Cherry-picking Science
This article was first published in Front and Finish, 2013
The Cherry-picking of “Science”
There was a point in my training career when I embraced the Cult of The Constant Cookie. I wrote quite a lot about my adventures and experiments in long-past columns of this publication. Having started with traditional pop and praise training, I chose to focus entirely on rewards in training. It was a personal challenge that taught me a ton about management, incremental skill building, and prior planning. I am glad I did it for my own education, and I still utilize generous and pointed use of positive reinforcement in my training. But I thought at the time that this approach would be difficult for a novice pet dog owner, and eventually even for me and my very willing and carefully managed and carefully trained dog, the limitations of that choice became evident. It led to missing information for the dog, which led to creative technical failures in distracting circumstances or when he felt that no treat must mean “Try something else”, which led to my personal frustration, which led to several months of angst and disappointment (for me, not for him) when it turned out that “Knock that stuff off and do what you are told” was the missing ingredient that worked. I survived, I adapted, and we were successful, but it's a lesson that motivated me to keep my toolbox open. As an instructor, I will support people who are exploring that pure positive road, but I will also point out the by-ways they have available to them when the chosen route isn't working.
Unfortunately, this limited approach to training is the most prevalent one being marketed to the general pet training world of busy people with busy families and demanding jobs, the very population least prepared to successfully implement the details of careful management, error prediction and prevention, constant preparation and perfect timing required to succeed by relying on positive reinforcement only. Class presentations are often full of emotionally-charged animal-protective hyperbole or science mini-bites that either bore people stiff, or frustrates them to the point of feeling too inadequate to even have dogs, especially if the dogs already have well-established bad habits. Instructors lead them to believe that they have no right to tell their dogs “NO! Don't do that!” for fear that their dogs will become fearful and/or aggressive. The methodology is often so complex that a child of 10 or 12 years old wouldn't recognize the vocabulary, let alone be able to accomplish the task of teaching the dog anything. And believe it or not, there are many people out there who greatly resent it when they discover that their dogs won't do ANYTHING if a cookie isn't dangling in front of them, even though they are pet owners who have the freedom to use all of the cookies available at all times. They don't want to have to live entirely in the barter system.
Proponents of the approach call it “Science-based training”, and the common tone implies that anyone who uses punishment or negative reinforcement is, at best, ignorant, even abusive. But to me, the dogma ignores a lot of behavior science, and it's time to remind people about the rest of information that is carefully suppressed or ignored.
The science of behavior modification recognizes 4 quadrants of consequences that modify behavior. FOUR! Not one, not two. FOUR! Choosing to ignore two or three of them is not a scientific choice. It is at best a philosophical choice. Go for it, for as long as it works! It is in my experience (and many other trainers' too) a choice that may work in the teaching phase for certain behaviors in controlled settings for a large percentage of dogs. But if you personally would choose failure to meet a training goal for real-life efficacy, rather than effectively use another quadrant when you have hit the limitations of your preferred one, that is not a very practical decision. That is a failure to adapt to the prevailing conditions. Given that some folks have publicly said they would rather euthanize a dog rather than use punishment to address some behavior problems, I consider it a weirdly selfish philosophical choice. The trainer won't adapt, so the dog pays the price. How humane is that?
The idea of not rewarding an “undesirable” behavior and waiting for a more desirable one to happen and reward it will work in some paradigms; but the theory ignores the fact that there are behaviors that are rewarding to the dog whether we like it or not. Standing there and letting a dog sniff the ground or the agility table and just waiting for the dog to look up and then rewarding the return of focus is fine at very initial stages of puppy training; but never doubt that the sniffing was enjoyable to the dog. He'll do it again. I've even known dogs who learned to deliberately sniff the ground and then look at their owners because the sequence of events always garnered cookies. In performance worlds, at some point the dogs need to know that when they've been told to heel or told to sit on the table, choosing to sniff rather than comply is wrong. In the real world, ignoring a sit command and choosing to jump up on people is wrong. Those are OUR arbitrary choices, but we need to embrace them, and teach them, and be steadfast with our choices. Adjusting a dog's personal comfort level while he's doing something “wrong” is an effective way to modify that behavior when the intrinsic value of the bad behavior outweighs the dog's perceived value of the behavior we really want or the reward we have to offer. Sometimes for some dogs, deer poop or the smell of female dogs is way more interesting than any food we have to offer. But when diving for those scents rather than doing recalls becomes uncomfortable personally, smart dogs will stop doing that diving, and do their recalls when told, and remain comfortable...and earn praise, and admiration, and even some of those treats which suddenly are valuable again.
