The Difference Between Dog Poop and a Bagel:

 

 

It may surprise you to know that behavior modification isn’t the same thing as “learning” in the sense used by scientists and modern dog trainers. If this sounds like splitting hairs, it’s not. It is a significant point that leads people down the garden path. I’ll give you an example.

Example 1: “Punishment does not teach an animal what to do.”

The above statement is a widely expressed belief of behavioral scientists and modern, scientific dog trainers. This would be a very simple truth if it was taken literally. Almost as simple as saying that lead doesn’t float on water. However, the people who say this add a subjective value judgment that makes the statement false. To see the error in their conclusions we need to look at what they mean by the words “teach” and “learn” and how it is influenced by a bias that comes from behavioral science. B.F. Skinner, the person most responsible for the underlying principles of “learning theory” believed that one could only have a science of behavior by examining a cumulative record of behavior over time. To accomplish this he created experiments that counted how often a rat pressed a lever or a pigeon pecked a lighted key. He picked rats and pigeons because they were small and cheap to sustain, not because they were the best animals to gain insight into behavior. The experiments were equally chosen to match his desire for data rather than some significant advantage in understanding fundamental principles of behavior. What the experimental process did was generate data that could b062014_2012_TheDifferen1.gife analyzed in the form of charts and graphs like the one on the right. You can encapsulate Skinner’s mantra as no data, no charts, no graphs, no science. This works if you are studying ways of increasing behavior. The problem started when Skinner looked at punishment. Punishment stops behavior. Now we have a serious road-block and an explanation of why behavioral scientists have a unique bias about how to view a roughly symmetrical dual phenomenon – reinforcement and punishment.

Let’s take a moment and see how the bias in favor of collecting numbers has influenced our language and our perception. To teach assumes that someone is going to learn. Therefore the individual must be in (X) state when the teaching starts and arrive at the state of (X+Y) when the teaching stops. We will call that Dog A, Dog A = (X+Y) a  dog with some addition to it’s repertoire. (The assumption in this is that the process of learning is additive rather than “different.” )This has been integral in the minds of behavioral psychologists for so long that they are incapable of understanding any other perspective. However the term “behavior modification” means a change in the existing state of an organism’s repertoire. It does not specific additive, merely different which would also imply subtractive. So, to Skinner, the following is not an example of “learning” but by any definition it is certainly behavior modification. Take a dog and select an arbitrary behavior (Behavior X) to inhibit. We are going to call this experimental animal Dog B prior to creating the inhibition and Dog C after. We can see that Dog A and Dog B are already dissimilar. One had a behavior added to its repertoire and the other is in the same state Dog A was before we added a behavior. Dog B has an existing repertoire that encompasses many behaviors both instinctive and learned. We will call it’s original repertoire (R). If we inhibit Behavior X, Dog B changes to (R-X) which we will create Dog C. This isn’t a rare thing. It is so common we have many terms for this form of behavior modification, usually represented by the prefix, “ex”. Someone can be an ex-smoker, ex-gambler, ex-drinker or ex-fighter. They once included those things in their life and not they don’t. To Skinner, the process whereby someone becomes and “ex-anything” is not learning – but it is certainly behavior modification. Example: Dog D jumps on people. The behavior is eliminated from his repertoire through a punishment procedure and he becomes Dog E – an ex-jumper. This is not considered learning even though there is a clear difference between Dog D and Dog E. We can examine 100 dogs and classify them as being with or without a specific behavior. Some are Dog D and some are Dog E. Not only is there nothing odd about this, it is a pretty important aspect of the human-dog relationship. Consider how many times someone asks, “Does your dog bite?” If dogs that bite are Dog F types and dogs that don’t bite after a punishment procedure are Dog G type we have a significant reason to know which is which. If ignoring subtractive relationships sounds goofy, here’s the paper trail that explains where this “distinction without a difference” comes from.

The Paper Trail:
Here is a quote by Dr. Murray Sidman – acolyte and colleague of B.F. Skinner in his book Coercion and Its Fallout. Sidman is often quoted by modern dog trainers as an expert on the topic of punishment.

But just punishing the animal for doing something else does not teach it to sit. At most, punishment only teaches it what not to do.”

