Acts of punishment are stronger than rewards

We sometimes assume that choosing punishment or rewards in training is merely a matter of personal preferences. We use the tools that we deem the most effective, and that we have grown proficient with. If we can teach a dog to sit by applying pressure on his hindquarters, why use treats? If leash corrections can inhibit the dog’s attempts to walk out of position or to jump on a person, why use treats? Treats are often considered a bribe for behaviors that could just as well be taught through a savvy application of an aversive that will discourage those undesirable impulses. Others consider the use of rewards as the preferred way to go and avoid inflicting discomfort, authority or pain to the animal. Why not encourage the animals’ creativity and problem solving in training? Why not teach the dog that sitting is fun and enjoyable instead of something to do or else? The choice of rewards over punishers however, is not just a matter of efficiency, ethics and personal choices. Studies on people and animals have shown that negative experiences have a faster and more lasting impact than positive ones. Why? Because one the brain’s function is to keep us alive and learning how to avoid painful or deadly situations is a matter of survival. As a result, using an aversive has a greater risk of impacting the animal’s emotional experience thus creating problems down the road. Many researchers believe that an important function of emotions is to trigger quick and adaptive behaviors to keep us alive. In humans this phenomenon has been the subject to much attention over the past 40 years. Since we share similar brain structures when it comes to regulating emotions, what we see in humans can help us understand what happens to our pets.

Couple fightingIt’s a well-known fact in marital counseling, that to keep our relationship strong, we need a five to one ratio of positive interactions to negative ones (Marano, 2004). It only takes one harsh word, one critical or sarcastic comment to overshadow several declarations of appreciation and love. We tend to hold on to that negative comment, feeling deflated, disrespected and invalidated. Similarly, losing money or friends has a greater impact on us than the opposite (making new friends and winning money). In the same way, even a brief contact at a fancy meal with a cockroach will render the meal inedible. But what is the antidote to this? There is no equivalent. As a Russian saying goes: ‘a spoonful of tar can spoil a barrel of honey, but a spoonful of honey does nothing to a barrel of tar’ (Rozin & Royzman, 2001). Studies on conditioned taste aversion in both people and other animals showed that it only takes a single trial for the learning to occur and have a lasting effect (Garcia & al., 1974). The bad generally wins over the good. Why such asymmetrical effects? Because it’s much better to overreact to a perceived threat, than to not react to a real one! When it comes to staying alive, we might have just one and only chance to react and to learn from the experience. On the other hand, if we don’t pay much attention to a few berries on the side of the road, we’re probably going to be just fine.

Mountain lionSo this negative bias has an adaptive value. For one, negative events are more harmful than positive events are beneficial. There is no positive equivalent to death for instance. So learning which events can lead to a fatal and irreversible consequence is of the highest priority. Negative situations generally also require a quick response and don’t give many options for trial and error, where positive ones can be experienced over and over again. The faster an animal can identify danger, the longer it is likely to live. But in the wild, the brain can be quick to react to a danger, but also quick to go back to a balanced homeostatic state. A wild horse will run for his life when chased by a mountain lion, but will go right back to grazing as soon as the predator is gone. In between the attacks, there may be days of careful watching of the horizon, but no real stressors.

Scolded dogSo how does this apply to animal training? Any aversive, in other words, anything that is perceived by the dog (or the horse, cat, ferret, etc) to be unpleasant, scary or painful will have a more lasting effect than anything pleasant. Moreover, since negative stimuli get the animal’s attention, they also take away from their ability to problem solve. How do you try to think through a situation and understand what to do when your emotions are up and your attention is on the aversive? So using leash corrections is not the equivalent to rewarding the dog for walking in position, even if in the end, the dog could learn to walk on a loose leash and in a heel position in both methods. From a neurological perspective, every correction will negatively impact the dog’s psyche, inducing stress and potentially even feelings of helplessness (due to the required number of repetitions of the leash corrections). As we’ve discussed, in a natural situation, animals can quickly go back to a more relaxed state of mind. Our brain has not adapted to continuous stressors. In fact, animals are predisposed to avoid negative experiences, by moving away or going after the source, thus minimizing their long-term exposure. What happens when they have to endure them on a continuous basis? One of the arguments in favor of using an aversive that I’m often given, is that life is full of negative experiences, so adding a punisher here or there will not have that much of an impact on the animal. But it is precisely because life already comes with it’s own set of challenges that we should do what we can to minimize how many negative experiences our companions have to overcome.

