Each of these photographs shows a dog jumping up on a person…..with the person appearing to be quite happy about the interaction. Yet, jumping up to greet is a frequently cited complaint that dog owners make to dog trainers. While I am sympathetic to owners’ frustrations, the underlying cause for jumping up in most cases is simply a dog who is exuberantly saying “Hello! How are ya! How about some lovin?” And of course, a contributing cause is that, as seen above, we are frequently inconsistent in our responses; some of the time we enjoy it and encourage it, and in other circumstances (such as when Muffin knocks over Grandma or places muddy paws on a visitor’s white dress) we are not quite so happy about it.
I AM A BAD DOG – I JUMP UP ON PEOPLE
Although trainers’ opinions certainly vary, my general view of jumping up is that “Yeah, dogs do this. Yeah, many owners dislike it and would like to stop it and we can help them with that. However, in the overall scheme of things, having a dog who is a bit too friendly and exuberant really isn’t such a terrible problem to have now, is it“? So, my training school’s approach to the problem is to first frame it in terms of what it actually is…….a dog being friendly. Maybe even an overly excited, young, goofy maniac of a dog, but still, he is just trying to say hello, albeit with an enthusiastic demeanor.
I AM A GOOD DOG AND I LOVE YOU!
We then address ways to modify a dog’s behavior to reduce jumping up to greet. One approach with unwanted behavior is to train the dog to offer an alternate response that is incompatible with the undesired behavior, a technique called “response substitution”. For example, in the case of jumping up to greet, we teach Mr. Exuberance to “sit for his lovin'” rather than to jump up for it. First the owner, then visitors, crouch to greet and reinforce with treats and affection, eventually shaping the behavior to achieve sitting for greeting with folks staying in an upright position as they reinforce sit or simply “keep four feet on the floor”.
Photo used with permission (Sophia Yin, DVM, MS )
An alternative approach is extinction. Extinction refers to purposefully and consistently preventing reinforcement of the unwanted behavior until the dog stops offering the behavior (i.e. the unwanted behavior is extinguished). Using the example of jumping up, this means removing the person. Because greeting the dog (providing petting, love, interaction) is what positively reinforces and thus maintains jumping up, extinction involves telling the owner to ignore the dog by turning her back, backing away, or walking away from the dog whenever he attempts to greet by jumping up. (Note: In this particular example, extinction is almost indistinguishable from negative punishment).
EXTINGUISHING JUMPING BEHAVIOR
Does Extinction Work? However, actually putting this technique into practice presents some problems. It is associated with what traditional behaviorists (of the Skinner “can’t talk about internal emotional states” ilk) refer to as an “extinction burst”, and what trainers typically refer to as “frustration and emotional distress”. In practice this means that the dog increases his bid for attention by following the owner and jumping more emphatically, becoming more active and frantic, barking and whining, nipping or mouthing. Generally, you end up with a mess – a dog who went from jumping and happy to jumping and distressed and who may have added one or two new and equally undesirable behaviors to his repertoire.
Why we need data: Which all begs (pun intended) the question – Does extinction work well as a practical training approach with dogs? If it does effectively reduce unwanted behaviors, does it come at the cost of unnecessary emotional stress and the risk of creating new problems? Regardless of my personal opinion of using extinction in dog training, all that I really could say about its use has come from personal experience with my clients and their dogs and discussions with other trainers. And despite the existence of an effective alternate technique (response substitution), ignoring a dog who jumps up (extinction) continues to be frequently recommended to pet owners by a wide variety of trainers, behaviorists, bloggers, veterinarians and authors.
Published Study: Until recently, there were no data that specifically examined the use of extinction with pet dogs. However, building on work from a previously published study, a group of researchers recently asked the question “Does the use of extinction with dogs produce an aversive emotional state [even while it may effectively reduce the targeted behavior]?” (1,2). In other words, does the use of extinction in dog training cause emotional distress?
Study Protocol: Altogether, 45 dogs who lived as companions (i.e. in homes with people) were studied. The protocol included three phases: The first was a warm-up during which the dog met the trainer and became accustomed to the training area. The second was the training (acquisition) phase. The trainer used dried liver treats to positively reinforce “gazing behavior” (looking into the trainer’s face) each time that the dog offered it. Three sessions lasting 2 minutes each were completed with each dog. The liver treats were kept in a container, located on a shelf, next to the trainer. The third stage was the extinction phase. The experimenter continued to stand near the shelf but now ignored the dog when he/she offered the previously reinforced gazing behavior. Three extinction sessions were conducted for each dog. Each session was video recorded and rated by an impartial observer following the sessions.
