Title: Clicker increases resistance to extinction but does not decrease training time of a simple operant task in domestic dogs (Canis familiaris)
Authors: Shawn M. Smith, Ellen S. Davis
Published: 2008 Applied Animal Behaviour Science
These researchers studied the effects of using a clicker during dog training, as these devices are popular with dog owners but have not been scientifically evaluated. The authors identified three possible mechanisms that clickers may serve in the dog training process (definitions found here):
- A conditioned/secondary reinforcer: A neutral stimulus paired with a primary reinforcer until the neutral stimulus takes on the reinforcing properties of the primary.
- A marking signal: A signal used to mark desired behavior at the instant it occurs.
- A bridging stimulus: An event marker that identifies the desired response and “bridges” the time between the response and the delivery of the primary reinforcer.
In order to investigation the mechanism of clicker training, the authors recruited 35 basenji dogs that had never been exposed to a clicker. Eighteen dogs were assigned to the clicker group and the remaining 17 were assigned to a control group. The authors used a trainer to condition the clicker group dogs that the click was associated with food delivery in a nearby bowl, while the control group dogs were simply given rewards in the bowl with no click. Then the trainer taught the dogs a behavior (nose touch a cone) and conducted strengthening trials to reinforce the behavior where the dogs were intermittently rewarded for correct responses to the nose touch cue. These trials were followed by extinction trials in which the dogs were given the nose touch cue but not rewarded.
The authors found that there were no differences between the clicker and the control groups in the number of trials or the time needed to learn the nose touch cue, which suggested that the clicker did not serve as a marker or bridging stimulus. It did, however, the clicker group dogs significantly longer to achieve extinction – that is, to stop nose touching the cone after the nose touch cue was no longer reinforced with food. This suggested that the clicker did serve as conditioned/secondary reinforcer, possibly because the clicker dogs were facing “double extinction”: they had to unlearn both that the clicker did not result in a reward and that the nose touch did not result in a reward.
The authors noted that dogs in both groups were obviously responding to hand movements from the trainer that were associated with delivering food to the bowl, which may have interfered with the dog’s ability to associate the clicker with the food delivery instead. The authors also had a significant difference between the age of dogs and the time/number of trials needed to learn the nose touch behavior: younger dogs were faster than older dogs.
This was an interesting study because, as the authors noted, it is one of the few that scientifically evaluates the mechanism by which a clicker may facilitate learning in dogs. (How do dogs learn?Check out Crash Course Psychology Episode 11 and Episode 12.) However, I had some concerns about how the authors chose to evaluate the clicker in the context of owner-dog training.
First, the protocol that the trainer used to teach the dogs a nose touch behavior involved the reward being delivered in a dish rather than directly from the trainer. Why?? The authors did not explain this adequately. I wonder if this may have impacted the dogs’ learning based on the information I have read from Dr. Yin about the importance of timing and posture during treat delivery when training dogs. The authors even noted that the dogs were apparently watching the trainer for hand movements that indicated treat delivery to the bowl, which aligns with Dr. Yin’s assertion that treat delivery is highly influential to a dog’s learning experience. Furthermore, I have never heard of an owner training their dog in this manner – therefore, the validity of applying the results from this study to a typical owner training their dog is questionable.
Secondly, if the goal of this study was to evaluate the utility of a clicker in training by the average owner, using a professional dog trainer likely produced results that are not valid outside of that context. A dog trainer will likely have skills that the average owner does not, such as effectively applying cue-reward protocols where a regular owner may have faulty or incorrect application simply because of experience. Because owners may be less accurate during the cue-reward process of training than dog trainers, it is possible that dogs could pick up on the clicker as a marker or bridging stimulus in the absence of other effective instructions. Additionally, other researchers have found that dogs are more attuned to their owner’s social cues vs. a stranger’s behavior. If the purpose of this study was to evaluate clicker’s use for dog owners, the dog’s actual owner should have been used in the training process.
In summary, I agree with the authors that the mechanism of clickers in dog training is an interesting and useful avenue of study. Clickers may be a useful tool in promoting humane, effective dog training for the average owner but incorrect usage of the devices may frustrate owners. Understanding how clickers help (or don’t help) dogs learn could aid behaviorists and trainers to advocate effective and safe training methods to owners. While I do not believe this study’s findings can be applied to a owner/dog training scenario, it is an interesting contribution to the understanding of how clickers impact canine learning.