10 Problems with positive reinforcement

  There is a list out there on the internet of 10-12 reasons why punishment is problematic. As with most unrealistic ideologues, those who wish to espouse that perspective never use logic to examine what they have said. As an exercise in balancing the scales, I offer this list as complimentary to the other one. I no more believe these things than I do the other list. That is because I don’t look at abuses to determine proper use and I have no bias in favor of either behavioral effect. Neither can be determined good or bad without providing a context for their use and an examination of their outcome. So, here goes. These are things that are problematic about using positive reinforcement. I have added the most important problem with positive reinforcement at the end of this list.

  1. You can cause physical pain/damage to your dog. Many dogs are physically injured by being reinforced for activities that are inherently dangerous. Competitive sports such as agility rack up cruciaBrokeLegte ligament, spinal and orthopedic injuries because the dogs are  encouraged to race at their highest speeds. Dr. Chris Zink, a veterinary orthopedic specialist has a number of books, videos and gives seminars on avoiding injury in competitive dog sports. Why is positive reinforcement for obviously dangerous behavior acceptable?
    2. It is difficult to gauge the appropriate intensity. Finding a level of reinforcement that will motivate a dog is a difficult thing. Some dogs are motivated by chasing moving objects. This can lead to chasing children and other pets. As the dog becomes aroused at their instinctive chase-response, they become more likely to become aggressive. This can lead to serious injury or death in companion animals and severe bites in humans. This isn’t a joke. There are dogs that do not crave what you are offering. You can only give a handful of treats so many times during a single training session.
    3. The dog can develop a “reinforcement callous.
    Once positive reinforcement is the established motiva
    tor, it becomes forever necessary. The common term for this is being “spoiled.” Many dogs trained with exclusively positive reinforcement will not work for anything but food. They understand a one-to-one ratio. A click always means a treat. One very prominent clicker guru says that when you click you are obligated to give a treat. I don’t know who made up that rule, but it’s a problem because if the dog senses the absence of treats they quit. The minute you stop continuous opportunities for reinforcement you get petulance and resistance.
    4. The behavior may disappear when reinforcement stops.
    This canard goes back to B.F. Skinner, himself. He suggested in Science and Human Behavior that a problem with punishment is that it might not stop a behavior forever. This is ironic. Positive reinforcement cannot maintain a behavior on a single offering. Of the two behavioral effects, punishment is far more powerful if the future is your concern. A behavior can be stopped forever using punishment. Positive reinforcement requires clever “schedules” to trick the dog into behaving over time. They are predictably unpredictable in their ability to control behavior reliably.
    5. It is difficult to have perfect timing. Using positive reinforcement requires skillful application according to the vast majority of experts.
    The American Veterinary Society of Animal Behavior says it has to be delivered in 1-2 seconds. The average person simply does not have this kind of timing.

    6. It is difficult to be perfectly consistent. It is virtually impossible to have treats at all times. Additionally, after meals, a treat-bound dog is not hungry and may offer no behaviors at all. In essence, prefect consistency is impossible with positive reinforcement.
    7. It can suppress desired behaviors; inhibit offered behaviors. Positive reinforcement has the ability to destroy a dog’s focus on desirable behaviors. For instance, a Jack Russell Terrier in the park may go ballistic over the sight of a squirrel – at the expense of all other behaviors. Such a dog is the victim of positive reinforcement – the reinforcing effect of being allowed to chase other animals.
    8. It doesn’t teach the dog what not to do. Positive reinforcement cannot create inhibitions. Therefore, it can never teach a dog to avoid chewing an electrical cord. Neither can it stop a dog from bursting through an open second-story window. Sadly, thousands of dogs are injured and killed because positive reinforcement leaves the illusion of control where no real inhibition exists. Dartin
    g out the front door is a great example of this. Treats for sitting can gain a semblance of control. Eventually, if the dog is full or a cat runs past, the reinforcement is not powerful enough to maintain the obedient sit and the dog flies out the door.
    9. The compliance effect of the reinforcer is limited to the presence of the discriminative stimuli. I won’t bother parsing this. What it implies is that if you aren’t there the behavior won’t happen. That is true with positive reinforcement.
    10. It is rewarding to the human using the reinforcement. This causes a multitude of problems. Dogs are reinforced for inappropriate and dangerous behavior so often that one can write volumes about the destructive power of the process. By enjoying the use of positive reinforcement, many owners over-feed their dogs to the point of early death and severe tooth decay. Additionally, teaching a dog to dance on its hind legs leads to cruciate ligament surge
    ries. Finally, reinforcing a puppy for growling or barking at guests coming to the front door normally leads to the adult animal reacting incorrectly and biting an “intruder.” This also usually leads to death – because the trainer has been positively reinforced for creating a dangerous behavior regardless of the outcome for the dog.
    11. It can trigger aggression when you try to cut back on the positive reinforcement. There are plenty of scientific studies that show that “extinction” schedules can trigger aggression. You don’t need to look those up. Take a sucker away from a baby and watch what happens. Or, fire someone who happens to be a gun-nut. When people develop an assumption that their behavior will be reinforced, they come to expect it. When it is interrupted it causes anxiety, fear and aggression.