Study Results: While using extinction significantly reduced the targeted behavior (gazing), it also led to an increase in behaviors that are associated with frustration. These included withdrawing from the trainer, lying down, increased movement (ambulation), whining, sniffing (often considered to be a displacement behavior), and avoiding the trainer. The researchers note that these results are especially relevant given the common use of extinction for discouraging unwanted behaviors in pet dogs.
Take Away for Dog Folks: For trainers, this study (and the researcher’s previous work) showed that extinction can effectively reduce a previously reinforced behavior in pet dogs. The results also showed that, while effective, extinction causes stress and potentially leads to displacement behaviors that can be problematic. I finished reading this article thinking about two important differences between the study protocol and the use of extinction in everyday life with the dogs who live with us:. These are:
1. The trainer in this study, while friendly and pleasant to the dogs, was unfamiliar and had no previous relationship or enduring bond with the dogs.
2. The targeted behavior, “gazing” was trained for a very short period (6 minutes!), and so had a very short and weak reinforcement history.
Together, these two facts suggest that the behavior that was trained (gazing) and subsequently extinguished was not a persistent behavior that held a lot of significance to the dog. It had an almost ridiculously short reinforcement history and involved a person who really held no importance to the dog. Yet, extinction still caused frustration and emotional distress in the dogs in this study. Wow. Soap Box time……..
So, one can imagine the degree of frustration felt by a dog who loves his people (and visitors) and who has a long history of being reinforced for showing what he considers to be just normal doggy affection (jumping up), when suddenly, all of the positive stuff abruptly stops. His pals begin to ignore him completely, turning their backs, walking away, not speaking to him, whenever he attempts to say hello.
Personally, I found the results of this small study compelling if simply to suggest that it may be time to consider the extinction of extinction in dog training. This is certainly not a difficult call, given that we have available other, more effective and less stress-inducing approaches, such as response substitution. Although training a dog to sit for greeting takes a bit more time, patience, and tolerance (jumping up is NOT the end of the world, after all), certainly it is preferable to using a technique that causes emotional distress to the dog, has the potential to cause other problems, and removes yet another opportunity to interact positively with our dogs. Nuff’ said. Off soap box. (Read the papers to learn more!)
1. Bentosela M, Barrera GT, Jakovcevic A, et al. Effects of reinforcement, reinforce omission and extinction on a communicative response in domestic dogs (Canis familiaris). Behavioral Processes 2008;78:464-469.
2. Jakovcevic A, Elgier AM, Mustaca AE, Bentosela M. Frustration behaviors in domestic dogs. Journal of Applied Animal Welfare Science 2013; 16:19-34.
Excerpted from: “Beware the Straw Man: The Science Dog Explores Dog Training Fact & Fiction” (2015).
27 thoughts on “Is it time for the extinction of extinction?”
Pingback: If you don’t eat yer meat, you can’t have any pudding! | The Science Dog
Pingback: You Barkin’ At Me? | The Science Dog
Pingback: Biting clothes - Page 2 - My German Shepherd Forum
Thanks Erin! (BTW – There will be a collection of the “best” behavior and training Science Dog essays published in book form later this year).
Brilliant! How did I miss this post last year!??! Thanks as always Linda!