How can we help our pet overcome a negative experience? Fortunately, through repeat exposure to positive experiences, new neural connections can gradually develop over time. That’s how animals learn. But it takes many more positive experiences to replace one negative one and the more traumatizing the event, the longer it will take. This is also why using an aversive can be risky. Depending on the level of sensitivity of the animal, but also on what associations the brain has developed (e.g.: the person administering the punishment and the negative experience), we might end up with more work than benefits. Undoing the effects of negative experiences takes time and commitment. Without going into any ethical considerations, from a practical standpoint, it’s a lot easier and more efficient to favor positive associations over the negative ones.

This ‘negativity bias’ is also one of the reasons why it’s so important to keep the animal under threshold when working on fear and reactivity issues (for more on this subject, you can read the post “Reactivity and Aggression in dogs – Managing and Treating“). If every negative experience has more impact on the brain than many positive ones, any time the dog is scared and/or reactive, we have just made things worse. The neural connections in the brain are strengthened. It will take even more repetitions of positive exposures to the scary stimuli to overcome the fear.

With the animal brain wired to give particular attention to negative experiences, it is essential to understand how much this phenomenon impacts our pet’s behavior. Helping them feel safe, develop coping skills and positive associations with their environment, is a prerequisite to their wellbeing. They will endure discomfort, fear and pain no matter how careful we are. Although we don’t need to become helicopter guardians, using an aversive, should always be of last resort when working with animals and more generally, when relating to others. For more information on the subject, Rick Hanson Ph.D., neurologist and psychologist gives an insightful presentation on the subject in the video below.

Jennifer Cattet Ph.D.

 

 

Animal training is about animal welfare

When is Controlling your dog too controlling

logo

Jennifer Cattet Ph.D. has been working with dogs for over 30 years, as an ethologist with the University of Geneva (Switzerland), a trainer and a behaviorist (in both Europe and the US). As Director of Training for a service dog organization in the U.S, she supervised and taught offenders in the training of service dogs. Today she's the owner of Medical Mutts (MedicalMutts.com), a company dedicated in the training of rescue dogs as service dogs for conditions such as diabetes, seizures, PTSD, autism, etc. She's also part of a research team working on understanding the ability of dogs to detect changes in blood glucose levels through scent. Jennifer also works with Smart Animal Training System on the promotion of reward based training and the development of technology to support it (SmartAnimalTraining.com).

Tagged with: , , , , , ,
Posted in Dog behavior, Dog training, Dog/human relationship, emotions, Fear, Misc, Positive Animal Training, Psychology
18 comments on “Acts of punishment are stronger than rewards
  1. Gerry says:

    While I agree in principle and with many of your general points, there remains a question of perceived degree or intensity of an aversive. And every negative experience does not have more impact than many positive ones, although some will do so. If your statement were always true, then most dogs being told “no” would have behavior issues, instead of forgetting the incident shortly after. (Garcia & al. seems to more compare conditioning to homeostasis as separate from reward/pain, as regarding the lasting effects. )

    People tend to like simple rules that can always be applied. I’ve seen people who were taught to always control their dog during a walk, to the point where the poor dog almost never gets to sniff or explore anything. They think they’re doing the right thing, but one could very easily call that aversive as it causes the dog distress and frustration.

    When Aimee Sadler started her Playing for Life program, a number of people objected to her use of what they considered aversives. After speaking with and watching her, I found that another case of people taking simple rules to the extreme, with no evidence of any issues in the dog’s subsequent behaviors. The “negativity bias” cannot be applied to external perceptions of adversity, but only internal ones of sufficient intensity and importance.

    Yet another case of controlled aversives in teaching dogs can be found with older puppies, where their behavior is now being corrected by socially skilled adults. While those corrections are aversive, the limited intensity and controlled recovery times prevent issues and promote learning.

    Now a request: please define the term “leash corrections”. I find this to be one of the most misused and abused terms. When I gently jiggle the leash & my dog stops sniffing, turns and walks along with me, why isn’t that a leash correction, as I am using the leash to “correct” his behavior? When he wants to stop and sniff, he may gently tug. If I give him slack he goes to sniff, and if not he continues to walk. And I do not become upset over having been given a leash correction by the dog. So many people speak against leash corrections, and so many others have no idea what they’re really talking about.

    In general, communications, companionship and reward to promote learning, while mild aversives are briefly used primarily as interrupters, and are not usually intended to modify subsequent behavior.