    12. Positive reinforcement allows gurus to thrive by  giving questionable to false advice. Positive reinforcement is an amoral process. If someone is reinforced for saying specific words, they will say them again. If there is no punishment for false statements there is no check on this behavior. Consider that one of the most successful clicker trainers in the world has never clicker trained a dog to any standard. Much of that person’s information is incorrect or damaging to a student’s knowledge of the process. However, here’s where it gets really destructive. Once the guru has acolytes, their status and money is dependent upon protecting the guru. See #11. They will attack anyone who contradicts their idol because it will also undercut their livelihood and/or ego.

    nigerian2

    Somali Pirates – created by positive reinforcement

    Trying to create a world where everything is ice cream without tooth decay is a fruitless task. Positive reinforcement in the absence of punishment is responsible for heinous crimes and acts of brutality. Think pirates, conquerors and every other low-life that ever swindled old people out of their retirement. Punishment in the absence of positive reinforcement destroys souls and creates lives of terror. We, as a species, developed in a world where we must be sensitive to both nice and nasty. That allows us to live and thrive on planet Earth. Punishment for doping off makes us stay focused while riding a bike. The joy of arriving safely at our destination increases the likelihood we will ride our bike again allowing the threat of punishment to keep us safe. It’s not either or. It’s almost always both. The best way to end this is to remember something that Romans said, a long, long time ago. Abusus non tolit usum. That is Latin for, “The abuse of a tool is not an argument against its proper use.” Learning to correctly use reinforcement and punishment is the wise option.

6 thoughts on “10 Problems with positive reinforcement

  1. Very good conclusion. I agree that this goes beyond just dogs, punishing bad behaviors is becoming abnormal now. We as the generation of “everyone gets a trophy” have to make everyone feel good in order to succeed within societal norms. Just tonight I spoke with a man who raised foster kids here in AZ, he said that working for the foster system they could not as care providers punish them in anyway. I was taken aback by this and couldn’t believe it–how are you to correct a child’s life that has already been messed up because of a probable lack of discipline without being able to set limitations and going outside of those causes consequences?

    • Brandon, you hit it pretty square on the head. I have been working in social work for about 14 years. Gary has explained society as a whole but in society we sometimes reward bad behavior or the lack of any behavior. We have punishment that doesn’t work apparently or we wouldn’t have so many repeating behaviors. The polarization in this world is worse now than I ever remember in my 60 years. As you said we reward an effort not the success with same reward. To me it is like shaping, I will reward movement towards the desired behavior, but after awhile only turning your head won’t cut it when a complete turn is the desired outcome. Punishment today is akin to your dog jumping through the screen to get to the squirrel in the back yard and for punishment you make him stay in the back yard. Then when your dog repeats the behavior we medicate him so he doesn’t have enough brain function to know what a screen is.

