A provocative question
Several weeks ago I was reading Dr. Patricia McConnell’s blog, which is always insightful and fascinating. I was intrigued by her topic, When is it Acceptable to Break the Basic Rules of Dog Training? (November 30, 2015). She asked her readers to weigh in on this question, and it prompted a lively discussion.
In response to Dr. McConnell’s question, I thought I would share a few “rules” that relate to the use of a clicker. Some are misunderstood rules, and others are not really rules at all, but are perceived as such by those who misunderstand the purpose of the clicker.
“A positive reinforcement trainer must use a clicker”
Recently, I was contacted by an organization to present a seminar for their group. But, when they saw “Clicker Training” in my company’s title, they almost cancelled the engagement—because their organization does not use clickers. I assured them that although we certainly find clickers to be a useful tool, we do not insist that they must be used. In fact, it is important to remember that there is no magic in the clicker. The clicker is simply a toy noisemaker. It is a useful tool for marking behavior and for communicating information to the learner.
For simple behaviors like “sit” or “go to a mat,” I may not be looking for precision, so sometimes I train those behaviors without using a clicker. With some cats I have worked with, the click was too startling and aversive, so I avoided the clicker and either eliminated the marker or used a softly spoken “yes” as my marker.
Animals are always looking for feedback that can lead them to reinforcement. If we don’t teach a deliberate marker, they will interpret other things in their environment as markers. The trainer reaching into the food bag, the trainer stepping forward, or something as insignificant as a big smile on the trainer’s face, could all be misunderstood as markers. Of course, this same misinterpretation occurs if we use a marker badly. Animals will find something that, to them, “accurately” predicts reinforcement.
The use of a marker is encouraged, because of how much it aids the training process and sets up the animal for success when used properly. However, good training can certainly be accomplished without a formal marker.
“The click must ALWAYS be paired with food”
I follow this rule almost all of the time. To be explicitly clear, though, I would state the rule differently: the click should always be paired with reinforcement.
If inexperienced trainers are uncertain of what an animal finds reinforcing, or they don’t know how to evaluate various reinforcers, it is safest to suggest that they always follow the click with food. Once they truly understand reinforcement with their animals, the click can be followed by other reinforcers (a tennis ball, a tug toy, or any number of other non-food reinforcers— and, in some cases, even the cue to do the next behavior). However, jumping to these steps can be challenging for a beginner; there is the risk of making an assumption about what an animal finds reinforcing. If we follow the click with something that is not reinforcing, we will see the power of the clicker weakened. I created my Karen Pryor Academy Smart Reinforcement course specifically to help trainers learn how to evaluate and establish effective non-food reinforcers.
“The click ends the behavior”
When we click to mark a specific behavior, usually it causes the animal to stop exhibiting that particular behavior and return to the trainer for reinforcement. Unfortunately, I encounter trainers who take this statement to mean that what happens after the click is irrelevant. It is important to remember that the dog is still behaving, and other behaviors continue or begin after the click. If we fail to pay attention to what happens after the click, we can reinforce unwanted behavior inadvertently. A common example is when a dog, after having performed a behavior correctly and hearing the click, starts barking in excitement as he approaches the trainer for the treat. If the trainer takes the statement that the “click ends the behavior” too literally, s/he will reinforce the dog, and the dog will learn to bark after that behavior.
“If you click, you must reinforce”
A variation on the previous theme is the beliefs that the click is a promise of reinforcement and that once you make that promise to the animal you should never break it. This is an excellent way to look at the clicker, and a good rule to follow. However, what happens if the animal performs a behavior incorrectly, but the trainer clicks the behavior accidently? The trainer is now facing a conflict between two different principles or rules:
- I clicked so, I must reinforce.
- The animal performed an incorrect response, so I should ignore that behavior.
We face dilemmas like this one all the time in training. In these instances, the trainer must recognize the principles underlying each rule as guidance in making the right decision. The answers are not absolute and the rules are not intended as inflexible edicts.
If I click by mistake, I recognize that I have already reinforced the animal with the click accidentally. I don’t want to compound the error by adding more reinforcement. If the clicker has a strong reinforcement history, failing to pair the click with a reinforcer one time will not impact the power of the click negatively. However, it is important that I do not keep making that mistake. If I repeat that mistake too often, then I would consider trying to work on my mechanical and timing skills, through training games or exercises without the animal present. In general, the “If you click, you must reinforce” rule is one that I insist on for beginners and continue to encourage for experienced trainers. But like so many rules, there can be times when it is wise not to follow the rule blindly.
“If you use a clicker, you must use it forever”
There are times when I choose to discontinue the use of the clicker for certain behaviors. In many cases, the termination of the behavior serves as a marker for the animal, and there is no need to keep clicking for it. For example, when a dog’s rear end hits the ground after being asked to sit, that completes the behavior and a click is not needed; when a dog retrieves a toy, taking the toy from the dog serves to mark the completion of the behavior. When I trained agility exercises, I used the clicker to shape precision for complex behaviors. But once I was performing with my dogs on the finished course, I did not use or need the clicker. The clicker is not necessarily required after behaviors have been learned.
“Discontinue using the clicker once behavior is learned”
The converse to the previous rule is the belief that you should quit using a clicker once any behavior has been learned. However, there are plenty of instances when a clicker can be useful well after the behavior has been learned to fluency. If a behavior is not self-terminating, meaning it could continue until the animal is instructed to stop, the click could be useful forever.
Frequently, I train behaviors to be continuous. Behaviors such as spinning, barking, waving, and targeting are examples of behaviors that I teach animals to keep offering until I click. Certain types of behaviors designed to help with medical treatments (ear cleaning, nail clipping, eye drops, etc.) are other behaviors that need to be worked on for the duration of an animal’s life. The animal must learn to accept different types of touch, different durations of treatment, new people (veterinarians and technicians), and various objects. In my opinion, this area of training is never actually completed, because the desensitization process never ends. It is helpful to keep using a clicker in these cases.
Rules exist for a reason, but there can be good reasons to break them
As an instructor and mentor, I give my students rules to follow to help them succeed as trainers. While I do expect my students to follow the rules closely when they are just getting started, I want them to learn the underlying principles that brought the rules into existence in the first place. I also want them to know when it is acceptable to break or bend the rules.
Good training is based in science, and there are laws of learning that are absolute! But the rules that we put in place often are simply our own interpretations of how to work with those laws and how to interpret specific situations. As rules are subject to error and human bias, we have to be flexible and smart about how literally we take some of the rules of training.