    • Gerry, thank you for your comment. The point of the bog is merely to point out that animals are predisposed to react differently to negative vs positive experiences and that we need to be aware of the potential impact of using an aversive. No matter how positive we try to be, we all use aversives at one point or another (just putting a leash on a dog can be an aversive). There are also times when the good outweighs the bad, it all depends on the intensity of one or the other. In most casts however, the negativity bias takes over.

      Anything can be taken to the extreme, but the world isn’t black and white, good or bad. The wisdom is to understand what is at play and make decisions based on the animal’s welfare and the efficiency of the method. There are models available to help us make those choices of when to use an aversive (like O’Heare’s LIEBI model http://www.associationofanimalbehaviorprofessionals.com/liebi50.pdf). The more tools we have in our toolbox, the better we are at making those decision while staying flexible and willing to make changes along the way.

      I’ll write more about punishment in training and behavior modification in future posts. As you pointed out, there are many subtleties and nuances. Too many for just one post.

      • Gerry says:

        That the world is gray is perhaps the most important point here. Thank you for O’Heare’s LIEBI model, which appears to be an expansion (as he noted) of others, such as the LIMA principal of Cynopraxis from Lindsay, which I use as a basis.

        On determining the threshold of negativity bias, I suspect we can just agree to disagree. As you said, there are many subtleties and nuances here.

    • Wes Anderson says:

      Gerry, Good point about the factor of intensity! Not all punishment (or reinforcement) stimuli are equal. I will say however that I have heard at least one very well known trainer of zoo animals say that even mild aversives can create a situation of distrust. Distrust is not a good thing when working with killer whales or bears. Dogs are a more forgiving species.

      • Great point Wes. While dogs can recover quickly from a mild aversive, it is still a stressor. The dogs sensitivity, the intensity and the frequency of the aversive will determine how much of an impact it had on the animal. Just like we can feel uncomfortable with a person because they simply looked at us with a certain facial expression, the animals we work with can experience anxiety to very mild forms of punishment (a look,’no’,’ah ah!). So we need to understand how they can affect our pet and be mindful about their use.

      • Gerry says:

        Wes, I agree and feel your point also relates to my comment on internal perception. And that the stressor distinction between eustress and distress relies on both nature and nurture. While dogs may be potentially more forgiving to common human imposed stressors (co-evolution, etc.), I’ll bridge your example with dogs who are entirely feral. Where initially many types of aversives must be avoided, as they may rapidly cause the situation of distrust you mentioned. Later in the process, however, their internal perceptions of these same stressors will often change. Similarly with very fearful or phobic reactive dogs, distress stressors are defined not by a book, but by their reactions to mild forms. With differences also found in type between your case of zoo animals, and the same ones in the wild.

        • Wes Anderson says:

          Ah yes biology is dynamic and static! What is true for a specific animal today may change tomorrow. The systems are very complex and we try to describe them in simple ways…and they refuse. 😉 Thanks

      • Christine says:

        Gerry, just because dogs are more forgiving, does it make it right to use aversive a when positive reinforcements can work instead?

        • Gerry says:

          Christine, let’s look at each piece of what you said.

          On “forgiving”, each animal by their nature and history will tend to “forgive” some types of aversives more than others. As I had noted, feral dogs will often not forgive actions that your own dog would simply ignore. So dogs are not simply more forgiving and I don’t treat them that way.

          On aversives, what are they really? If you check Jennifer’s reference to Garcia & al, you will see there are many more events that are aversive than you may realize, and Garcia refers to their relative intrusiveness. Some aversives are so mild they may interrupt or prompt behavior but are immediately forgotten or have no significant or lasting effect. Garcia makes that point, and consistent with it I have argued here for a threshold for negativity bias. But that’s all I argued on aversives.

          On positive reinforcement, it is not a panacea, and even cases that use it effectively will still contain components of other types of conditioning or desensitization. If it always worked alone in all cases, Garcia’s paper would not have been written or referenced by Jennifer. Still, I mentioned as a guiding principle Lindsay’s LIMA, using the Least Intrusive, Minimally Aversive technique. That’s very similar to the LIEBI from Garcia that Jennifer spoke about. Those are the guiding principles that determine “right”.

          Returning to my earlier example of adult dogs teaching puppies, why do the dogs use corrections to inhibit unwanted behavior? Are they not coercive aversives? Are they damaging the pups? Could they possibly do all that training with only positive reinforcement (which they do use)? As Jennifer said above, one post is too short to fully explain this.

