Interview with Claudia Fugazza. Do as I do and imitation in dogs

Interview with Claudia Fugazza. Do as I do training technique based on natural abilities of dogs to imitate.
Does the Do as I do training technique really teach a dog to imitate?
Do as I do is a training technique that relies on the cognitive ability dogs possess. So the Do as I Do is not about teaching dogs to imitate, they already possess this cognitive ability and they use it in their lives. So, basically, if a dog wants to learn to open a door and go in the garden he can learn it socially. And he possesses this ability.
With the Do as I do you give a cue for the rule that Do it means copy, but this relies on the cognitive ability that dogs already possess.
If dogs already have these inborn cognitive abilities that allow them to learn socially, why some dogs use them without any training while other dogs don't?
I think that all dogs could do it. And may be there are some predispositions or also some differences in raising environments, previous experience of dogs. So, for example, if a dog lives alone in the garden and has very little social contact with the owner… he might possess this cognitive ability, but this cognitive ability may not surface, because the environment and the experience of the dog do not allow it to surface. This might be the one of the hypothesis for these differences between dogs.
What do the dogs usually learn by themselves socially, observing their owners?

I think that in a very general way dogs learn what are the social rules in a family, and [they do it] in a mixed way of social learning and individual, associative learning. [They learn] if the behaviours are punished or rewarded and if other people in the family or other dogs behave in a way. So there's an influence from social learning and there's an influence of individual learning for this very general learning of how to stay in the world, let's put it so.
But then what they usually learn more by imitation or these more social learning processes is that usually has something to do with the goals they also want to achieve. So, for example, if the owner opens a litter bin and in the litter bin usually there's something that dogs like may be they'll observe and at least understand that the litter bin can be opened and they'll learn by themselves how to open it. Or they can even learn how to open the litter bin… socially. So this is a kind of behavior that is most frequently observed.
But then also to dig holes. This is something dogs like. And many times owners reported that they were nicely planting some plants in the garden and then they left. The dog was in the garden. Of course, the dog was observing, because they really observe us more than we think. And then you go back and you find a mess in your garden, because the dog has also [dug] some holes, bacause he behaves as you.
Are there dogs that are unable to learn socially?
I don't think that there are dogs that don't learn socially. I mean in my experience there are no dogs that cannot learn socially. Basically, I think this training method relies on cognitive ability that is part of what a dog possesses. But then to start the training protocol, of cource, a dog needs to learn these three behaviors on the verbal cue. So this is part of the training history of the dog. But a baseline, this cognitive ability… I think it's possessed by all dogs.
Are there any differences between breeds that can influence the outcome of Do as I do training?

I'm not studying breed differences, but I have experience with many many dogs (because of seminars and courses) different breeds. And from my experience it turns out that I didn't find any differences in breeds.
Of course, it's easier to work with working breeds like border collies, because they are really predisposed to cooperate with the owner. And they are really motivated to do something besides any rewarding. But I haven't found any differences in breeds like czechoslovakian wolfsdog or italian wolfdog, which are really not predisposed to work with humans. And I had no difficulties in training this kind of dogs. So this is my experience.
I don't have experience with such breeds like livestock guarding dogs, because they usually don't come to training classes. They stay with the sheeps in alpine pasture. So I cannot really generalize the answer. So for this kind of breeds I don't know. But for breeds like Yorkshire, German Shepherd or Poodle or, say, Chihuahua… in my experience there's no difference.
Nontheless, it's interesting that you started Do as I do training with czechoslovakian wolfsdog which has often been called an «untrainable» dog…
The fact that I started with czechoslovakian wolfsdog, this was my dog, this is my dog, so, of course, I started with her. And she is really willing to collaborate with me, especially if there's some food for her. And she seems to enjoy training very much. So I wouldn't say that this is `untrainable' [dog], but this demonstrates that if you really know the learning theory and if you're able to apply it in practice, you can train a wolfdog.
What can negatively affect dog's ability to learn through Do as I do?
At the first place I would put inhibition. So if the dog… and his experience of training has been inhibited, which should never be done, he will not feel comfortable, of course. He will not feel so good to try to perform something. And he will have severe difficulties in grasping the idea of imitating or doing what you do. Because he is inhibited, he is basically sitting there and hoping nothing happens. So that's te first.
And the second is… those dogs that are always trained to do the same thing. So as an example, I don't know… Schutzhund or this kind of training. Because usually the owners that do this don't practice other sports and don't make their dogs do other experiences. It's not good to generalize. I just take it as an example, so I could take also agility or obedience. But if the dog is really stuck to some association, because also the dog has always been trained with associative learning. So training is an association, and training is always doing exercises A, B, and C and that's it. These dogs might, at least at the beginning, have some hard times to perform something new and to work cognitively.
Have you ever seen dogs who seemed to be unwilling to work with owners in Do as I do tasks?

