Video of a dog doing the Imitation Experiment:
http://youtu.be/PHb9ZFkuv1s
An Answer from Dr. Adam Miklosi:
Are dogs able to imitate the action of others?
In 2006 we published the first scientific study on imitation in dogs (Topál et al 2006). This was followed by some debate whether one can really speak about ‘imitation’ in this situation. We defined imitation as the ability to perform a functionally analogue behaviour after seeing in done by others (humans or dogs).
We believe that dogs, as a social species, are able to imitate an action shown by others but it is very difficult to provide a clear evidence for this under controlled experimental conditions. Our procedure was the first to provide strong evidence, despite the fact that dog trainers or people working with dogs (e.g. shepherds) probably relied on this ability already many 100 years ago. Given the assumption that dogs have the ability to imitate, our procedure only makes this phenomenon more explicit. In addition to the possibility of studying this skill by scientific means, it has turned out that this so called ‘Do as I do’ procedure provides also a nice way to train the family dog. As a consequence Claudia Fugazza (a PhD student working with us) has developed a new way of dog training based on the imitative abilities of dogs.
The basic idea is that the dog should be able to copy the action of a human (or dog) demonstrator. This can be achieved through a few simple steps in training. First, the dog should be able to perform a few actions (sit, turn around, lie down etc.) on a verbal/visual signal. Second, by small steps the trainer (owner) should replace these signals by the appropriate behavioural cues. For example instead of giving the ‘Turn around’ verbal signal, the human should actually turn around before the dog is allowed to execute the same action. Third, the dog should learn that (1) it should wait until the action is finished. Usually, it is a good idea to use a specific signal (Listen!) before starting the action (demonstration). (2) The dog should learn that it is allowed to perform an action only if it hears a new signal (Do it!). The ‘Do it!’ signal becomes a general command for the action to be carried out during this procedure.
If the dog performs well with the already known 2-3 actions (he is ‘imitating’ the actions), then we can introduce step by step novel actions. At the beginning the dog learns probably only the link between the specific action demonstration and his action but slowly it may also grasp the idea that his actions should be similar to the demonstrated ones. We could try this out by showing the dog a new action (never used in the training so far), in order to see what he does as a response.
After becoming familiar with this procedure, dogs can be trained relatively easily, independent from their previous training experience. Claudia Fugazza has also shown that using this ‘Do as I do’ method dogs can be trained to perform new actions that can be then associated with a verbal cue in subsequent training. This method seems to be as good as other methods, like clicker training.
We have also shown that dogs are able to remember the demonstrated actions for more than 10 minutes at least, even if they are distracted by performing other activities (Fugazza and Miklósi 2013).
We believe that that there are some advantages in using this method with dogs. First, learning by observation reflects probably a biological tendency in dogs, thus this method relies on a natural interaction between dog and owner. Second, learning by observation may rely on using differ faculties of the mind, so this may enrich the mental life of dogs. Third, the owner has to play an active role in this case (I DO something – now YOU DO the same) that may enhance dog-owner bond and improve cooperation.
References
Fugazza, C., Miklósi, Á. 2013. Deferred imitation and declarative memory in domestic dogs. Animal Cognition, in press.
Topál, J., Byrne, R.W., Miklósi, Á., Csányi, V. 2006. Reproducing human actions and action sequences: “Do as I Do!” in a dog. Animal Cognition, 9: 355-367.
Anonymous
An interesting article. I have a principal doubt:
This type of training based on observation, is not a bit limited in its functionality?
There are many behaviors that we are not able to perform to serve as a model for the dog.
Hope you can answer me
Thank you.