      • Larry, good points but a minor tweak. Punishment works. If it has to be repeated it was done incorrectly. I don’t touch cactus. If I continue to touch cactus after the prick of the spine, there is something wrong with the process – like wearing heavy gloves that block the ability of my skin to feel the prick. Meaning the goal is contingent punishment – an adjective that is almost never used. Conversely, if someone threw treats at a dog an no behavior developed it would be a tip-off that it was a purely respondent conditioning procedure. I make a differentiation between respondent aversive events that lead to simple associations (I don’t like the mall because I can’t find a place to park) and non-contingent punishment that has the effect of a general suppression of behavior across a broad spectrum that usually negatively reinforces high levels of alertness and active brain activity to identify the “cause” of the punishment. (My car was broken into at the mall so I lock my doors.)

    • man its like you are just moving from one extreme to another… but at least overall the positive reinforcement methods seems like a greater way for people to train dogs over the traditional methods because people who incorrectly use punishment is even more sad than that of reinforcement i think. the lesser of evils? and hopefully eventually a proper balance is realized.

  2. John, go back and read the intro paragraph. “As an exercise in balancing the scales, I offer this list as complimentary to the other one. I no more believe these things than I do the other list. That is because I don’t look at abuses to determine proper use and I have no bias in favor of either behavioral effect.”
    As for the specifics of your comment, you have assumed that traditional training is all negative – an incorrect assumption. As for the dangers of positive reinforcement, my neighbors dog was killed by a car, recently. He slipped out a gate to follow his family as the drove away. His motivation came from months of positive reinforcement that bonded him to them. There is no “lesser of two evils” unless you examine a context. However, positive reinforcement is the motivation for every thug, pimp, conqueror and swindler in history. The way to stop anti-social, evil behavior is the punishment, sweet punishment, used correctly. A world without punishment is anarchy where the strong are allowed, ironically, to punish others with no restraint on their behavior.

  3. I completely agree. My two are therapy dogs, multi-team. I trained them in the Cesar Milan style of “touch cotrection” if commands were not obeyed (but only if they KNEW the command). The touch mimics the “bark and nip” behavior a mother uses with pups as well as dog-to-dog correction communication. It works, and it LASTED. In the last three years I’ve only had to apply it three or four times. On the other hand…
    I have witnessed multiple accounts of other therapy dogs who were trained ONLY in positive reinforcement classes (cookie-pushers) and frankly their levels of training were, to say the least, disgusting. No control, impulsive, both dog and human agressive (I have WITNESSED children bitten in the face, myself bitten, and had my dogs attacked, not just growls or snaps – attacked by these treat-slingers only dogs). I NO LONGER visit, nor allow my dogs around these groups or handlers.
    Many of the points you listed above I’ve PERSONALLY WITNESSED. Many of the dogs have to take further classes as well as rehabilotation.
    The most obvious of the above mentioned problems, imho, is the scrnario of the squirrel. It’s called “Environmental Competition”. Roughly it’s when something means MORE to the dog than a treat… or even a juicy stake. Then all control is tossed out the window.
    Lastly, there is a thing called “respect”. When you were a child, did you have “respect” for your parent? In other words, when you were young and repeatedly told not to run in the street, and received “correction” for it – you gained respect for your parent and this influenced FUTURE lessons. I.E. running in the street got you punishment, and this brought respect. THEN when your parent told you not to do something you LISTENED.., and did not disobey.., because you KNEW the consequences. Pre-emptive behavior.
    Positive reinforcement trained dogs do not see the owners as parent or alpha or top dog, but as equals or subdominants- this is a recipe for disaster.
    Final word: Ask any owner whose ever had children if they punished their child for DISOBEYING them, they’ll tell you yes. And yet, they do not accept that the psychology is the same for our furry-faced friends. A dog’s mother can punish their puppies, or one dog can correct another dog… is this “cruel”? Certainly not! It’s only ehen the human does it.
    I think the answer lays in the METHOD OF PUNISHMENT – which is why I like the Cesar Milan method – the psychology of canine behavior modification.

Leave a Reply

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