Interesting article! I’m definitely having a hard time grasping the concept that we should get rid of extinction. My view is that all learning (behavior change) causes some frustration and some stress. After all, you’re changing a strongly reinforced habit (in the case of jumping). I strongly support setting the dog up for success and things like baby gates and leashes help up prevent the dog from jumping on the person, but a huge part of the equation is to stop rewarding for jumping. At some point, the dog will resort to a previous behavior in the course of learning, since learning is not linear, and the dog will jump – and you have to ignore it, and it may confuse the dog. It is a part of the learning process. I think extinction can be necessary in some cases, but should be accompanied by proper reinforcement of desired behaviors to limit frustration. Thank you for making us rethink this training strategy though! 🙂
Hello Ms. Crossover! Thanks for reading and for your thoughtful comments. I think that the basic difference in philosophy that we may have is found in your first statement “My view is that all learning (behavior change) causes some frustration and some stress”. Why do you think that all learning is stressful? Are there not examples in dogs (and in humans) when learning is an inherently pleasurable experience? For example, when I learn something new about a topic that interests me, I would not say that I feel stressed or frustrated, but rather that I enjoy learning and feel a great sense of accomplishment (pleasure) when successful. Similarly, as a clicker trainer, I would argue that most of the new behaviors that my dogs learn with a clicker are enjoyable to them and that the only time they may be stressed is when I set criteria too high or when I attempt to free-shape something that confuses them. If you would agree that there are at least some instances of learning new behaviors in which our dogs are NOT stressed/frustrated, then why would one choose to use a technique (extinction, in this example) that has been shown to cause frustration and stress over another technique (response substitution) that does not? If an alternate, less stressful (and imho more successful) technique is available, why continue to use the technique that stresses and frustrates the dog? Regarding your point, and it is a good one, that at some point in the training process the dog WILL jump and that the owner must not reward this, again with respect, I disagree. When this happens, the owner can simply ask the dog to sit, reward with a very low level +R or even just praise, and then practice four or five “sit for greetings’ for a higher level +R. Thinking that we must NEVER reinforce an unwanted behavior, is, in my humble opinion, not so very different from the rigid thinking of correction-based training that dogs must always comply, with our (often quite arbitrary from the dog’s point of view) commands for obedience. Again, this is just my opinion as a trainer, and reflects a philosophy of training rather than a point that is based upon evidence. Regardless of our minor disagreement, thanks for your thoughtful note and for following the blog – I enjoy the conversation! Linda
Love your way of thinking! I’ve tried for three+ years to teach my very social dog #1 not to jump up on people and #2 not to demand-bark. He’s learned a great deal of self control in the jumping department and was recently able to pass his CGC evaluation. Yay! but I know he’s not “cured” of jumping. We have more work to do, but we’ll get there. BTW, his jumping up is a BIG problem because he weighs about 90 pounds! The demand barking is still problematic after all this time. I’ve always been advised to ignore it, but my gut finally told me that he’s needy and ignoring it just adds to his frustration. I also felt inadequate as a trainer because I hadn’t been able to extinguish his behavior. I remember when he was a wee pup and I was talking casually with my neighbor (who has no dog training expertise) about ignoring the barking. She thought that maybe – like with babies – I could go to him and comfort him until he settled down. Of course my response, was “oh no, that would be reinforcing!”
Hi Debbie – Thanks! I am glad you enjoy the blog! (Feel welcome to suggest topics; I love hearing from trainers and am happy to investigate specific topics for folks when I have the time available). Just a thought regarding the demand barking – Have you tried “preemptive redirection”? We use this term at our training school to refer to response substitution that occurs before the unwanted behavior is actually triggered. As I am sure you know, most dogs, like people, follow and seek patterns in their behaviors, so we can often identify those patterns and the triggers that precede the unwanted behavior. If you can identify even a few of those with your boy’s demand barking, perhaps you can redirect him with a toy/stuffed bone/alternate behavior before he tunes up (I also think that the barking itself is at the very least stress-relieving, if not actually enjoyable for our dogs, so in that respect once it is triggered the barking itself is reinforcing). Just a thought, as you very well may have already tried this with your boy. What I especially like here is your attitude – love reading this and knowing that you are spreading such a positive outlook about our relationships with our dogs to your clients!
“Yeah, dogs do this. Yeah, many owners dislike it and would like to stop it and we can help them with that. However, in the overall scheme of things, having a dog who is a bit too friendly and exuberant really isn’t such a terrible problem to have now, is it“?
Couldn’t agree more! This is why, as you so well explained, we should find and alternative behavior. Jumping is a friendly gesture at its core and we don’t want to risk damaging the intention behind it.
Hi Amy, Thanks for reading and for your comment! Love your final line “jumping is a friendly gesture at its core”. Very well-stated and even though we know that a lot of folks, especially those with large and/or extremely exuberant dogs, still want to stop or reduce jumping, I think that framing it in these terms brings a less reactionary and negative perspective to the whole issue.
Pingback: A Scientific Look at Dogs Jumping Up - Hostal de los Perros
Hey, I totally think dogs can feel “frustration” and over time it can have dramatic affect on their well being. After all, just read that scientists are learning dogs are self-aware.
In the mean while, my dogs jump all over every one in our family.
That’s why we have dogs! We just do a lot of laundry and have great abs.
Hi Bonnie, Thanks for you comment (and for the smile)! Glad to have you following the blog!
Interesting – of course we cannot simply do away with extinction, especially if you’re doing any free shaping. Free shaping depends upon extinction, of the previous criteria in order for the new criteria to be learned.