          Skipping back to my question on leash corrections, many people have spoken out against its use, but caused confusion by not defining the term. Much of my work follows that of Dr. Karen Overall, who qualifies leash corrections by the use of a choke or prong collar, or forcefully changing the dog’s position or behavior with the leash. Often the hardest part of leash training a dog is to get the person to stop pulling them, and adding confusion there makes it that much more difficult. While a single post may be limited in scope, we should at least attempt to define our terms, and not imply that things like aversives or stressors are necessarily bad.

    • Melissa says:

      Thanks for the stimulating discussion, Gerry and others. I thought you raised some good points, Gerry, especially on leash corrections and when does a negative experience provoke negativity bias? I did my PhD on cognitive bias in dogs, and I do not know the answer to that last one. As far as I am aware, no one has figured out yet what can alter whether dogs expect more positive or more negative outcomes. I should note that I would generally think of “negativity bias” as a bias towards remembering negative experiences over positive, and being better able to recall memories of negative experiences over positive, which is not really the same as a negative judgement/expectation bias, which is what we are looking for when measuring pessimism. The former I think is useful in explaining how easily fear conditioning and the likes can occur, while the latter more useful in explaining why repeated negative outcomes are not good for a dog’s emotional wellbeing. But, either way, we still don’t know what kinds of conditions result in negativity biases or negative judgement biases in dogs. AFAIK, it’s surprisingly difficult to pick up a negative judgement bias shift in dogs. It may be how we are measuring it, or it may be that dogs are pretty resilient in general on this front. Having said that, I have met dogs in the course of my research that do have a negative judgement bias, and appear perfectly happy most of the time. The effects are kind of subtle, but they do seem to be more fragile in training than dogs with a positive judgement bias. At times they can find some pretty mild things that are not even supposed to be aversive quite distressing. It’s not that they are more sensitive to potentially aversive stimuli so much as they are sensitive to outcomes that are not resoundingly in their favour. An optimistic dog doesn’t really care if they get hits and misses, as long as they are getting some hits. Pessimistic dogs care, and they seem to feel the misses far more keenly. It doesn’t have to be a miss as in an aversive stimulus. Just a miss as in they were hoping for something good and they didn’t get it.

      A great blog post, as usual, Jennifer.

  2. Sonya Bevan says:

    Brilliant. Thank you. Sharing. Keeping.

  3. Thank you, thank you, thank you for this brilliant and beautiful article. I have come to understand that for some folks, the need to use aversives/corrections/negative feedback, etc. is ingrained in them like a belief. As if the Old Testament is an important part of dog training. Rather than even try a month without it, they will spend years arguing for its importance. It becomes, for many, an issue akin to religion, or political party, or pseudo-scientific beliefs: their need to believe in it is more important than their need to learn about the possibility of something different or better. The 5 to 1 formula is a lovely example, and it is for adult humans; I think it is closer to 10 to 1 for young dogs. There is no more annoying a zealot than a convert, and I count myself the converted. I am ashamed of things I did to dogs 30 years ago; if I have an excuse, it is that I was doing what I was taught, but there is no excuse any longer.

    As to the question regarding degree to aversives, or of defining a leash correction: a leash used properly is never used by the handler or felt by the dog. It is there as a back-up plan in case something unforeseen happens. Of course aversives happen in degrees, but being a little bit of a brute, or a little bit scary, or a little bit annoying, doesn’t make it okay. As Bob Bailey said, “Pavlov is always on your shoulder.” What is that you are trying to condition your dog to feel? That you are only a little bit scary? Only a little bit annoying? Only slightly violent?

    The question to ask: If I can so this without aversives (whatever the degree), why am I using them? We can do anything with rewards that we can do with aversives. There are enough dog trainers doing any and everything with dogs, without any aversives at all, that we know we don’t have to use them. The need is not from the dog, the need comes from the human.

    • Gerry says:

      Michael, you might look up the definition of pseudoscience in Wikipedia. While I have disagreed with Jennifer at times, it has been more a question of degree or scope, and often in only details, rather than the type of extreme opinion you seem to present here.

      For an overall comparison of the Force Fee Purists against scientific approaches, I would suggest Brad Griggs at:
      http://www.k9services.com.au/2014/punishment-vs-abuse-philosophy-vs-ideology/

      • ladychaunceybarkington Me says:

        I don’t see a comparison, I see an emotive rant against something the author doesn’t have a grasp on. I’ve also yet to meet a behavior professional using purely positive reinforcement, or purely positive reinforcement and positive punishment.

        Just calling professionals “purely positive” shows a lack of comprehension of operant conditioning quadrants, as well as their respondent conditioning counterpart.

