A Rebuttal

Earl1Disclaimer: I am a force-free trainer and do not use pain, intimidation, or fear to modify dogs’ behavior. The following is a rebuttal to Gary Wilkes’ article titled “Modern, Scientific, Positive Dog Training and Dirty Little Secrets.” For brevity’s sake, this post will only focus on the first six points of his article.

While full of myths that pervade current dog training culture, Wilkes’ statements simply do not hold up in the court of science.

“Let’s pretend for a second that you are a positive trainer who uses methods based on science. That is a pretty big club. It’s also a pretty handy business strategy. For one thing, it differentiates your style from “those other trainers,” who are assumed to be un-positive and unscientific. It also implies that you are more educated than your competition and have diligently studied learning theory – the science of behavior as accepted by behavior analysts, ethologists and comparative psychologists.”

While I’m sure there are trainers who view reward-based training as a viable business strategy, it should not be the primary motivation of anyone going into the field. Business strategy is different from ethics. Certainly, ethics can and should be incorporated into a trainer’s business strategy. But first and foremost, a trainer’s motivation should be to use training methods that result in efficient, effective results with as little wear and tear on a dog’s emotional and physical state as possible.

Herein lies an example of consumer protection: There are plenty of force-free trainers who lack a solid understanding in animal learning theory and the skills involved in identifying behavior problems and implementing behavioral modification plans. Buyer beware.

“In science, especially behavioral science, positive doesn’t mean nice, good or beneficial. Modern trainers use the term (and all of its derivatives using adaptations of “paw-sitive” to imply dogs + nice.)”

Wilkes is correct in stating that “positive,” when used in terms of animal behavior analysis, does not refer to a moral choice. It simply refers to the addition of a stimulus. This isn’t a training issue. This is a marketing issue, which is why I tend to use the term “force-free training,” and why others choose to use “reward-based training.” Semantics.

No animal can be trained to a high level of performance without the use of aversive control. At least, none ever have. That is because if there are no “negative” consequences for errors there is no incentive for the animal or human to improve their performance. 

While certainly a myth that pervades current dog training culture, this statement simply does not hold up in the court of science.  And, as Dr. Susan G. Friedman so eloquently states, “Behavior doesn’t flow like a fountain; behavior is a tool to produce consequences.” In other words, dogs are consequence-producing machines. They don’t produce a behavior for no reason; they produce a behavior to get something, whether it be food, attention, social access, or in cases of fear and aggression, to increase distance from a scary stimulus.

Wilkes’ statement also flies in contradiction to The Matching Law, which states that an animal, when faced with two possible behavior options, will choose the behavior that has the strongest history of reinforcement. This isn’t just opinion. It’s a law.

Writes Friedman, “It is the nature of animal behavior to change what they do, based on the outcomes of doing it. In this way, behavior is selected by consequences: Behaviors that produce desired outcomes are repeated; behaviors that produce aversive consequences are modified or suppressed. Behavior is a purposive tool, part of every animal’s biological endowment, used to affect the environment. Even bacteria change what they do based on the consequences of doing it.”

In terms of consequences, force-free dog trainers have two highly effective options. The trainer can reward the behavior, upping the odds the dog will do the behavior again. If a dog is doing a behavior the trainer wishes to decrease, the trainer can remove the consequence the dog was trying to achieve via “negative punishment.”

Let’s discuss punishment for a moment. By definition, punishment decreases the occurrence of a behavior. When a trainer uses “negative punishment,”  that trainer removes a reinforcer (like food, access to humans, access to play and other dogs) immediately after an unwanted behavior occurs. The dog realizes that the unwanted behavior results in the removal of highly coveted things, rendering the behavior irrelevant.

Dogs do what works. If a dog consistently realizes that doing one behavior gets him the exact opposite of what he was trying to achieve, he will do the behavior less. That dog doesn’t have to endure pain, fear or intimidation to get this result. He simply has to understand what behaviors produce which consequences. And the consequences certainly don’t have to inflict fear to produce reliable behavior. In fact, the more fear involved, the less reliable a dog’s behavior is likely to be, as I will discuss below.

Also, there are a plethora of examples of dogs trained to high levels of precision without the use of aversives. Confirmation bias and cherry picking are dangerous things for anyone, particularly those who espouse the virtues of science.

EG: The next bomb detection dog you see was trained with some element of punishment for failure or negative reinforcement to compel the dog to do something it didn’t want to do. You better hope it wasn’t trained “all positive.” Why? Because the dog never had anything to fear if it screwed up. If the word fear bothers you and you instantly have “negative” feelings about it, I suggest you consider how much of your life is governed by fear – such as not stepping in front of a bus. The assumption that fear is a bad thing or that it is always traumatic is simply another rhetorical distortion of modern trainers and behaviorists. If your life is being protected by someone who has no fear of missing an explosive device you are not really protected.

