Aggression and Operant Conditioning
By Gary Wilkes Copyright 2009
About 30 years ago I was a brand-new shelter manager of a small humane society in Oregon. Soon after my hiring I was responsible for performing and supervising euthanasia - a necessary part of the job. One morning I was attempting to give a fatal injection to an adult Lab mix. I was still learning my craft and a little hesitant with presenting the needle. As the needle kissed his skin, the dog jerked his foreleg back and tried to bite my hand. As he lunged, the syringe slipped from my fingers. By complete fluke, at the instant the dog’s teeth tried to close on my hand, the syringe was balanced on the table, pointing straight up. By trying to bite my hand, the dog inadvertently stuck himself in the lip with the needle. A behavior analyst would say that the lunging behavior was contiguous with the presentation of an aversive stimulus that likely constituted a positive punishment procedure. That is a lot of words to simply say that the dog punished himself accidentally. Regardless of how you describe it, he recoiled backward exactly as if he’d been jabbed in the lip with a 20 ga. needle - which is exactly what happened. Now I was faced with a dilemma. I was less than confident about my ability to euthanize this dog. I mustered up all my courage and put a fresh needle on the syringe. As I placed the needle along the cephalic vein, I plainly saw the dog flinch. As the needle entered the skin, I saw something else. At the instant that mirrored his attack on the last repetition he clearly pulled his head backwards. He was attempting to avoid the jab of the 20 ga. needle - a needle that was no longer in a position to jab his lip. By any analysis, the bite of the needle stopped the attempted bite on the first occurrence and created a latent inhibition against biting on the second repetition. This is a perfect example of controlling aggression with an operant conditioning procedure, specifically a contingent punishment procedure.
Myth vs. Facts
It is common in modern training and behavior philosophy to assume that aggression is a mystical phenomenon that cannot be directly confronted. Commonly, the fundamental aspects of aggression are side-stepped through a process of naming all the various triggers for aggression – dominance aggression, food aggression, predatory aggression, territorial aggression and a host of “aggression du jour” categories. This is like asking someone how to make ice-cream and they recite a list of flavors. That doesn’t answer the question. Dividing aggression into myriad subcategories merely tells us the context of the behavior. Unless treatment is fundamentally different for these various types there is no need for the classifications. One wonders the purpose of telling a client that their dog has “resource guarding issues” after the dog bites to defend a treat. It is doubtful the owner needs that information or the task of learning a fancy term. 1 In reality, there are only two forms of aggression – the kind you can stop and the kind you can’t. A behavior analyst might term this operant (controlled by its consequences) vs. non-operant (not influenced by consequences) aggression. The problem is that modern experts do not know how to stop behavior. That is because they openly vow never to use the tools that can stop a behavior and openly attack anyone who does. (A major irony of the “all postive” movement is that they hate punishment - and use it with gusto to silence anyone who wishes to challenge their assertions or use punishment to save lives.)
Current Practice and Pseudo-ethics:
Today, the primary tools for controlling aggression are positive reinforcement and the removal of positive reinforcement that may cause a behavior to become stronger or weaker, respectively. Fifty years ago, the primary means of dealing with aggression was some form of physical retaliation. Neither of these polarities offer an ideal solution to aggression, but the older perspective was in the ball park. The reason both perspectives are off the mark is in the assumption that a practitioner is limited to one tool or the other. (i.e. reinforcement vs. punishment.) In fact, no such obligation exists and no ethical or logical rules suggest that dichotomy. In veterinary medicine, a doctor may use radiation, surgery or chemicals to treat potentially lethal cancer. The vet may use them in a combination or sequence that is most likely to stop the cancer or affect an improvement. Reapplication of the various treatments may be necessary in order to save the animal’s life. Sometimes the sequence of treatment makes a difference in prognosis – such as cleaning a wound before applying a topical antibiotic. Why a behavior therapist would not use all the effective tools and combinations/sequences of tools available may be a matter of ideology, personal preference or whim, but is not a matter of ethics, logic or science. Consider the illogic of a widespread dictum - punishment can only be used as a last resort. If a cancer is known to be fixable with surgery, would a veterinarian go through every less invasive treatment first, even if those treatments had never been proven effective in controlling cancer? That’s nuts - and plainly unethical because it delays effective treatment and needlessly risks the animal’s life. Why would a professional not use the methodology most likely to be effective rather than attempting to be least invasive? How does least invasive save the dog’s life it it is also least effective? For instance, putting a topical herbal remedy on a shattered leg simply because it’s less invasive than surgery is a cruel joke. Avoiding the use of punishment to stop aggression is no less cruel. Every moment the aggression remains unstopped and uninhibited puts the animal and others in jeopardy. In this specific case, the wisdom and logic of medical ethics are curiously absent from behavioral therapy.
