The days of training a dog with treats and pats could be numbered, according to a new study which shows that dogs are sometimes quicker at learning by imitation than by the old fashioned reward and punishment, or clicker learning approach.
The study is the latest to show that the most popular approach to training dogs underestimates the canine ability to imitate behaviors that they see their masters perform. It's also the first to put what's called the "do-as-I-do" approach for training dogs right alongside the old operant conditioning approach, in which dogs behaviors are directly reinforced with the aid of a clicker.
If the new research is confirmed, it could mean more legitimacy for the do-as-I-do training method.
"We compared the efficiency of the do-as-I-do method, which relies on social learning, with that of a training method that relies on individual learning (shaper/clicker method) to teach dogs three different kinds of object-related actions," explains Claudia Fugazza and Adam Miklosi of Eotvos Lorand University in Budapest, Hungary.
In order to keep the two kinds of training comparable, the researchers enlisted dog/owner pairs who had previously earned a certificate for either type of training.
They were tested on training new actions ranging from simple to complex and sequences of two actions, in three separate sessions, using the training method for which they were certified. In each case the owners had 15 minutes to train their dogs to perform the new actions.
"Enrolling unexperienced dogs would have meant first exposing them to preliminary training, before both the methods would show their best results," said Fugazza. "Testing (pairs) that achieved a training certificate for either method, we were able to have two groups that were comparable regarding their skills and experience in training. Of course this cannot completely rule out individual differences in owners' skills, but this was the best way to eventually minimize them."
The traditional shaper/clicker approach first trains dogs to associate a loud clicker sound with a treat, in true Pavlovian style. Then the clicker can be used as a positive reinforcement for the right behaviors as a dog is trained to perform parts of what together become more complex tasks.
The do-as-I-do method, on the other hand, involves initially conditioning dogs to pay attention to the owner who performs whatever action they want the dog to learn. The dog is trained to then try and imitate an action when told "Do it!" The owner's performance and then the dogs attempt to imitate it are repeated until the dog gets it right.
"While we did not find a significant difference between the two training methods with regard to simple actions, we found that subjects using the do-as-I-do method outperformed those using shaping/clicker training in the case of complex actions and sequences of two actions," the researchers reported in the latest issue of the journal Applied Animal Behavior Science.
Just how useful the distinction is between training methods remains to be seen, however.
"In reality you're pulling from different learning methods," said veterinarian Liz Stelow of the University of California, Davis. "Very rarely are you seeing 'I'm going to teach in one particular way.' We all intuitively go with what will work with a particular dog."
That said, Stelow says there is always a push to find a better, quicker way to train dogs -- especially assistance dogs. And since dog breeds, as well as individual dogs, vary in how they learn, she would like to see a larger study comparing the training methods which involved more dogs and more breeds.
"We believe that studies like this should benefit the practitioners working with dogs," said Fugazza. "This is a very first step in this direction, at least regarding the do-as-I-do method, and unfortunately such studies on training methods are lacking in general."