An oft-repeated axiom I read and hear a lot is “Punishment will cause aggression.” Hmmm...So now wait a minute. I am pretty sure that the scientific definition of punishment is an event that the subject considers unpleasant that, when associated with a particular behavior, stops and/or reduces the frequency of that behavior. I am willing to agree that abuse can lead to aggression. What is abuse? Abuse is when punitive action either continues past the cessation of the “bad” behavior and becomes an outlet for human emotion rather than a behavior modification tool; or when the physical force of the action is so over-done that the subject is emotionally and physically in a state of fear or pain that overshadows the information and association with behavior intended by the trainer. Applying any punishment without also showing the dog how to avoid it is abusive. But it's unwise to define specific actions or tools as abusive. I've had dogs who needed nothing more than a verbal “NO!” to stop any naughty behavior; but I've had a couple who were not impressed by anything short of a big pop on a long line when they had the choice to come to me for a cookie or chase a squirrel. Prey drive won out over food drive. Personal discomfort changed the balance. Perception of the consequences taught the dogs to control their personal comfort by choosing to comply with my commands when presented with the same choice again. Resulting reliable recalls made all of us happy and safe for many years.
“The only correction you need to do is not reward bad behavior.” I'm good with this simple statement up to a point in the learning process; but in our sport, we eventually have to deal with the fact that our dogs cannot tell the difference between a delayed reward for a long sequence of correct behaviors, and not being rewarded for a mistake. To the dog, it looks the same: Cookie absence. For dogs who have been taught that the presence or absence of cookies is the way to gauge their success or failure, what we mean to be reward delay or duration extension is very stressful. This stress leads to superstitious fidgeting, vocalizing, and the extremes of active creative behavior throwing, or lack of effort. That is why many dogs who have been trained very nicely indeed are very worried in trial rings. No treats, no toys, that must mean that every thing is wrong. That is science in the dog's brain. At least for performance dogs, they need to know that silence is golden, and that we can be relied upon to clearly identify an error and show them what to do instead. I think pet dogs benefit from the same courtesy. Additionally, we have to make sure that the motivators we use are clarified as rewards potentially available after cooperation, and not bribes that must be present as a cause of behaviors. This is problematic for both pet and performance training. In my experience, the transition is made easier by teaching the dog to respond to physical guidance while still using motivators blatantly. When the motivators are shifted to rewards, the physical guidance reminds the dog what he should do in response to the command or signal. Controlling his exposure to the physical guidance by cooperating with the commands negatively reinforces his response, which is a consistent reward. (And yes, I know that pure shaping with rewards rather than lures would avoid that problem. I get it; but as an instructor, I deal with people who don't have that many hours in the day to wait for desirable behaviors to happen. I need methods that allow my clients to accomplish what they need with clear goals, and quickly enough to avoid human frustration with canine experimentation. So do you. If you don't believe my words, you should post spies outside your beginning pet classes and hear the comments people make to each other when the instructor isn't there. I've had that opportunity in my travels, and it's an enlightening reality.)
“Aggression begets aggression, so you should never physically punish your dog.” That's another bit of propaganda that I find particularly disturbing when it's being recited to me by people with a young dog who is barking and lunging at the end of a leash, and then being fed the moment he stops to take a breath on the grounds of rewarding that instant of quietness, when in reality the dog was just inhaling and get ready to continue the action. I agree that proactively being ready to reward a dog for a good trained behavior, like sitting and looking at his handler instead of the lunging and barking, is the approach of choice, preferably before the dog performs that other sequence. But on the occasions that management fails, the bad behaviors need to be made unpleasant to the dog. The dog is the first aggressor. The trainer needs to make the outcome of that choice unpleasant enough to convince the dog to reconsider that option the next time. And I believe the dog needs to be presented with the choice again to finish the conversation. Sometimes he needs to be corrected again, and again, and again, so that he figures out that yes, every time he snarks, growls and lunges, he flies through the air or he gets a snoot full of lemon juice or mint spray. If he's been controlling people and dogs in his space with those behaviors for a while, it's going to take repetition to convince him that it's not working for him so well this time. When he accepts that reality, he'll make a better choice. That choice deserves praise, but the dog will still be in a conflicted state of mind that could be shifted further into a cooperative zone by requiring the dog to obey some simple commands. Active cooperation deserves admiration and rewards, but appreciate that the dog's discovery that not doing those bad behaviors prevented flight or spraying also was rewarding. Controlling one's personal comfort level is very reinforcing.