To anyone using simple logic Sidman’s comment is bizarre. Punishing a dog is intended to stop a specific behavior. We want that (R-X) state.  It is not supposed to teach a dog to sit or do anything else. The elimination of one behavior and the creation of another are two separate processes. If the objectionable behavior is jumping up on people, I can punish the dog for jumping up as a discrete procedure. Just like touching a hot stove doesn’t require that you also learn algebra as you learn to not touch hot stoves. If I at some point I teach the dog to “sit” then “sit” is a part of the dog’s repertoire along with “don’t jump on people” and a thousand other things like knowing where and when the food bowl is full. Meaning there is no dog that jumps on people that does not already have other behaviors in its repertoire. If jumping up is blocked, the dog accepts the new rule and adapts with some other behavior or with no behavior at all.Sometimes all that is necessary to adapt is to stop doing a single behavior, now. No other modification of the repertoire is needed. That is as true in nature as it is in captivity.

The main concept is that if at any point I decide to stop the jumping behavior the process is independent of any other behavior. Once punished the dog can do something else it already knows how to do or nothing at all – because nothing at all is the purpose of punishment. If the dog pivots and sits instead of jumping up fine. If not, it doesn’t matter. It can wander away, stand still, lie down or many other behaviors it already knows. The idea that a punished behavior is somehow linked to learning another behavior is a non sequitur.  The two things are not related – until you read into it the bias of all behavior analysts. To them, increasing behavior is the only kind of behavior modification acknowledged as valuable. Subtracting something is not considered at all. In Sidman’s quote you see their solution to problems based on bias. They must teach a new behavior to prevent an existing problem – an (X+Y) solution. Even though the two processes, reinforcement and punishment,  are roughly symmetrical opposites both are valued by their ability to increase behavior. (Except that punishment cannot subtract behavior) By the definition of those who created “learning theory” this is a contradiction. Reinforcement increases behavior and punishment decreases or stops behavior.

The problem is that this perspective has trickled down to academics in related fields, dog trainers and then became “fact” to the general public. Here are some quotes that display the echoes of Skinner, Sidman and the orthodoxy of modern behavior analysis. Note: I cite these people because they make money giving advice to dog owners. They are no more immune from criticism than Dr. Phil or anyone else who publicly claims to have and pronounces “expert knowledge.”

Ian Dunbar: Veterinarian and PhD in dog ethology (his PhD is not in behavior analysis)
“Any punishment for inappropriate behavior is an obvious advertisement of insufficient, or ineffective training. You have yet to effectively teach your dog to want to do what you want it to do.”
Victoria Stilwell: Popular television trainer
“Modern behavioral science has proven that forceful handling such as physical punishment, leash yanking, or making a dog submit by rolling it on its back is psychologically damaging for the dog and has potentially dangerous consequences for owners. Instead, the most successful modern training theories suggest that reinforcing good behavior with rewards while using constructive discipline is much more successful.”
The American Veterinary Society of Animal Behavior: (An association of veterinarians, primarily board certified behaviorists.)
“AVSAB recommends that training should focus on reinforcing desired behaviors, removing the reinforcer for inappropriate behaviors, and ad¬dressing the emotional state and environmental conditions driving the undesirable behavior. This approach promotes a better understanding of the pet’s behavior and better awareness of how humans may have inadvertently contrib¬uted to the development of the undesirable behavior. Punishment should only be used when the above approach has failed despite an adequate effort as part of a larger training or behavior modification program that incorpo¬rates reinforcement of appropriate behaviors and works to change the underlying cause of the problem behavior. “
Pat Miller: Popular “Positive” Dog Trainer
“It (punishment) doesn’t teach the dog what to do.” (emphasis hers)

Dr Susan Freidman PhD behavior analyst:
“Punishment doesn’t teach learners what to do instead of the problem behavior.

Consider for a second that two learned, well-respected behavior analysts, a veterinarian with a PhD in ethology, a scientific organization of academically trained veterinary behaviorists and two popular “positive” dog trainers are unaware that the purpose of punishment is to stop a behavior from occurring. They all assume that problem behaviors are the result of not teaching something else. OK, I’ll bite. Australian Cattle Dogs were bred to bite the heels of cows. That’s why they are called ‘heelers’. WhHeelerHeelingat would one teach in advance that would make that instinctive behavior not happen? Sit? OK, let’s try it. Take a well-bred cattle dog and show it some cows. Now say “sit” and watch  as the dust kicks up in your face from a dog going lickety-split after a bunch of cows. You can’t teach a behavior that will remove this behavior from the dog’s repertoire. That is because the dog has a genetic obsession with chasing moving objects and then biting the critter’s foot to drive it. You didn’t create the heeling behavior. It was there when the animal was born in every cell of its body. It woke up one day when conditions were right. It’s not going away by teaching something else. This is true for all dogs bred for a behavioral purpose – pointers, retrievers, hunting hounds and other types of herders. This is not known to behavioral scientists.