No, I don't think so… I mean… to work with the owner… in my experience no, but I have a sort of biased sample, because at seminars usually come owners that work with gentle methods, not choke chains, collars or such things. So I have a biased sample.
And there are dogs that don't want to work with other people, because may be they are shy of other people, usually. And I've tried to use this technique to improve the situation, because it might be that at the beginning the dog is trained by the owner, so he learns the rule of imitation, he learns that Do it means copy with the owner. And then you can try to train the dog this way with another person. And my experience with really few dogs is that it works in a sense that the dog can anyway have a relationship or communication with this stranger he usually is fearful of in the situation he knows. But having sort of strong relationship, because the person is the model in that moment, but without physical contact. So the dog is not forced to have physical contact. He just has to perfrom the same action.
It may seem that social learning is to some extent incorporated into every learning process. Is that the correct?
Basically, I put it in a sort of opposite way. Because we all, humans too, can learn different tasks in different ways. And it is possible to learn a very simple task spending fewer energy, for example, by association, which is a simpler process, we use association. And if we need some more cognitive reasoning we use it. So probably depending on how it is easier to learn a task we use either social learning or associative learning, trials and errors. It really depends on the task and the situation.
Basically, social learning strategy tells that we use social learning when the task is difficult or when we don't have, or haven't learned a strategy that is functional, that is efficient, or when it's dangerous.
How do you think what is the best practical application of the Do as I do training technique?
I think that Do as I do's best deployment is for dogs for disabled owners. I have no experience with dogs for disabled owners, because unfortunately, in Italy there is no such a demand for these dogs, and the training is quite expensive for the owner. My colleague from the Universit of Budapest Márta Gácsi trains dogs for disabled owners and she uses much social learning in the training.
Apart from training dogs for handicapped owners, and other very practical applications, what is Do as I do technique good for?
The story maybe goes a little bit back. So the idea of the Do as I do training comes from the scientifict discoveries of the cognitive abilities of dogs. And we discovered that dogs have really fine and complex social cognitive skills almost similar to humans' cognitive skills. And I was really surprised that in training these skills are never used. Because in training the traditional training techniques, gentle or not gentle, are always only based on associative learning, which is a very simple process possessed by basically all species in which we have searched for it: insects, fishes, crabs, mammals (humans, dogs etc.).
But if the dog is really so predisposed, as we have found in our scientific studies, to learn socially from humans, and if we know now that he has this ability of functional imitation, why shouldn't we use this in training? So yes, it's true that the dog can easily learn a complex task this way, but our aim as dog trainers or dog lovers or as dog owners might also be just to exercise these cognitive skills our dogs have in order to provide some good experience, mental experience for them.
In your book you say that Do as I do training improves relationships between an owner and a dog. How did you realize this? How do these improvements feel like?