If you’ve ever done Karen Pryor’s shaping game with humans, you know how excruciatingly frustrating that can be, even for the most simple of target behaviors. Having done this often in a Clicker Training Certificate course, it was enough to confirm my view, that when I do actually do shaping, it’s not of the “Free Shaping” type, but rather what Kay Laurence calls “Intelligent Shaping”, where one can help a dog over rough patches with things like pointing, touch an object to be touched, targeting where necessary to avoid just this type of extinction frustration. For the frustration comes from “Well I just did it 5 times and it was right, why is it not right now and WTF DO you want me to do?”
Hi Leonard – Excellent point about free-shaping! In essence, we are “extinguishing” each behavior that a dog offers in hope that is “the one”, and this can be extremely frustrating for dogs. I agree also that true free-shaping is difficult to achieve and perhaps not something that works well with many dogs. Last week, one of our trainers taught a single-session “101 Things to Do with a Box” session at our training school. All of the trainers who participated are clicker-trainers and it was so interesting for us all to realize how much we wanted to help our dogs by luring or adding environmental cues to help them to choose a behavior. In that way, it was a great exercise in self-assessments of our own training, as well. Dogs differed a lot also – My youngest Golden, Cooper, constantly looked to me for help, finally figuring that he should just offer eye contact (aka laser-beam eyes :)) instead of interacting with the box. I finally had to help him as he was so upset that he did not know what I wanted from him with that danged box! A friend’s Akita, on the other hand, was our “box star” of the evening – every time I looked over, he was trying something else – I was certain he was going to be spinning around on top of the box at one point :)! Thanks for the comment – Happy training! Linda
This all makes perfect sense to me if we’re talking solely about jumping. I’m working with demand barking with a dog who is bonded to me. He begins the demand barking often after we’ve done a short training session. Would you still find that extinction isn’t recommended with that behavior? If you think an alternate behavior would be a better approach what alternate behaviors do you think work here? I’m having trouble thinking of one that is incompatible with continuing to bark in the way that sitting is incompatible with jumping.
Hi Kizzbeth – Personally, I avoid using extinction for any problem behavior for which access to or interaction with a person is the reinforcing stimulus. Not only does it cause unneeded stress (as data seem to support), but also because it does not work. (I have yet to meet the owner who says to me “Yeah, I ignored my dog for doing **** and it worked great!”. Rather, we typically hear “Yeah, I try ignoring him, again and again, and it just is not working”).
Consider that teaching our dogs alternate and incompatible behaviors should, as much as is possible, be a proactive rather than a reactive activity. In the case of demand barking in this example, you know that at least one of the triggers for your dog’s barking is the excitement and enjoyment of training and perhaps the anticipation that the fun is ending for him and he prefers that the fun continue. 🙂 Since you know the trigger, a proactive use of an incompatible behavior might be to end the session with a down stay on a comfy bed with a stuffed Kong to keep him busy for 20 minutes or so; another alterative might be to end each session with a short walk or a game of “find it’. While these measures may not change his behavior 100 percent of the time, they often have the effect of removing the stimulus (the training session ending in a predictable way each day), and providing an alternate activity to barking.
Similarly, the “sit for greeting” that I discussed in this blog piece works best if the sit is requested before the dog even thinks of jumping up. I find the entire scenario of “dog jumps up, so then I turn my back and walk away” not only to be an awfully rude way to treat a family member who loves us, but also a method that simply does not work very well because it is reactive rather than proactive. In other words, the dog already offered the jumping up (i.e. one more practice session for him), before anything was done; much better to prevent it altogether and teach him another way to get his lovin’ IMHO.
If we remember that dogs, like humans, are pattern seekers and pattern followers, we can usually find one or more triggers or daily patterns that precede an unwanted behavior and either change up the trigger or remove it altogether to alter the patterns that are not working well for us or for our dogs.
Last, a disclaimer……..I am not citing studies for the above; this is just my opinion as a trainer and I realize that trainers vary (more than a little bit!) in their approaches. Studies are what The Science Dog is all about, so for the most part I am going to try to refrain from doling out advice unless specifically asked as in the above. Thanks for reading! 🙂
That makes sense. In this case it’s a friend’s dog and I recently realized that with my dog I have a pretty specific “training is done” ritual that encourages him to go shake off the brain work and do his own thing. It is based on a vigorous rubdown and snuggle which is great for my dog but doesn’t work for the friend’s dog who has some touch sensitivity. I was trying to figure out what the best ritual was for this particular dog and not getting to anything satisfying or workable yet. I do think that a game of find it would be great as he’s an amazing nose worker and, of course, the frozen kong is also good for him. Thanks for the direction!