Apparently if I do not want my cattle dog to chase my cats I must teach it to do other things so powerfully that it overrides 150 years of breeding. That begs the question, “with what”? What positive reinforcer is more powerful to a cattle dog than doing what it was bred to do? The answer is none. If you doubt this, do the one thing that learned scientists are supposed to do but behavioral scientists never do – try it and report your findings. Get a half dozen Blue Heelers and teach them to sit instead of chasing cows. Then tell me the results and what specific things you did to accomplish the improbable. Now the other shoe drops. Australian Cattle Dogs are never allowed to ‘heel’ horses with riders. That behavior is punished early in their working career for obvious reasons – drovers ride horses. Apparently savvy Australian drovers teaching their dogs to ‘not heel’ horses with riders isn’t an example of “learning”, according to Skinner, and is proof of inadequate training, according to Dunbar.

This real-life example leads to the next problem for behavioral scientists and modern trainers  – positive reinforcement cannot block a behavior from occurring specifically because it does not subtract anything from the dog’s repertoire. This is supposed to be the idea – never remove something, only add behaviors. That doesn’t work. No matter how scientifically I work to reinforce the absence of a behavior, it’s still there. It doesn’t go away simply because some other behavior is reinforced. If that were true, you would forget old behaviors as you learned new ones. EG: No one would be able to play multiple musical instruments because as you learn piano you would forget how to play violin.

The important thing to glean here is that we live in a sea of behavior. Understanding fundamental principles is as simple as opening your eyes. When people say things that are intended to sway your opinion the best action is to reduce the information to basic principles. For instance, here are some examples of rephrasing Sidman’s logic applied to other aspects of life.

But just hitting the brakes does not cause your car to go anywhere. At most, hitting the brakes only causes the car to stop.

But just turning off the burner on your stove does not cook anything. At most, removing the heat merely causes the stove to not cook.

But just hanging up the phone doesn’t allow you to talk to anyone…

But just falling from the sky doesn’t teach you to walk upright…

If you are burned by a stove it is only because you didn’t correctly teach yourself how to use pot holders.

The serious problem in modern training is the avid acceptance of concepts that are at their root, false. Some of these things can only be differentiated if you go back to the foundation. Examining the end statement may actually appear to be splitting hairs while actually describing a deep divide. These differences are significant in the long run. For instance, at first glance you might thing these two things are identical. Trust me, only one of them is edible in the traditional sense. (We shall call your introduction to the stuff on the left P. The stuff on the right is PR) If you bite both and one teaches you to never eat it again, is that an example of learning?

It is every trainer’s responsibility to know the difference between a pile of poop and a pretzel. Quoting someone who doesn’t know the difference just because they have a PhD doesn’t help – it’s still poop. Learning the fundamental truths about behavior is the key to subtracting the pile on the left from the pile on the right – which would be (PR-P=Yummy) No additive combination comes up with an acceptable outcome because the dog poop never goes away, even after you eat the pretzel. That would be a very good thing to learn.

8 thoughts on “The Difference Between Dog Poop and a Bagel:

  1. This whole argument reminds me of an article written by a famous and oft-quoted “Veterinary behaviorist” in an article that discusses why punishment is counterproductive and psychologically damaging to the dog.

    I blogged about it at some point, noting how I completely agreed with this person because, in the cases outlined, the negative consequences outlined did NOT fit the criteria of “punishment” in that they did not decrease the behavior in question. In the descriptions, the actions taken were poorly-timed, incorrectly carried out and often with the wrong tool & technique. In short, everything wrong with punishment was the *people,* and not the dog. Which, then again, leads back to the Behaviorist Manifesto that people cannot be trusted with all the information because, inevitably, it will be farked up and the dog pays the price with psychological discord that will ruin its life forever and ever, amen.

    “Punishment does not teach an animal what to do,” because *that is the point* of a properly-applied and well-timed negative consequence. We “punish” a behavior to decrease its frequency and overall occurrence, and reinforce to increase the frequency of a behavior. That sentence and its derivatives indicate to me that people who claim to know something about behavior actually…don’t.