This example is rife with flaws. First and foremost, Wilkes shows a fundamental misunderstanding of fear in relation to animal behavior. Competent trainers do not use operant conditioning on a fearful dog. Instead, they use classical (Pavlovian) conditioning to address a dog’s emotions. Not until the dog shows a positive emotional response to a stimulus in the environment does the trainer transition to operant procedures. (See my previous articles here and here for more information).

Second, instilling fear in a dog doesn’t produce reliable behavior. As Jean Donaldson teaches her students, when dogs are afraid, nothing else matters. Fear overrides a dog’s ability to respond to operant contingencies. Fearful dogs may not want food. They could care less about their owners telling them to sit, go down or stay. They may start to display aggressive behaviors that aren’t present at other times. They are unable to cope and paralyzed by one thing: fear.

“An animal’s functional behavior is made ineffective whenever we ignore its fears, force it to go where it resists going, and coerce it to do things against its will. Even locking a dog in its crate with a fear- eliciting toy, based on the rationale that “he’ll get used to it,” renders the dog unnecessarily powerless to escape. When a lack of control becomes a life-style, it may result in the aberrant behaviors dogs do such as excessive barking, repetitive licking, and phobic behavior,” Friedman explains.

Thirdly, Wilkes’ analogy is a flawed one. A person’s choice not to step in front of a bus is not comparable to a bomb detection dog not “screwing up” because of a potential explosion. It simply goes against the principles of animal learning theory and dips dangerously into anthropomorphizing. A bomb-sniffing dog doesn’t do a “good job” because he is inherently fearful that if he won’t, a bomb might explode. A bomb-sniffing dog reliably detects and alerts handlers to explosives due to a history of learned relationships between antecedents, behaviors and consequences.

To quote Dr. Murray Sidman in his book Coercion & Its Fallout, “An overworked and incorrect bit of folk wisdom pronounces the carrot to be of no avail unless backed up by the stick. But the carrot can do the job all by itself.”

“Behaviorists and modern, positive trainers offer methods that please them and forget that dog owners do not have unlimited funds, time and patience to live forever with low-expectation non-solutions. There is no effort to offer knowledge aimed at the “end user” in the world of modern dog training or behaviorism.There is no thought to the damage that is done by offering bogus, never proven advice. There is only the quest to rein supreme in the public eye while ignoring the dog and owner – the very reason to offer behavior services in the first place. The word “fraud” comes to mind.”

This isn’t a force-free vs force-based training issue, this is simply a consumer protection issue. There are plenty of incompetent trainers on both sides of the aversives vs. non-aversives argument. Wilkes is correct in stating dog owners must be wary of who they choose to train their dog, but it’s a problem across the board, regardless of a trainers’ philosophy.

“By definition, positive reinforcement increases behavior. It cannot stop behavior. That is done by punishment. If you preach that you do not use punishment it means that there are limitations to the services you offer. This fact is covered up by the fictional and unsupported claim that modern scientific methods can control all behaviors. That is a logical contradiction. Those methods can’t stop anything. To believe their rhetoric you must assume that there is never a reason to stop a behavior, now.”

Wilkes is confusing the phrase “positive reinforcement training” with the idea that a positive reinforcement trainer uses only reinforcement. There are many positive reinforcement trainers that use punishment (see above about removal of desired consequences to reduce a particular behavior).

Wilkes also fails to address the many nuances of animal training, including concepts like behavioral momentum, differential reinforcement of an incompatible behavior, and the pitfalls of punishment. To quote Friedman again, “The focus on replacing the function of a problem behavior with an appropriate alternative is fundamental to understanding behavior and respecting behaving organisms: If the behavior didn’t matter to the animal, it wouldn’t keep doing it.”

If we only focus on telling a dog what “not” to do, we leave a massive void. The dog knows not to jump, but what does he do instead? For dogs that have high energy and low ability to control impulsive behaviors, we need to help them fill the void. We must teach them replacement, desirable behaviors to do in place of the unwanted ones we’ve taught them not to do. And we teach them via – you guessed it – positive reinforcement of the desired behavior.

– Maureen Backman, MS is the owner of Mutt About Town dog training in San Francisco. She is also the founder of The Muzzle Up! Project and Muzzle Up! Online. To get in touch, email her at muttabouttownsf@gmail.com.