More Experience to Ponder:
About 25 years ago I was working with an Australian Cattle Dog named Molly and a Terrier mix named Punkie. They were both fully adult females and had a short but violent history of aggression. The younger dog, Molly, was reaching social maturity at about 18 months of age. The older dog, Punkie, was about three and fully mature. This is a common scenario for inter-female aggression. Molly, the Cattle Dog, was a very tough animal. She had received very damaging wounds from the collective fighting with no lessening of her will to attack. During one fight her jugular vein was exposed. Punkie, though technically a poodle-terrier, resembled the poodle only by the texture, length and curl of her coat. Under the skin she was a barrel-chested Jack Russell.
My then na´ve self followed the contemporary and still modern goal of using primarily positive reinforcement to solve the problem. All of the books I had read suggested that punishment isn’t effective in stopping aggression and that a myriad of negative side-effects ruled it out as a viable therapy. It was universally promoted that science confirms that positive methods are superior to punishment.2 The slogans attached to the topic were compelling: violence begets violence, punishing aggression merely punishes the symptoms without stopping the actual behavior, punishment causes rebound aggression and alienates the animal from the owner/punisher. To avoid these dire consequences I crafted a subtle, brilliant and utterly stupid solution using positive reinforcement. I taught Punkie to wag her tail, on command. Like all ideologically motivated solutions it couldn’t work in the real world.
The tail was my focus because one of the triggers for the dogs’ fighting was Punkie’s tail carriage and movement. Her tail carriage was typically terrier – straight up, high over the back and flagging – a sign of aggression in most breeds and more specifically, a sign of aggression to Molly. Cattle Dogs, by contrast, stand comfortably with their tails pointed at the ground. When a terrier has its tail over its back, it’s a normal posture. When a Cattle Dog elevates its tail, it’s a threat. Molly felt constantly threatened by Punkie.
As a first step toward fixing the problem I decided to use positive reinforcement to weaken one of the triggers. I taught Punkie to wag her tail on a horizontal plane rather than hoist it up over her back. As I envisioned the process it worked brilliantly. When the owners noticed Molly bristling over Punkie’s tail, they commanded “wag.” Punkie would dutifully drop the elevation of her tail and twitch it back and forth in a traditional wagging pattern. Boy, I thought I had a brilliant solution. In reality, it was a stupid waste of time because it did nothing to actually inhibit the aggression. Punkie’s overly-muscled rear-end made it impossible for her to maintain a wagging pattern for more than about 20 seconds. After her muscles fatigued her tail gradually elevated above her back into a perceived threat. More importantly, positive reinforcement doesn’t cause an inhibition that would carry into the future. Unlike my lab-mix who withheld the bite on the second experience with the needle, neither Punkie nor Molly were in any way inhibited from either high-tail carriage or major throat ripping in the future. My solution was missing the key ingredient because I believed “experts” who really weren’t.
We Meet the Bottom Line:
By the third session my use of positive reinforcement had lessened the aggression but not stopped it. At the end of the session I got up to leave. Punkie was about three feet from me, facing away. Molly was about three feet farther, facing Punkie. My eight years working in shelters allowed me to anticipate what was about to happen. Molly was about to rip Punkie’s face off. At that moment I was almost helpless to stop the fight. The only tool at my disposal was the knowledge that I had correctly associated the word, “No!” with a blast of water from a spray bottle. (As with most trainers who pander to positive reinforcement, I called it a “correction” rather than what it was, positive punishment.) At the instant the dogs were about to clash, my water bottle wasn’t going to be of any assistance – it was in my brief case and they were too far away for a blast of water to be effective. Fearing that doing nothing wasn’t an option, I yelled. As I yelled “No!” both dogs froze for a split second. My preconditioning with the water bottle had at least achieved that. As my mind ruled out throwing my hard-sided brief case at them, I looked down at the couch in front of me and found the solution -- I picked up a throw pillow and allowed it to live up to its name. The pillow grazed the hair on Punkie’s back and hit Molly right between the eyes with some force. She spun around and ran down the hallway, submissively wetting the whole way. She went under a bed and couldn’t be persuaded to come out for more than 30 minutes. I said to myself…”Hmmmmm.”
A New Day Dawning:
According to all the experts, what I had just done couldn’t have happened. There was no rebound aggression. The dogs didn’t perceive the throw pillow as a participant in the fight. They didn’t explode into battle as it hit them. After the fight, no aggressive overtures occurred for more than two weeks. (Prior to the “bonk” they were fighting daily.) Neither of the dogs hated or feared me after the incident. This process defied conventional scientific wisdom.
Regardless of popular rhetoric, this revelation changed my business immediately. If it happened once, it could happen again. I needed to learn the rules of how to apply this new tool effectively and safely. Soon, I integrated precisely applied positive punishment to aggression despite the virtually unanimous warnings against its use. I discovered that some forms of aggression are very sensitive to operant conditioning. Food aggression, for instance, can be stopped with an almost 100% assurance of success – if you know what you are doing. If you believe it can’t be done, you will never gain the experience to know how to stop it. After using positive punishment on more than 7500 dogs, I have only seen three cases of rebound aggression – all directed at the “bonker” rather than a human. In some cases, a single experience of properly applied punishment can stop a dangerous behavior for a lifetime. In other cases, more lengthy procedures must be used, often including an initial punishment procedure followed by positive reinforcement for correct behavior.
