If a dog is punished for growling, he will learn to suppress such warnings and go straight for a bite.
Or so I’ve been told innumerable times, most recently in a promotional blurb distributed by Whole Dog Journal. The advertisement was for a new Ebook (just $7.95) that among other things, promises to explain
How punishing a growl will lead to biting without warning
Every trainer is familiar with this caveat. It’s promoted by experts, recited like gospel, and has become a staple argument against the use of positive punishment in dog training.
But where did it come from, and what proof exists to back it up?
The Prevailing Wisdom
An online search yields no shortage of confident exhortations, but few clues as to the origin of this claim. Here is a sampling (emphases mine):
Never Punish a Growling Dog. Many dog owners get understandably upset when a dog growls. Their first reaction is often to suppress the growling by scolding or punishing the dog. This is never a good idea. By teaching your dog that growling is not an acceptable behavior, you are taking away his ability to warn you that he may bite. Often we hear stories of dogs who bite with no warning. In many cases, this is because the owners trained the dog not to give a warning growl first. –About.com
Never punish your dog for growling. This may seem counter-intuitive and may even go against the advice of your dog trainer or dog trainers you have seen on TV. If your dog growls at your child he is sending a clear warning that he is very uncomfortable with the actions or proximity of the child. Be grateful that your dog chose to warn with a growl rather than going straight to a bite. If you punish the growling, you may inhibit the warning growl the next time and the dog may bite without growling first. Punishment or scolding will not make the dog feel better about the child, in fact he may even feel more anxious and be even more likely to bite in the future, especially if you are not there to control the situation. Dog trainer Yamei Ross says, “Punishing a dog for growling is like taking the batteries out of your smoke detector. You don’t want to hear the noise, but the danger is still there.” –doggonesafe.com
Indeed, by punishing a growling dog, owners are not only depriving a dog from a very important warning tool, but are also proving to the dog that the owner is unreliable and unpredictable, significantly increasing the level of stress in the dog. More stress does not solve the problem, and next time the dog encounters the same situation he may appear to be OK with the situation when he is not. Eventually, a bite will come and this time there will be no advance warning. It will come straight and mean to the owner’s eyes –Ask A Behaviorist
Surveying the above, one might assume science had demonstrated this conclusively. After all, how else would such an oddly specific idea have become the prevailing wisdom?
More to the point: Why, in fifteen years of training dogs professionally, have I never seen this happen?
Maybe I’m an outlier, but regardless, something doesn’t add up. Right or wrong, dogs are punished for growling routinely. I’ve done it myself. Dogs do it to each other. And pet owners do it a lot. If the typical result of such temerity were in fact a dog that bit without warning, then why is the world not overrun with such dogs?
I began asking around among seasoned trainers and behavior consultants. Where did this pearl come from, and did they think it fake or genuine?
Responses ran the gamut, but the mean seemed to lie somewhere between moderate skepticism and luke warm endorsement. Some admitted to sharing my disbelief; some defended the claim’s veracity; and several offered up competing interpretations of how aggression sometimes escalates in situations where punishment has been applied.
Among the theory’s higher profile advocates is Pamela Dennison, who not only boasts first hand experience with such a dog, but actually took the trouble to write a book about him. That dog is Shadow, an adopted human-aggressive border collie and the subject of Bringing Light To Shadow: A Dog Trainer’s Diary, published in 2005.
Here is the most salient excerpt, highlighted at the start of chapter 5:
A growl is just a warning–nothing more. It is obvious to me that Shadow had been punished for growling by his previous owner. It is important not to punish the growl out of a dog because if you do, you end up with a flash biter–a dog that bites with no warning. A growling dog doesn’t want to bite–that’s why he is growling. Punishing the growl out of the dog is like telling a police officer that he doesn’t have to say “Stop or I’ll shoot.” A growl is a wake-up call to YOU, to show you that you need to work on desensitizing your dog to whatever it is that he growled at. The important part is not to take it personally, a “How dare you growl at me!” kind of attitude.
I have to wonder what made it so obvious to Dennison that Shadow had been punished for growling, or that this specifically had generated his flash biting, apart from her understanding that such a history could explain the behavior she was witnessing. To my mind, it seems the cause of Shadow’s biting was largely inferred from the same theory his case is enlisted to support.
Many of the examples raised in defense of the theory shared this cyclical format. Relatively few included both a before and after component. Those that did mainly showed only a loose correlation between the punishment and biting, and hardly any were clearly distinguishable from cases involving more straightforward escalation from growl to bite.
As for a scientific basis, no one pointed to any beyond suggesting the claim had been extrapolated from scientific principles, probably by Ian Dunbar.
I dug a bit further and found more references to Dunbar, so I contacted him.
He recalled introducing the idea in the late 80′s or early 90′s, and forwarded me a number of his articles from that period. I found the below (emphasis mine) within “Provocative Behavior,” based on a behavior column from the June 1989 issue of the AKC Gazette.