        I train without coersion because, to me, it’s logically the best choice. Ethics are my own and not necessarily shared by clients so emotions don’t enter into my arguments for it. I also train without coercion because I don’t just work with dogs, I work with species not domesticated to put up with coercive liberties on our part. I can’t try to force them to do my bidding and walk away in one piece. That leaves me with no need to use them on dogs either, even though I very well could. I can do better.

  4. Kate says:

    The author sites psychological studies and terms and then completely miss uses the terms. To me this discredits the author right off the bat! The worst offense is the word “positive” and “negative”. They just mean add and subtract not reward and punishment. You can have both positive and negative punishment and both positive and negative reinforcement. She makes a lot of guesses with absolutely no support. She states the punishment is stronger than reward. She doesn’t back this up at all or anything else she says for that mater. Not once does she reference one real study, just declares they exist. She clearly has never heard the endless stories of dogs running through invisible fences over and over just to get to something they wanted! Also she states that missing rewards is not as important to an animals survival. Clearly she hasn’t seen predictors take a beating from their pry so they won’t starve to death. For once I would love to see a dog training article based on real since and backed up with real studies and for the love of god get the terms right. If you don’t know what the terms mean then how the hell can an author correctly interpret real studies!!!

    • It is true that when we’re talking about the four quadrant, the terms positive and negative are used in a mathematical sense and refer to adding or taking away something. This does not mean that these terms can’t also be used in other contexts. The four quadrants take into consideration the effects on behavior, in this blog I was only referring to whether something was pleasant or unpleasant.
      If you click on the links in blue in the text you will find a few studies on the subject.
      For a review of the four quadrants (positive punishment, positiver reinforcement, negative punishment negative reinforcement), please read the following post http://blog.smartanimaltraining.com/2014/03/24/learning-theory-basics-part-2-operant-conditioning-if-it-works-do-it-again/

      Jennifer
      P.S.: I appreciate constructive feedback and love hearing what my readers have to say. If you disagree or do not like the subject or the content of a blog, please state your point. If your arguments are valid, there is no need for personal attacks. We can always agree to disagree.

3 Pings/Trackbacks for "Acts of punishment are stronger than rewards"
  1. […] In a blog posted by Jennifer Cattet Ph.D. in http://www.smartanimaltraining.com she says that for us, it only takes one harsh word, one critical or sarcastic comment to overshadow several declarations of appreciation and love. We tend to hold on to that negative comment, feeling deflated, disrespected and invalidated. We need a five to one ratio of positive interactions to negative ones (Marano, 2004). It takes at least five good things to wipe out a bad one […]

  2. […] For me it was a somewhat gradual shift in thinking. The biggest push, mentally, was when I got my JRT. When he was about five months old I put a prong collar on him to teach him to heel. He froze and wouldn't move when the collar tightened. I'd never had that experience with the dozens of dogs I'd trained using a prong. I wasn't ignorant about positive reinforcement at that point. I have always used it when training but I also used aversive methods. Seeing my young feisty little terrier freeze with a look of abject fear on his face slapped me in the face. At that point, I was still training retrievers though not as actively as I had been. I was using a shock collar for advanced training and on other aspects of retriever training. I had also around that time joined a dog forum that had several very educated trainers who were actively promoting force free training. They shared many articles and studies about the use of punishment and fallout. I realized that some behaviors ( and some were very subtle) I was seeing were indeed the result of my methods. Oh and I never put a prong collar back on my JRT after that first time. I kept mulling over my methods which were the methods almost everyone in my dog world used at that time. The more I read on the subject the more convinced I was that I had to abandon using aversive methods if I wanted to consider myself a humane trainer. To this day, I am still learning about how dogs and all animals learn. Punishment does work. All animals are wired to respond to pain with avoidance. So you can stop behaviors you don't want using punishment. Where it often breaks down is that for the average dog owner and for many trainers, punishment is not delivered correctly and with enough pain/force to effect a result that is long lasting. The use of pain to stop behavior may be more efficient in the animal world but at what cost to our relationship with our dogs? Saw this article today on FB. Very timely,eh? Acts of punishment are stronger than rewards | Smart Animal Training Systems… […]

  3. […] die Möglichkeiten der praktischen Anwendung von Zuckerbrot und Peitsche im Hundetraining und deren Folgen […]

Leave a Reply

Follow via Email

Subscribe to our mailing list

* indicates required
I am interested in:
Disclaimer
%d bloggers like this: