Dog Training Methods: Clicker Training
Why Clicker Classes Are Ineffective And Potentially Dangerous
Pavlov's Dog
“Clicker training has not taught me a whole bunch, other than that people can get wrapped up in fads and catch phrases.” —Bob Bailey
Can Clicker Training Be Bad For Dogs?
Originally published in different form on September 16, 2010 at PsychologyToday.com.
I Can’t Get No Satisfaction
About ten years ago I wasted a fair amount of time on line discussing training issues with other trainers and trainer-wannabes on various dog behavior boards, and came I across an interesting anomaly. Those who extolled the virtues of clicker training in one thread would commiserate with each other, on another, about their inability to curb their dogs’ pica (the compulsive ingestion of non-food items), scavenging and counter-surfing behaviors. This suggested to me that clicker training might cause food-related behavior problems in dogs. This has turned out to be the case. Every single dog owner I’ve talked to recently, whose dog exhibits serious food-related behavior problems as an adult, was either mistreated, abused, or clicker-trained as a young pup. In fact, last year I had three clients whose dogs had all been to the same clicker-training academy in New York City, and had all developed food-related behavior problems as a result.
One case was quite severe. The dog was compulsively eating rocks and had been hospitalized several times, and had nearly died twice. When II asked the client if her dog had been clicker trained she said yes. She’d taken her dog to a clicker class when the pup was 8 weeks old.
“That’s probably why she’s compulsively eating non-food items.”
She didn’t understand, so I asked, “Did the clicker trainer say anything to the class about Pavlov’s dogs?”
“Pavlov’s dogs?”
“Yes, when she introduced the clicker?”
She thought a bit, then said. “Oh, yes. They salivated at the sound of a bell?”
“Yes, sort of.” (Pavlov never actually used a bell.)
Then I explained how unpleasant it must have been for those dogs to salivate every time the bell rang. And, since the dogs were salivating, that means the sound of the bell stimulated production of the hunger hormone, ghrelin. Ghrelin stimulates hunger but it doesn’t act alone. If the feeling of hunger isn't satisfied (by eating something), then ghrelin stimulates production of cortisol, the body’s stress hormone. This shows that clicker training—if not done with absolute precision—can be stressful for dogs. And it's why some dogs develop eating disorders as a result of bad clicker training.
So while clicker training does create a Pavlovian response, that’s not necessarily a good thing because it essentially creates the expectation of imminent satisfaction, but doesn’t deliver the goods.
Deferred Gratification & Artificial “Rewards”
The idea of using a clicker to shape animal behavior originated in 1946 when Keller and Marian Breland—both former students of B. F. Skinner—noticed that Skinner’s operant conditioning chambers (also called “Skinner boxes”) made small noises when delivering “rewards” to the animals trapped inside them. The Brelands wondered if these sounds might be acting as an acoustic “secondary reinforcer” for the behaviors the animals were “learning.” (Primary reinforcers are said to satisfy deep survival needs while secondary reinforcers are said to be the result of an association made between a primary reinforcer and an unrelated stimulus; thus the clicking sound used by clicker trainers is sometimes called a secondary reinforcer, sometimes a “bridging stimulus” and sometimes a “marker.”)
Keller Breland was the first to apply a clicker to dog training. He used it when working with field dogs at a distance. Despite the fact that he believed the clicking sound acted as a “bridging stimulus,” it could also be seen as a way to communicate binary, i.e., yes/no, information to the dogs in real time. But that’s not the way clickers are used in puppy classes. They’re used to tell the pup, “I can’t give you a reward right now, so please use this clicking sound as a temporary substitute for real satisfaction.” This means that clicker trainers are essentially expecting puppies to engage in a process called deferred gratification, a form of impulse control. And puppies are notoriously bad at it. Why?
Because they’re puppies!
A Feeling of Flow
Keep in mind that good working dogs, like the ones Keller Breland trained, will keep working no matter what. They don’t do it to gain an external reward at some later time but to maintain the pleasure of being in a state of flow, of being on the brink of excitement. That feeling is their reward. So the lack of immediate pay-off inherent to clicker training is irrelevant to them. That’s because when dogs are in this kind of state, any information, even something that would normally feel punitive in other contexts, will actually increase the dog’s feeling of flow. This is probably why clicker training has fewer side-effects when it’s used in training dogs for agility. In fact, when working dogs are working (and agility training gives most dogs that feeling), they’re capable of converting whatever form of stress they’re feeling into a feeling of flow. (This concept, of converting stress into flow, was first proposed by Kevin Behan in his 1992 book, Natural Dog Training, and has recently been validated by a newly-discovered principle of physics called the constructal law.)
