Punish or Perish: The Bias that is Destroying Millions of Client Animals
By: Gary Wilkes, DVM
This year, millions of dogs will die because of the absence of proper care. Their numbers exceed all the animals treated in all the veterinary hospitals across the country. Their common failing is behavioral, not medical or nutritional. This behavioral train-wreck is composed of several innocuous but highly lethal behaviors: jumping on people, darting out the front door, destroying property, tugging on leash and biting. Each of these behaviors can be stopped through the use of operant conditioning – but not if your only tool is positive reinforcement. Positive reinforcement cannot create inhibitions that will prevent an animal from offering normally occurring behaviors. Only positive punishment, the presentation of a stimulus that causes a behavior to weaken or cease, is capable of stopping a behavior, cold. If you wish to slow down or stop the slaughter, you must be able to skillfully punish a behavior. That presents a bigger problem. Virtually every academic institution and many professional psychological associations tacitly endorse and enforce a bias against the study or practice of positive punishment. This bias is integral to the version of operant conditioning created by B.F. Skinner and now universally used as the template for operant control. This quote, from the Association for Behavior Analysis International, a scientific organization founded by Skinner, acknowledges this bias.
“Throughout his career, Skinner opposed the use of all forms of punishment; he advocated positive ways of changing behavior.”
The obvious question is why a scientist would “prefer” one behavioral effect over another, regardless of context? Physicists don’t prefer momentum over inertia, biologists don’t prefer plankton over fungi and veterinarians do not prefer medicine to surgery. Without a context, preferences are meaningless. By definition, punishment weakens or stops behavior and reinforcement increases or strengthens behavior. These are Skinner’s own terms for these roughly symmetrical behavioral effects. If your goal is to stop a behavior, why would you use positive (reinforcement) methods? If your goal is to increase a behavior, why would you use punishment? This pro-positive bias is plainly illogical but worse, completely unscientific. It is also the most widely held belief about behavior modification.
To see this reflected in modern veterinary circles, this is an excerpt from the position statement of the American Veterinary Society of Animal Behavior (AVSAB)
“AVSAB’s position is that punishment should not be used as a first-line or early use treatment for behavior problems. This is due to the potential adverse effects which include bat are not limited to: inhibition of learning, increased fear-related and aggressive behaviors and injury to animals and people interacting with animals.”
This statement implies that behaviors that need to be stopped should not be exposed to the behavioral effect most likely to stop them. The harmful side-effect of jumping on children, destroying property and biting people is almost always death. To shun a tool that may prevent a premature death because it might impede learning is literally ludicrous. You cannot teach a dead dog. Additionally, this statement implies that punishment, an effect that decreases a behavior, somehow increases aggression. The vast evidence is to the contrary. If Cindy Mears, a teenage kennel worker, knees a 90 pound Chesapeake in the chest, what does the scientific literature say about her application of punishment triggering “rebound aggression”? The science should confirm what actually happened. As the dog hit the ground, he quickly scrambled to his feet, darted toward Cindy and sat, wagging his tail. He was “sucking up and getting straight.” I have seen that scenario play out thousands of times. Is the knee to the chest an application of punishment? Yes. Does it result in aggression? No. Dogs that do respond to this form of punishment with aggression are anomalies. If the scientific literature somehow demonstrates that Cindy Mears was bitten by the Chesapeake the scientific literature is either incomplete or incorrect. If a high-school student can apply punishment to stop an unacceptable behavior and not trigger aggression, why are learned scientists unable to do the same? Why is Cindy’s experience not part of the body of knowledge used to formulate acceptable training protocols? (As a side note, the Chesapeake was adopted the next morning. The family selected him because of all the dogs in the kennel, he was the only one that didn’t try to jump up on them.)
Another common belief stemming from the Skinnierian ideology is that punishment should only be used after extensive positive reinforcement protocols have failed. Again from AVSAB…
“Punishment should only be used when the above (positive reinforcement for desired behavior, removing reinforcers for inappropriate behaviors, changing the emotional state and environment.) approach has failed despite an adequate effort as part of a larger training or behavior modification program that incorporates reinforcement of appropriate behaviors and works to change the underlying cause of the problem behavior.”
Again, this implies that the two behavioral effects have the same influence on behavior. If a dog is in danger of euthanasia because it bites people, we must use positive reinforcement to correct the problem. After positive reinforcement fails, a process that is not actually designed to stop behaviors, we are to hesitantly use the tool that is most likely to stop the behavior. This makes punishment a linear extension of failed positive reinforcement procedures. That is the first speck of logic in this “positive first” argument. Apparently, punishment is more powerful and can do things that positive reinforcement cannot. If that is true, it leads to a host of questions. If punishment has properties that differ from positive reinforcement, why would either be universally recommended or shunned? Who decided on the specific sequence of reinforcement before punishment? Who decided that only one effect should be used? What happens when you reverse the sequence and use punishment to inhibit the behavior followed by positive reinforcement for acceptable behavior? Why wouldn’t we attempt to match the correct protocol to the specific behavioral problem? Why don’t they know which behaviors are most likely to benefit from reinforcement and which are most likely to benefit from punishment? No answer.
One would think that research into these questions would be balanced and provide practical answers. One would be wrong. At the ABAI international conference in 2009, the ratio or presentations regarding some aspect of positive reinforcement to punishment was 750:1. Major learning institutions currently provide no courses in the use of punishment to change behavior. The obvious question is, if no universities teach the use of punishment, what is the source of behaviorists’ knowledge about the pitfalls of punishment? What credentials do they provide that would indicate their competence to dictate the tools of behavior modification? The answer, regrettably, is none. The reality is that their opinions are based on reading reports from other people and not from actual experience or direct observation. If they actually wished to discover how punishment works in the real world, they need only look at animals in their homes. Dogs that are kneed in the chest stop jumping up on people. Dogs that live with electric containment systems don’t leave the property. Dogs that run into sliding glass doors stop running into sliding glass doors. Aversive control is a part of life that all animals use to learn what to avoid. The suggestion that this process is unneeded suggests that we have no need to inhibit behaviors. On the contrary, the absence of inhibitions is the deadliest state of mind.
While millions of carcasses are hauled to land-fills, major institutions decry the behavioral effect that would save their lives — positive punishment. If you happen to decide that punishment is a valid tool for behavior modification you are faced with the complete absence of academic instruction regarding this tool. In the mean time, the killing goes on and leaves you with the knowledge that an ideology of “nice” stands firmly in the way of resolving the problem.
Originally published at DVM360.com