    After all, as you’ve stated here multiple times, “Reinforcement does not teach an animal what not to do.”

    “So now what?,” I would ask these “behaviorists.” “Now that we have established the most basic of basic concepts in behavior modification, let’s start applying them. What can you teach me about stopping behaviors in a humane, effective and PERMANENT manner in a *short amount of time* in addition to your obviously vast knowledge of how to make other behaviors continue lifelong?”

    • Instead of using aversive techniques (+P and -R, that have been shown in many studies to lead to unwanted behavioral fallout and negatively affects the welfare of the dog, even if “only” with a slight increase in anxiety) because it is easier and faster, learn how to manage and control the situation to prevent the bad behaviors, until you can gently and humanely teach the dog an appropriate way of behaving. You are the human and you are (or should be) in control of the dog. That is what a behaviorist would suggest.

      • Alisha, I think you also need to define exactly what you mean by humane. I personally have an issue with trainers who would leave a dog in a stressed out mal-adaptive state of mind for months and years at a time while they exercise their bias against appropriate effective P+ that would quickly allow the dog to move forward.

        • What stressed out and mal-adaptive state of mind would that be? If it is the result of anxiety or fear, positive punishment is absolutely not the way to go. Classical conditioning and then using positive reinforcement would be infinitely better for the mental welfare of that dog, and when done correctly (as with anything), would not take months to accomplish.

          Gary – I do realize my comment wasn’t really about the topic (sorry!), it was simply a direct response to the last paragraph that Viatecio posted. …I don’t think we should flip the coin and consider the very human fallout of positive reinforcement, because dogs are not human, and we can control what they are reinforced for.

          • You must be new here.

            Have you checked out any of Gary’s videos, in which he works with dogs that are in a mal-adaptive state of mind, and through the judicious use of reinforcement and punishment, has made them into thinking beings that suddenly have control over themselves?

            I’m confused, because you seem to be under the impression that I would take a fearful, anxious dog and just give it a good crank on the collar and TA-DA problem solved.

            Dogs aren’t human. That’s the beauty of working with them, and why I went the way I did after my formal schooling, which actually taught me how to work more with people. (And still does come in handy a lot of the time, but I prefer dogs, thanks.)

  2. Alisha, the topic is not whether to use aversive control or attractive control, but the illogical statement that punishment doesn’t teach anything. As for the advisability of using either attractive or aversive control, that is dictated by the specific context of what you are trying to solve. If you must stop a behavior to resolve a problem then your solution must include aversive control. Also, the “studies” you mention were not performed by people who have any training in the correct use of aversive control. The fallout of correctly applied punishment is always good. I don’t touch cactus spines. I do not touch hot stoves. Punishment protects us from doing behaviors that are dangerous and potentially lethal. That is its function in nature. To flip the coin, consider the “fallout” of positive reinforcement – hubris, thievery, violence, deception, fraud – and virtually all other human vices that start at a baseline instinctive reaction and then become stronger based on contingencies of reinforcement.

  3. Alisha, I punish fearful dogs all the time. That is because I work with dogs that are always fearful. If a fearful dog is injuring itself by busting out of a crate when the owners leave, there isn’t any “positive” procedure that will stop the behavior – because positive reinforcement by definition can’t stop a behavior. Additionally, many fearful dogs will not take treats. They do not want to be touched. Your presence is aversive to them. Your choice is to leave the animal in fear, forever, or use aversive control to correct the problem – that is the correct ethical choice. After that procedure the dog becomes relaxed. I’ve done it many, many times over more than 25 years. However, I would challenge you to ask the people who propose otherwise what experience they have using aversive control and where they learned to use it. You will find that they are not experts and have no training in that area. If you believe them, you are simply swallowing a catechism.

  4. I’ve been working with an anxious Great Dane. We used a combination of bonking and clicking and treating. The owner can’t believe the change he’s seen in her in just a few visits. He says she is much calmer and it’s the first time he can actually see her use her head and make better behavioral choices. She actually comes up to me now instead of hiding under the table now that she understands that it is her behavior that determines rewards and punishments. That is huge information for a fearful dog. Thanks always GW.

Leave a Reply to Gary Wilkes Cancel reply

Your email address will not be published. Required fields are marked *