Conclusions and Projections
The use of punishment and its complimentary opposite negative reinforcement as components of a process to control aggression is a viable option for serious professionals. The straw-man of anti-punishment ideologues is someone who would use exclusively aversive methods. No such person exists. I have created more practical tools for positive reinforcement than anyone else in the modern dog training movement over the last 25 years. I never use exclusively “negative” or exclusively “positive” solutions. I use positive reinforcement in balance with aversive control crafted to benefit the individual animal and the specific goals of my training program. The “all positive” ideology suggests that I may only pick one polarity and that there is no way to apply gradient aversive control – it’s always presented as so powerful that is is automatically traumatic and dangerous. They will not consider that the sequence of which behavioral effect you apply can have a great deal of influence on creating effective treatment. i.e. It is often effective to use punishment first to stop an existing potentially dangerous behavior followed by positive reinforcement for acceptable behavior. What if the dog needs a working repertoire before you can stop the aggression, such as my use of positive reinforcement with Molly and Punkie? That requires the use of positive reinforcement prior to any use of aversive control. What if the best solution is to apply concurrent polar opposite effects that create one of more contingencies? i.e. Jump on children and you get bonked. Sit quietly and you get treats. The sequence that is most likely to solve the problem is the one a skilled practitioner chooses based on experience – not some handcuffed process that offers only cookie-cutter methods based on ideological beliefs. In reality, creating negative consequences for bad behavior concurrently with positive consequences for good behavior is the most powerful way to change problem behavior rapidly, humanely and effectively. Simply giving treats for an alternate behavior isn’t functional at all. No pet owner is going to spend the amount of time and diligence necessary and starve their dog to make treats trump the sound of a doorbell. No treats can stop a dog from ingesting inedible objects. No treats can prevent a dog fight. i.e. Treats and punishment can. That is because reinforcement and punishment do different things. To be effective one must understand behavior as a dynamic, many-layered phenomenon that does not respect one-dimensional perspectives, either positive or negative.
The inevitable result of inadequate treatment for this problem is injury to a human or other animals and death to the aggressive dogs. In essence, the behavior is almost always fatal. In medicine, the treatment is always evaluated in terms of the risk of not treating the ailment vs. the risk of the most likely successful treatment. In behavior, this is not so. The difficulty of discussing aversive control, rationally, is a long-standing and illogical bias against the use of aversive control within the behavior analytic community, behavioral psychology, veterinary behaviorism and now, mainstream dog training. Reports of failure or dire consequences using these tools appear to me to be the result of failed methods or improper application. As there is no course, textbook, instructor, practical training, internship or certification within academia or veterinary schools to teach the correct use of punishment, this is undoubtedly the case.3 Instead of developing better procedures, as was done in the dawn of surgery, behavior experts wish to drown the baby in the bathwater. Using proper respondent and operant methodology opens up treatment to a large percentage of aggressive dogs normally assumed to be unfixable. To quote the Romans, abusus non tollit usum – the abuse of a tool is not an argument against its proper use. While “all positive” people continue to try to dominate the conversation using a mantra that translates as “Death Before Discomfort” I am of a different opinion - I choose life.
Author’s Note: If you liked this article, you’ll probably like some of the content on my blog... clickandtreat.com/wordpress/
1 Many modern experts seem to rely on jargon to imbue gravity to their words. Telling an owner that his dog bit him as the result of “misdirected aggression” is both unnecessary and useless in solving the problem. The term merely indicates that the dog attacked the wrong target. This would be like telling someone drenched by rain that they have a hydraulic avoidance problem.
2 The biggest irony is that within the literature of behavior analysis is a definitive, never contradicted description of how to stop aggression and the modality that “competely suppressed” the behavior, including rebound aggression. If you doubt me you can go to the archives of the Journal of the Experimental Analysis of Behavior. They are on-line and free to anyone with an interest. Look up Ulrich, Wolfe & Dulaney, 1969. Punishment of Shock Induced Aggression. (Just Google the authors and you’ll find the .pdf file) The scientifically valid method of stopping aggression is electric shock. The modality used by the researchers in 1969 has been greatly improved in the years since - it’s called an electric shock collar. You probably haven’t heard that from any modern trainers or behaviorists. Most of them want shock collars outlawed and would never use one. Meaning they intentionally oppose the use of the tool that is scientifically confirmed as effective in suppressing aggression. It makes you wonder why learned Doctors hide references to valid peer-reviewed research. Last I checked, that isn’t very scientific. This is by no means the only citation to studies demonstrating the effective use of punishment to control behavior, but it’s the one most connected to this specific topic.
3 Ask yourself why an academic who has never received any training in the use of punishment would be an expert on the subject. They aren’t. Their statements are the results of simply repeating what someone else told them. If they are wrong, you will inadvertently be helping to withhold treatment or knowledge of treatment known to be effective. Known by whom? 15,000 years worth of dog trainers.