The ‘Solution’ Becomes The Problem
Limiting treatment to punishing the dog for growling generally makes matters worse. The dog is growling because it feels uptight in specific situations, for example, when approached by a stranger. If the dog is punished, it now has two reasons to be uptight: 1. the initial reason – lack of confidence, and 2. the prospect of correction, or punishment, by the owner/handler, which further destroys the dog’s confidence.
In many cases, punishment may cause the growling to increase in frequency – a paradox of learning. A vicious circle develops, whereby the more the dog is punished, the more it has reason to growl and hence, the more it is punished etc. The ‘attack’ by the handler (the dog’s only immediate ally) tends to make the dog especially nervous and ‘spooky’. It must be extremely unsettling for the dog to have its best friend suddenly turn against him in times of need. At first the dog can establish neither rhyme nor reason for the handler’s outbursts. Soon, however, the dog learns, the handler’s uneasiness is contingent upon the specific situation, e.g., approach by a stranger. As such, the stranger’s approach now becomes a cue, which forwarns the dog, its owner is about to get upset and punish the dog. Most dogs do not like it when their owners become anxious, apprehensive, or agitated and so, the dog now has a third reason to growl – to keep the stranger at bay, in order to prevent the owner from becomming uneasy.
For cases wherein punishment successfully inhibits growling. This is disasterous. Firstly, punishment must be extremely severe to stop a worried dog from growling. Extreme punishments in stressful situations are inhumane. Secondly, although the dog still feels uptight, it no longer gives warning. The dog still does not like strangers and it wants to growl, but dare not. This is akin to a smoke alarm with no batteries, or a time-bomb that does not tick. The dog’s temperament is still extremely unstable but on the surface, all appears to be well.
Whether one agrees or disagrees with the above, it’s worth noting that Dunbar’s original hypothesis is more nuanced than many subsequent iterations, including those listed at the start of this post. Dunbar draws a clean line between punishing a dog for growling and punishing the growl itself (in the operant conditioning sense), suggesting the average effort to deter growling via punishment will not be successful at suppressing the behavior over time.
Pat Miller, editor of Whole Dog Journal, is also careful to maintain this distinction, as in the introductory Ebook blurb as well as the below from a frequently cited 2005 article.
Worst of all, and most significantly, if you succeed in suppressing the warning signs, you end up with a dog who bites without warning. He has learned that it’s not safe to warn, so he doesn’t. –Pat Miller, “Understanding Why Your Dog Growls: Why you should never punish a dog for growling”, 2005 WDJ
Yet the prospect of generating a stealth biter is continually promoted as a substantive risk, despite its depending on an unlikely coincidence of technical aptitude and poor judgement, whereby the punishment is skillfully applied but the overall strategy is woefully incompetent.
I put my doubts directly to Dunbar, who offered the following (emphasis mine):
Does punishing a dog for growling create a dog that bites without warning? Well, we are talking about two things here:
1. Inhibiting the dog’s growling 2. Provoking the dog to bite
Non-verbal and especially, aversive feedback often cause a bunch of additional problems. However, by its very definition, aversive punishment would absolutely inhibit the dog’s growling; if the frequency of growling is not reduced, then the aversive stimulus cannot be classified as a punishment! The most commonly used definition of punishment is “a stimulus that causes the immediately preceding behavior to decrease in frequency and be less likely to occur in the future.” And so, if the aversive stimulus were used effectively as a punishment, the dog would be less likely to growl and IF the underlying cause for the dog’s unease has not been resolved, the dog would be just as likely to bite, i.e., the dog may bite without warning (growling beforehand). Furthermore, should the dog associate the aversive punishment with a person, the dog may be more likely to bite without growling beforehand. (This may happen when a dog associates the punishment with the punisher, or when a dog is aversively punished in the presence of children for example.)
According to Dunbar, punishing a dog for growling may result in a bite without warning under certain conditions. Those would seem to include the punishment being relatively harsh, the punishment being highly successful, and the absence of productive work (such as a program of desensitization and counterconditioning) toward building confidence, tolerance, or positive associations.
This might explain why I’ve never experienced the phenomenon first hand, despite punishing dogs for growling myself a fair number of times. I can’t imagine using punishment as an isolated strategy, without training alternate behaviors or conditioning a better emotional response to triggers.
But I suspect the more fundamental reason I’ve not witnessed a dog’s warning signals being punished out, is that this is not how aggression mainly works.
Lost in Translation
Before looking at how aggression does escalate, let me briefly address the denial of growling as aggression. I’ve noticed a trend toward framing the growl as a highly desirable behavior, to the point of assigning it a near sacred status. Within this paradigm, the growl is acknowledged as an essentially peaceful alternative to an actual bite, and as such must be left unmolested (if not honored with rounds of applause).
Personally, I take issue with the notion that all growling should be interpreted as an excruciatingly polite request made under duress (as in the excerpt below).
A bite is at the far end of a long line of behaviors a dog uses to communicate displeasure or discomfort. To stop another dog, human, or other animal from doing what he perceives to be an inappropriate or threatening behavior, the dog often starts with body tension, hard eye contact, a freeze, pulling forward of the commissure (corners of the lips). These “please stop!” behaviors may escalate to include a growl, snarl (showing teeth), offensive barking, an air-snap (not making contact), and finally, an actual bite. The dog who does any or all of these things is saying, “Please don’t make me hurt you!” -Pat Miller, ”Teaching Bite Inhibition”, June 2010 WDJ
Undoubtedly, many dogs are pleading to avoid conflict with their growl. But I suspect others are saying something closer to “Bitch, don’t even think about taking my bone!” Bottom line, there’s a difference between the headspace of a dog that growls at his owner for attempting to move him off the couch, and one that growls at his owner out of fear of a beating.
Why does that matter? Because much of the weight of Dunbar’s theory seems to rest on one’s acceptance of the internal narrative assigned the growling dog. Give the hero a different script, and the ending no longer makes sense. Is the dog that growls and lunges at every dog he sees while pulling his owner down the street in fact repeatedly imploring, “Please stop, don’t make me hurt you,” or shouting “Get off my block or I’ll go medieval on your ass!”? I can’t answer that question and neither can Dunbar. But I can recognize one dog’s communication as potentially distinct from another’s, even if they both happen to include what we humans indiscriminately call growling. And I would argue that while a growl can be a laudable expression of restraint, it can also be something very different, something more genuinely offensive.
Shutting Off the Tap
In batting around these questions with trainers who work aggression cases regularly and hands on, some common themes emerged.
First, the implication that aggressive behaviors exist within discreet modules, such that one may be extracted while the rest remain functionally intact, doesn’t sit well. These behaviors are part of a natural sequence involving rising arousal levels. They may be let to proceed, may be accelerated, or may suffer interruption, but they cannot easily be rearranged or reassembled with parts missing.
As trainer Michael Shikashio, CDBC, describes the progression from growl to bite,
My thought is that it is more of a fluid escalation, one where you can “shut the water off” before it spills over the top of the glass.
I think most trainers would agree with this characterization, and in fact acknowledge that interrupting the “flow” with a well-timed punisher has the real potential to halt aggression in its tracks. In such cases, the punishment may be very mild, yet succeed at suppressing the arousal. Another trainer referred to this as response blocking.
Whatever one calls it, it’s a phenomenon with which trainers specializing in aggression are universally familiar. Yet the prevailing wisdom suggests a world in which it’s not even a remote possibility. In this world, punishment either doesn’t work at all (and possibly adds fuel to the fire), or it acts like a perfect surgical strike, annihilating individual behaviors with laser precision while leaving the rest of the picture untouched (and opening the door to biting without warning). And of course, the implication is that one can’t reliably predict which outcome will occur.
Meanwhile, the real capacity to suppress aggression via positive punishment without disastrous fallout is rarely acknowledged in public discussions.
Why is this?
Because a threat that is both profound and unavoidable, striking often and at random, is a powerful deterrent. And that’s ultimately the point, in my estimation, far ahead of enlightening anyone about anything.
An honest discussion of punishing aggression would reflect actual knowledge and actual probabilities. The prevailing wisdom regarding punishing warning signals, based on untested hypothesis and promoted by rumor, reflects neither.
Am I saying that punishing dogs for growling is typically a good idea? Not by a long shot. I may never have witnessed this strategy result in a stealth biter, but I’ve seen plenty of other unpleasantness.
Am I saying it’s impossible to generate such a dog via this formula? Again, no. But research and experience both suggest it is the least probable or even relevant consequence of punishing for posturing, growling, etc. Far more likely to get an immediate escalation of the conflict, a redirection of aggression toward the handler, or an interruption of events accompanied by a suppression of arousal. The fact that all of the above might also happen to include less growling is almost beside the point.
The claim that punishing a dog for growling may cause him to bite without warning is made so frequently, and with such authority, that it should hold up to a little scrutiny, in my opinion. Yet a casual inspection reveals little foundation and lots of loose ends.
At minimum, it’s an exaggeration of how typical this phenomenon is, and an overstatement of our understanding of how aggression works. At worst, it’s a questionable idea whose main attraction is its utility as a scare tactic.
Aggression is complex, both as a suite of behaviors and as a problem to solve. If it weren’t, every trainer with a basic understanding of theory would be prepared to deal with it swiftly. Meanwhile, our understanding of it, both as individuals and as a profession, is neither crisp nor uniform. It is dynamic and evolving, hopefully in the direction of reality and away from myth.
My thanks to the many accomplished trainers who shared their opinions with me on this topic, and to Dr. Dunbar in particular for responding to my questions.
© Ruth Crisler and Spot Check, 2013.