This is also why clickers “work” so well with dolphins and orcas. Most of the behaviors these aquatic animals are “taught” entail innately satisfying forms of movement similar to what they naturally exhibit in the open ocean. Even mirroring behaviors, where they’re asked to mimic the trainer’s stance, posture, arm movements, etc., are also inherently satisfying because both dolphins and orcas enjoy synchronized movements with members of their pods. (And if you want to see the feeling of flow in action, watch footage of a group of dolphins or killer whales zooming through the ocean at full speed!)
Meanwhile, when puppies are trained to do tricks in a clicker class the clicker actually impedes whatever brief feelings of flow they’re capable of experiencing. There’s only one behavior that comes close to the feeling that working dogs, dolphins, and orcas get, and that’s twirling around in circles. Owners of clicker-trained pups tell me it’s the only behavior their dog actually retained. The dogs will sit or lie down or roll over, all in one blur (rather than pairing the behavior with its actual command), but when asked to twirl, they twirl. They love to twirl!
I think the reason for this is clear. In a previous post, “How Dogs Learn, Part 1,” I pointed out that dopamine, the brain’s “reward” chemical, is not necessarily released in association with rewards. It’s often released in the absence of a reward, and especially in the absence of an expected one. So dopamine is not a “reward” chemical, per se. It wants us to pay attention to changing patterns in the environment, positive and negative. That’s how animals learn.
But dopamine gets its effects by creating pleasurable changes in internal feeling states. And those changes create a feeling of flow. (In fact, how does dopamine get released into the brain? It flows!) So there’s a deep connection between learning through pattern recognition and being in a state of flow. And that’s why puppies retain their love of the “twirling” command while forgetting most of the others. It’s the only command that gives them that weightless feeling.
The Known Dangers of Clicker Training
Of course, I could be wrong. The incidents of ADHD and pica that I and others have seen may not bear a direct causal relation to clicker training. The thing is, though, I’m not the only trainer to bring up these issues. Even positive trainers have warned about the known dangers of clicker training.
Here’s what one clicker trainer had to say: “The clicker is a precision instrument. After years in which I used the clicker intensively, I learned to use it only to teach a new behavior, or to reestablish a behavior that has been forgotten. I use it for a few minutes per week, after 7 months of age.”
This framework, of using the clicker once the dog has passed through his puppy stage, is much less harmful than clicker-training a very young puppy.
Bob Bailey, who was a colleague of Keller and Marian Breland, and who probably knows more on the subject than almost anyone, says, “The proper application of the clicker is that akin to using a scalpel to make fine cuts.”
It takes nine years of study to become a surgeon. What’s required to become a clicker trainer? Not much. An online course or two, maybe a seminar. (The trainers at the clicker school I mentioned earlier didn't study dog training with a knowledgable dog trainer, they were taught dog training by a former dolphin trainer!)
What about the average dog owner? I was shocked and appalled recently when, at a session with the owner of a new, 8- wk. old pup, the puppy piddled on her wee-wee pads, and the owner rushed over and started furiously clicking at the puppy, over and over and over, terrifying the poor little thing! I asked where she got the clicker and why she thought it was a good idea to shove it in her puppy's face, and she said she picked it up at the counter of her local pet store. And she was trying to make sure the puppy remembered that weeing on the pad was a good thing, and that she was being rewarded for doing it.
If the clicker is a precision instrument, why are we putting it in the hands of people who don't know how to use it properly? Most pet stores now have bowls of clickers available at the front counter. There are lots of internet sites extolling the virtues of clicker training. But there are also sites where dog owners ask for help undoing the problem behaviors caused by clickers.
A Clicker Can Easily Get in the Way
Nina Bondarenko is a highly-respected positive trainer living in London. In 2005 she wrote: “Are trainers ‘Shaping for Stress?’ Does your dog go on strike during a shaping session, or ‘get crazy?’ Does the behavior become unnecessarily complicated? Trainers have a responsibility to follow guidelines that ensure they are truly positive trainers, not ‘killing with kindness.’ … Behavior can quickly become frantic, aggressive or unfocussed, causing learning to get off-track.”
Does any of that sound positive?
Bob Bailey again: “Most pet owners seldom have need for a clicker, in my opinion; a clicker can easily get in the way of getting good behavior.”
Not only can it get in the way, it can potentially cause severe behavioral problems, and in some cases even death.
Lee Charles Kelley
“Life Is an Adventure—Where Will Your Dog Take You?”
Discussion:
I posted a link to an earlier version of this article on my Facebook page. Here are most of the comments I received, and my replies.
Johnny Fleckenstein: Are you sure the Daniel Badeski quote is a “warning” of the “known dangers of clicker training”? Not only does Badeski say nothing negative at all about clicker training, but he also says this:
“The clicker does not remove the social and emotional communication, its only purpose is to add useful information to the dog. The method works through the use of positive reinforcement with total respect for the dog’s physical and emotional well-being.” and this: “But the clicker can do much more: it can help dogs develop their mental skills (ability of association, memory, attention), emotional skills (stress, frustration, excitement, anxiety) and social issues, particularly communication with people. The clicker can increase self-esteem, confidence, self control, and even help dogs with behavioral problems.”