When the dog is making better choices, I agree that the giving the dog high value rewards is extremely useful for convincing a dog that it's really not bad at all to be close to a trigger. Desensitization and counter conditioning is absolutely necessary for many dogs, and in an ideal world it would have happened early and prevented the bad stuff. But when bad stuff is already happening, these techniques need to be done carefully to avoid accidentally rewarding aggressive action sequences or a volatile state of mind. I have seen dogs who showed excellent manners and self control in some other human's hands, but when with their owners would revert to the aggressive displays because the actions led to treat production. The owners would fail to direct and reinforce non-aggressive behavior, so the aggression became the way the dogs earned rewards. I realize that this is an example of an error of execution; but it shows how easily a slight misapplication of theory can backfire when the ONLY tool you've been allowed to consider is the food reward, and you haven't been given a way to stop a really bad behavior to make way for something that actually deserves a reward.
Citing incomplete examples of any technique to illustrate a chosen point certainly could happen for any argument, but it's prevalent in this sort of debate. Pure positive proponents on blogs and various social networks have posted video clips pulled from a much longer sequence of events that show dogs with big problem behaviors acting out with greater strength in response to corrections, or alternatively showing dogs doing nothing at all on the next try, and say that there is the evidence that the corrections won't work, that punishment caused aggression, or punishment caused shut down. Well, this is a normal aspect of the behavior modification process. This happens when anyone, human or animal, has had success with a behavior sequence for some time, and when that behavior is blocked or punished, at first the subject is very resistant to abandoning the pattern, and will in fact try the same behavior pattern with more intensity fueled by frustration. The second part, the stillness that actually is NOT “shut down”, is the first moment of adaptation and problem solving. When it finally becomes evident that a former behavior pattern isn't going to work anymore, it takes a moment to decide what to do with that information. The first crucial adaptation is to stop doing the thing that isn't working anymore and accept that change is necessary. That is a very hard decision for some dogs in certain circumstances, and the difficulty may temporarily block his ability to actively perform any preferred alternative. But a patient handler waits, restrains the dog and blocks more bad behavior if needed, and lets that emotional state cycle and dissipate, and when the dog's front brain is working and looking for an alternative, that is the moment to offer guidance. When learning a new approach to a familiar situation, caution is part of controlling old impulses and making better choices that are unfamiliar. It gets easier with time, with appropriate rewards for improvement, and repetition to build confidence and fluency; but it's a process, and that process is often ignored or deliberately concealed by those who preach a single approach to training. If the author or speaker using these chosen clips don't also show the final outcome of the training session that they've selected a dramatic clip from, then they are withholding data. That is not good complete science, and it is very misleading and deceptive to novice trainers.
ALL training is based on science, but science is about a full appreciation of realities. Cherry-picking only the facets of reality that you find attractive is bad science. My experience is that the more you know about training and dogs, the less you have to use punishment because you're better at building good habits with management, skill training and social conditioning. But in the real world where the conditions cannot be under your direct management at all times, this single-minded approach leaves less experienced dog owners ill-prepared to react effectively when things go wrong. So logic says that as instructors, it would be best to prepare our students for those worst-case scenarios, and educate them about those other three behavior modification quadrants so that they can be used correctly, fairly and effectively. Frustration from lack of information about how to react to bad behavior is a factor that leads many folks to give up on their dogs and surrender them to shelters. If we really care about the dogs, we should help people do what it takes to create the dogs they need by giving them a full and effective tool box.