When I started Do as I do this was a kind of pilot experiment for me, because I've read the study [1] and I wanted to try out the protocol with my dog, because I was really curious. And when I started I thought that I really had a good relationship with my dog, which was true. We've already done many thing together, sport or walks in the forest, we went skiing together, we went swimming together, we played a lot. So I thought it was good even if my dog is a quite difficult one, of a particular breed, because czechoslovakian wolfdogs are not so easy usually.
But then I started Do as I do and I realized that we'd really improved our understanding of each other in such a deep way. So basically it was like I completely knew what she was doing and she completely knew what I was doing. So it's more of understading each other thing. But, of course, this is not a scientific point of view, it's just my personal experience of my feelings.
Sounds like what people say about games and toys for dog's mental activation, doesn't it?
Yes. So Do as I do realies on cognitive abilities that are different from [abilities for] associative learning and therefore it allows your dog to exercise his cognitive abilities more than whatever has to do with associations. So this way, for this respect, yes.
Do results of the Do as I do training depend on relationship between an owner established before the training?
I think that there's a strong influence. But it's in both directions. So if you have a good relationship, in which you already are somwhat a model for your dog, because you are a coherent owner, because you have always behaved in a coherent way providing formations the dog can rely on, [then] it's easier for the dog to start training.
But it's also true the opposite. So if you don't have such a relationship with your dog, through this method you might improve a little bit this situation.
How far we can go with Do as I do training?
Actually, we don't know. Because we don't have such an experience that you really scientifically test how far we can go and what a dog can imitate and what he cannot. So may be we will get this experience in some years. But now I cannot answer.
Could it be that, once a dog grasped the idea of Do as I do, he or she will imitate even if the owner doesn't want?

This is a question that is often asked in seminars. And this is not my experience. So you might suppose that the dog exercise these social cognitive abilities and he might use them more often in everyday life. But basically I think that dogs are already do it before and without the Do as I do training. So basically they've already learned socially from us. It's just that we have never used this in training and now we can do it.
But in my experience this doesn't happen. Because with Do as I do the dog learns that you are happy if he imitates when you say Do it and not if he imitates whatever. But I think that the dog might learn to open the door or open the litter bin even if don't want also without the Do as I do if he has his own purpose for that.
How long does it take to teach a dog to Do as I do?
I really depends (how long it takes) on what is the baseline that you have. I always suggest that before starting the Do as I do training the dog at least knows three behaviors on the verbal cue. And if the dog knows these three behaviors on the verbal cue it might take six sessions of training for the dog to understand and generalize the rule of imitation. But if the dog doesn't know these three actions that he can use as samples for mental representation of the matching between the demonstration and his own action, then it's longer. So my suggestion is that you start from this baseline with at least three behaviors on the verbal cue. This speeds up the process.
Once the dog learned the rule of the training, is there room for further improvement?
I think that the dog… the more experience you give him, the better he gets. And the more your dog can face the new experiences in general, in life, the better he can find solutions for the tasks. Because he is open-minded. I mean, open-minded is not a scientific word, but he has many connections to use in his brain to face the situation. So I think the training is always going on.
In Do as I do training is it important for dog to undestand the goal of the action demonstrated by owner?

Basically with Do as I do training you can train your dog in such a way that he matches the goal. So, for example, if you train your dog to open a drawer you can reward the dog when he socially learns to open a drawer. And he can use his paw or his mouth, depending on what is the situation. And this way his behavior is more flexible, because he has socially learned what is the goal of the action, and therefore he can now adapt his behavior to what is the contingency. So, for example, if the drawer is high he can use his paw, because he cannot reach it with his mouth. But if the drawer is in the lower position he can use his mouth, because it's more functional. And we already know from some prevous studies that dogs can adapt their behavior according to the situation in what has been called selective imitation [2].
What dogs cannot learn by imitation?
We don't know… in a scientific way, because we are now studying what are the actions that best, or more efficiently trained with the Do as I do training and with others associative, traditional training techniques.
But my experience… the actions that are only body movements or not so efficiently trained with Do as I do, because our body schema [3] is very different from the dog's one. And because it is not so obvious that the dog can find, for example, a matching between my arm movement and his paw movement. So this is probably the actions that are not the best choice for the Do as I do.
How can dog's experiences of everyday life help in Do as I do training?
In my opinion, the more you allow your dog to experience different situations of social learning, of individual learning and of life, so also knowing the environment, exploring the environment, the [better] your dog can [learn further]. And it is really against what old-fashioned trainers told that if the dog is doing agility you should not try obedience, because the dog can mess up what he knows about agility.
But it's really the opposite. The more experience the dog has in different situations, not only training, but also traveling with the owner in the city, or having nice walks with the owner in different environments, the more he has psysiological chances of better succeding in different situations, also in training.
What we have to take into account when it comes to motivation for social learning, besides traditional rewards (food, toys etc.)?
Dogs and any [living beings] need to learn from the very moment they are born in order to survive. So it's about surviving. We all need to learn. Then of course, it's about what the experiences you are allowed to do or you are not allowed to do. But learning is about surviving. And for the dog in particular, from the evolutionary point of view, the human has always been a relevant stimulus in the environment, because at the beginning the food came from humans villages. Because it was the human waste, for example. So, probably in the dog's mind humans [themselves] are somewhat anyway relevant, [besides rewarding in training].
In traditional training it's been emphasized the importance of timing — in order to be learned the behavior should be rewarded quickly. And it's very different from what you have shown in studies on deferred imitation, when dogs were supposed to imitate owners' actions after a considerable delay.

