus