Training ‘Do As I Do’: Fun and Efficient!

Through interactions with the world, animals learn that certain actions lead to pleasant or unpleasant consequences. Most of our training theory revolves around our ability to manipulate the animals’ environment to reinforce certain behaviors rather than others. However, not all learning happens from direct experience. As humans, there are many behaviors that we learned indirectly, through observation of others. All children watch their parents and siblings and from a very early age, start to imitate them. In fact, a very large number of important skills are learned through imitation. That’s how we learn how to walk and talk! For the longest time, scientists have argued that imitation is what makes humans unique from other animals. After all, the ability to imitate another being involves a certain level of self-awareness and empathy, as one has to recognize in the other the same body parts and be able to move them in the same way. So what we do so easily is in fact a fairly advanced process. Studies are now showing that non-human animals are indeed capable of learning through imitation(‘Do As I Do’), but that when it comes to learning complex behaviors, teaching by imitation may also be quicker and more efficient than reward based training (associative learning). It’s time to add a new tool to our trainer’s toolkit: training through imitation!

I’ve often heard Ken Ramirez describe how we can teach concepts to animals. Dogs can learn to distinguish between left and right or between big and small. Since 1994, Ramirez has also developed a training protocol based on teaching animals to imitate one another. One dog would perform an action, like sit or go to a crate and another would repeat that action when given the cue ‘copy’. Although never studied from a scientific perspective, his training protocols clearly put in question the idea that humans are the only ones capable of such advanced cognitive processes.

What if dogs were not only capable of imitating other dogs, but also to learn through observing humans? We already know that dogs can respond to pointing gestures or that through social referencing they will adapt how they feel about a situation based on our reactions. So why not consider that we could also teach our dogs to perform certain behaviors by simply showing them first? Wouldn’t training be much easier if we could just show our dog what to do?

In 1952, Hayes & Hayes had developed a procedure, called ‘Do-as-I-do’ in an attempt to demonstrate how chimpanzees could learn through imitation. This study was put into question for the subjectivity of the interpretations of the chimpanzee’s behaviors, but it provided a method that has since been tested on other species. It’s only recently however, that the ability to imitate has also been tested in dogs. Using the ‘Do-as-I-do’ protocol, a group of researchers have demonstrated that dogs are indeed capable of reproducing a completely new sequence of actions from observation only (Topál & al., 2006).Teaching through imitation turns out to be both effective but also a faster teaching tool than shaping when it comes to teaching certain complex behaviors, such as opening drawers, picking up objects or turning on or off light switches. (Fugazza & Miklósi, 2014).

Before you try to teach your dog a new behavior by simply showing your pooch what to do, there are a few fundamentals to put in place. The dog needs to learn the concept first, in other words, unless you have a cue for your dog to imitate you, Fido may just sit there with a blank stare. Here are a few guidelines from Claudia Fugazza, but for a more detailed description of the protocol, I would recommend her seminars or getting her 8 hour DVD set ‘Do as I do, a new training method based on social learning.

How to get started:

1. You’ll first need three distinct behaviors that your dog knows fluently. Make sure the dog can perform the behaviors from the verbal cue alone (or hand signal), without any prompts, not even a slight body movement (like a head turn). Be sure to use behaviors that you can easily reproduce. Our anatomy is very different from the dogs’ so body movements can be very arbitrary. The behaviors also need to be distinct. A sit and a down, for instance can be problematic since many dogs will sit first before they lie down. So when they don’t get rewarded for the sit, they’re likely to lie down naturally and you won’t be able to tell if the behavior was triggered by the demonstration or by the sequence. You’ll also want to stay away from behaviors that are self-rewarding or a toy that the dog likes a lot. This may distract the dog from focusing on the protocol. Once you have found the right behaviors, practice them a few times before you start.
2. Place the dog directly in front of you. During the initial phase, when the dog is learning the concept, it’s critical that you and your dog always stay in the same starting position, preferably facing each other. Like other forms of training, you want to minimize cues that the dog may learn to rely on that have nothing to do with the concept itself. So if you look to the right when you say ‘spin’ and look to the left when you say ‘down’, your dog may rely on your orientation and not learn the concept.
3. If you’re using objects (retrieve or ‘touch’), make sure to have all objects present during the session and place them at equal distance to the dog. Here again, you don’t want to make one object more salient than the other.

The protocol:

1. With your dog in a ‘sit’ or ‘down’, give him/her the cue ‘stay’ (previously learned of course) then demonstrate the behavior.
2. Get back to your initial position. Give the cue ‘Do it’, followed by the cue for the known behavior. So if you demonstrated ‘spin’, you’ll say ‘Do-it’, then ‘spin’.
3. The dog should perform the known behavior.
4. Reward the dog.

Repeat this sequence with all three of the behaviors while avoiding to ask for the same behavior more than twice in a row. You’ll want to avoid the dog associating the words ‘do-it’ with the behavior. Once the dog starts to perform the behavior immediately after you say ‘do-it’, before you give the cue to the behavior, it’s time for the second phase.

You’ll need 3 more previously trained behaviors for this phase and will add them to the first thee all at once. This protocol helps the dog generalize the concept ‘Do-it’.

Once the dog consistently offers the behavior with the cue ‘do-it’ alone, the hardest part of the training is done. It’s now time to introduce new behaviors.

1. Demonstrate the new behavior.
2. Give the new cue
3. Give the cue ‘do-it’
4. The dog should perform the new behavior that you have demonstrated
5. Reward!!

To put this new behavior on cue, after a few repetitions, you’ll simply stop saying ‘do-it’ and only give the new cue. Finally, you can stop the demonstration altogether and the dog should respond to the new cue alone.

At this point, getting the dog to perform new behaviors is easy and you wouldn’t need to pay attention to your orientation or place different objects on the ground anymore. Once the concept is learned you can play around and be more flexible. You may, for instance, decide that your criteria is to simply reproduce what you did, regardless of how he gets there. You may also decide that you want the dog to match your behavior more precisely, so if you touch an object with your hand, the dog should touch it with a paw.

The ‘do-as-I-do’(DAID) protocol has been shown to be very effective and doesn’t require as many repetitions as with regular training, especially when it comes to teaching sequences of actions that are object-related, like ‘open a drawer’, ‘close a door’ or ‘pick up an item and put it in a basket’. Dogs seem to remember the cue better when linked to a demonstration. According to Ramirez however, the downfall could be that since the behaviors are learned in whole, it’s hard to fine tune via shaping. More studies on the subject will be needed however as there may be more differences that are task related. Nonetheless, for those of us looking for new and innovative teaching tools and new ways to interact with our dog, this is it!

Jennifer Cattet Ph.D.

The following is a collection of many videos demonstrating imitation training. You can exit at any time but be aware there are different videos in this collection.

About Jennifer Cattet Ph.D.

Jennifer Cattet Ph.D. has been working with dogs for over 30 years, as an ethologist with the University of Geneva (Switzerland), a trainer and a behaviorist (in both Europe and the US). As Director of Training for a service dog organization in the U.S, she supervised and taught offenders in the training of service dogs. Today she's the owner of Medical Mutts (MedicalMutts.com), a company dedicated in the training of rescue dogs as service dogs for conditions such as diabetes, seizures, PTSD, autism, etc. She's also part of a research team working on understanding the ability of dogs to detect changes in blood glucose levels through scent. Jennifer also works with Smart Animal Training System on the promotion of reward based training and the development of technology to support it (SmartAnimalTraining.com).