I often get this question. So it's about two different things. It is true that the dog in associative learning [4] has to be rewarded in very short time. Because he needs to associate the reward with the behavior he's performing in that moment. And this is about associative learning. And there are also some exceptions. Because if, for example, we know Garcia's studies about disgust, about learning disgust of the food [5]. So in these experiments what they did was they gave to some rats food that was with something in it that made them sick. But this made them sick after several hours. But still they associated the taste of the food with sickness. So they avoided this food. And this is an exception in associative learning. So remember that this is an exception. And this is something that then caused problem to the whole theory of associative learning.
But in my case, in my study on deferred imitation [6] we don't create any association. So basically the study on deferred imitation demonstrate that the dog is able to observe a demonstration and recall it after delay of ten minutes, for example. And this is not an association. This is just memory of actions. Which we didn't know. So the results are quite surprising, because didn't know how far could we go. And the ten minutes result is just surprising and we are going to study with longer delays and we want to see when they forget.
In the seminar you mentioned that you had to be very careful with using the word imitation when it comes to non-human animals. Now there's evidence for imitation in dogs, many scientists still do not accept it…
So they are I would say skeptic. But this is not bad, because this force us to really put together straight protocols that really test what we want to test. So I don't see it in such a bad way. I know that this is a sort of challenging issue… to test or to study dog's imitation, because many scientists also believe that dogs cannot imitate and that only humans can imitate [7]. But this force us to really use straight protocols that allow us to really test only for imitation and to control for other learning processes. So this is not so bad in a way. And I think in science it should be more or less like this.
Afterword
We all are grateful to Claudia for the seminar. It was an inspirational one indeed! The facebook group Do as I do Austria was established, so that people can share their experience of Do as I do training. Some participants have even extended the Do as I do technique to other species:
Do as I do exam and Research
There's an opportunity to take a Do as I do exam. The exam attests:
- understanding of theory behind the Do as I do training (it's sufficient to read Claudia's book to pass the theory)
- dog's comphrehension of imitation rule (the dog should be able to copy actions even demontrated by different strangers; taking a seminar for advanced Do as I do training is highly recommended)
Seminar for advanced Do as I do training is going to be held in October, 19th-20th, 2013 (Read more).
If the exam is successfully passed, participant receives the certificate signed by Claudia Fugazza and prof. Ádám Miklósi.
Once the participant's got the certificate, he or she is welcome to take part in ongoing research being carried out by Claudia Fugazza and her colleagues from University of Budapest.
Further information and registration for Do as I do exam: This email address is being protected from spambots. You need JavaScript enabled to view it.
Footnotes
- 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. (Download).
- Range, F., Viranyi, Zs., Huber, L. 2007. Selective imitation in domestic dogs. Current Biology, 17: 868-872. (Download).
- Body schema.
- The two forms of associative leaning are classical conditioning and operant conditioning.
- Conditioned taste aversion.
- Deferred imitation.
- Zentall, T. & Akins, C. (2001). Imitation in animals: Evidence, function and mechanisms. In R. G. Cook (Ed.), Avian visual cognition [On-line]. Available: www.pigeon.psy.tufts.edu/avc/zentall/.
Further reading
- Claudia Fugazza. Do as I do. Machs mir nach! Hunde lernen, indem sie uns beobachten. Verlag Dogfriend Publishers, 2012 (German edition).
- Ádám Miklósi. Dog Behaviour, Evolution, and Cognition. Oxford University Press, 2009.
- Eötvös Loránd University, Department of Ethology. Family Dog Project. Publications.
- Clever Dog Lab. Publications.
Intetview: Olga Kajarskaia
Transcript, text, photos and video: Vitaly Samigullin
Dogfriend Center / Verlag Dogfriend Publishers.