LCK: Yes, I know that Badeski is an avid clicker user, but he also points to the danger of clicker training puppies while they’re too young. And he makes it pretty clear that he uses it judiciously, where the training school that seems to have created the problem behaviors I’m discussing here doesn’t.
Also, like thousands of trainers, Badeski is wrong in thinking that learning takes place via positive reinforcement.
Johnny Fleckenstein: I enjoy following you and your articles because I think it’s nice to see things from a different perspective to keep my own opinions grounded to something (I respectfully disagree with many things you write). In following you and researching Natural Dog Training, I’ve seen a few of your critics talk about how you specifically tend to “cherry-pick” quotes from people and present them in a way that supports your point. It appears that’s exactly what you have done to Daniel Badeski here in your article (and curiously withheld his name?). The point of the article that you took the quote from is actually the complete opposite of yours, and I think it’s really unfair the way you misrepresented him.
LCK: I disagree that the point of the Badeski article is the “complete opposite” of mine for the reasons I cited earlier. I would argue that Badeski (and Bailey) are inadvertently—and I’m sure quite unintentionally—making several of my arguments for me.
And I have no interest in what my detractors are saying unless their arguments are on point. Then I might listen and engage in a discussion with them (as I’m doing with you). As for “cherry-picking” quotes, most biological scientists have been doing that for years with Darwin (I’m referring to the “differences of degree-not kind” meme). I also try to keep my posts under 1500 words so I’m not obligated to offer readers all possible arguments against what I’m saying, just to make my points in as economical and easy-to-understand way as possible. Plus, the same detractors who complain of my selective quotations do the same thing to me, and worse!
As for not citing Badeski by name, I looked for his bio but couldn’t find it. (I was looking up top instead of on the sidebar.)
Finally, the definition of learning via +R is that reinforcers increase the likelihood that a new behavior will be repeated. My definition of learning is that once a dog has learned a new behavior there’s no longer any need to keep “reinforcing it,” or, as Badecki writes, “to reestablish a behavior that has been forgotten,” which he does with the clicker. My questions is, if a behavior has been forgotten, has it really been learned? I learned to play the guitar when I was 18. My fingers may get a little rusty if I haven’t played in a while, but I have never forgotten how to play.
Here’s another conundrum. If a) positive reinforcement is the basis for “the science of how animals learn,” and if b) clickers are a form of positive reinforcement, and c) clickers are used to train marine mammals to do tricks for the crowds at SeaWorld, then why are clickers still used constantly during their routines? If these animals had actually learned the behaviors why would the trainers need to continually use clickers? Does positive reinforcement cause behaviors to be learned or not? The evidence from SeaWorld suggests it doesn’t.
Robert Vaughan: More evidence that every trainer needs a full toolbox of techniques ... one single method for every dog, is foolish.
Johnny Fleckenstein: Also, you quoted Bob Bailey twice, making him seem like he is “on your side,” but you didn’t mention that in the same breath he said “I do think that some time, down the road, most trainers will learn that the clicker is the most powerful single tool they have, and they will quit beating it to death and learn to exploit it to its highest potential.”
LCK: Thanks for the Bailey quote, Johnny. I wonder, though, what Bailey means by “beating it to death...”
Johnny Fleckenstein: I would guess he meant clicking too much, or becoming too reliant on it, but he still doesn’t think it’s bad or dangerous. (“I am not a fan of the “ever-clicking” approach to training. The proper application of the clicker is that akin to using a scalpel to make fine cuts. However, the increasing use of reinforcement to get behavior is good, so I guess the prevalence of sloppy “clicking” is a price paid for trainers thinking more about reinforcement rather than punishment.”)
LCK: And I disagree with Bailey when he says that “the clicker is the most powerful tool.” To my way of thinking emotional rapport is the most powerful training tool there is. And for the life of me I can’t understand how one is supposed to develop emotional rapport with dog by using a clicking sound.
And the problem is that he’s only seeing training through the lens of operant conditioning. He’s not taking the principles of Natural Dog Training into account. He sees things in a very limited, punishment vs. reward fashion. Mind you, his knowledge in this limited area of dog training is encyclopedic. But he’s working and writing from within a very narrow construct. And modern scientific research has shown repeatedly that the operant conditioning model is no longer valid.
Johnny Fleckenstein: How can you quote him to support your argument against clickers, when he is quite clearly not against using clickers?
LCK: My article is about the dangers inherent to clicker training. Bailey uses clickers, but he’s also aware of its pitfalls. I mean this pretty much sums things up: “Most pet owners seldom have need for a clicker, in my opinion; a clicker can easily get in the way of getting good behavior.” Plus, Bailey, like most advocates of operant conditioning, is apparently unaware that positive reinforcement isn’t a real process; it’s an approximation of how learning really takes place, and misses the mark by miles where it’s really important.