Just LOVE your “training is done” ritual with your dog! (I may have to borrow that one)! Good luck with your friend’s dog!
Interesting article. I have a rescued rottie, who was abandoned by her family. She is very much in need of active acceptance from me, to whom she became attached when we rescued her, a bit less so of my partner, who was there from the start too. She would jump to the point of launching herself at my chest/face. The vet says she shows signs of hyper-vigilence and PTSD, perhaps from past mistreatment/abuse.
Extinction was practiced diligently for the first 3/4 year she has been with us, but it only distressed her – increased efforts of jumping for attention, which of course still met with rejection, and then submission posture, only to spring back up if I now responded to her. But then we began the alternative approach, the positive alternative behavior of sit, and praise when this alternate behavior was shown. This has been practiced now for 1/2 year, and we have a happy dog, who still wants to jump but readily does the sit because she gets attention and praise. Even her “brother” now does the sit when we get home, even though he was not a jumper – he adopted behavior that he saw getting praise for the female!
Hi Polly – Thanks for sharing your experiences. It sounds like your girl has found a wonderful forever home with you. Love that you had success with her, as well as the great example of observational learning in her brother!
I have three major comments, for whatever they’re worth. Interesting article and thanks for posting about it.
1. Teaching an incompatible response to jumping (such as sit for greeting) doesn’t necessarily have to omit ignoring the behavior one is trying to extinguish. It is possible to do both during a training session, going for the “not this but that” route to better greeting behavior as long as one is careful not to inadvertently teach a two-step chain.
2. You (well, the researchers, but you summarized) list the dogs’ behavioral evidence of frustration as withdrawing from the trainer, lying down, increased movement, whining, sniffing and avoidance. I’m curious as to what behaviors you would have rather seen. A dog who loses interest and goes to do something else sounds pretty normal to me.
3. This whole frustration thing is problematic. We’re all being pretty anthropomorphic here, but learning can be a frustrating process, even when one is really enjoying what they’re learning. Learning to puzzle out what is going on in a time of frustration is probably a very valuable task as far as learning self-control goes. I question the value of attempting to remove minor frustration from the equation.
Hi Kristina – Thanks for your comments and for following the blog. I think the work that you are doing with Dog Science Inc. is exciting and important and look forward to learning more about your studies and work! In response to your comments:
1. Agreed; with just the caveat that the training bit precede the ignoring bit and that the alternate behavior be readily achievable by the dog, if only for those behaviors in which access to and interaction with a person is the reinforcing stimulus.
2. True, context is important. However, avoidance certainly can suggest an aversive response, especially in the context of having previously had a very positive interaction with the trainer, as it had in this study.
3. I am not sure I agree that discussing “this whole frustration thing is problematic”, nor that being anthropomorphic is a legitimate argument. I discussed the issue of non-human animals and emotions in a discussion with another writer in a previous blog, so do not want to repeat it here. However, to summarize, there is a large body of research and philosophical argument that validate and support the existence of emotions in non-human animals, to the extent that the onus is now on the naysayers to disprove their existence (at the very least in mammals and birds) than it is upon supporters to prove their existence. (My previous answer, in the Steve Series, provides an entire list of books, if you are interested). Second, I think that we may have a semantics issue with the word “frustration”. The authors of the study used a set of behaviors that signify a moderate stress response in dogs to suggest frustration. I am not sure that being frustrated is compatible with an enjoyment of learning. Rather, being challenged or stimulated certainly can encourage learning. Frustration, however, is generally considered to be an unpleasant emotion and one that may inhibit rather than aid learning. So, I would disagree with this comment, both in your use of the rather pejorative term anthropomorphize and in your argument that frustration is conducive to learning or to a positive human-dog training interaction.
But hey, two outta three ain’t bad, is it? 🙂 Thanks again for your thoughtful input!
Just to clarify, I consider anthropomorphism really helpful at times and I do not mean to use it as a pejorative at all. Further, when questioning the behaviors, I’m really trying to ask what a better response would be, because it seems to me that those behaviors are equally likely to indicate disinterest. So seriously, what behaviors would one rather see in that situation (assuming we’d rather the dog not be frustrated which is probably a teaching-inhibition issue we can save discussing for a later date, :-))?
Ahhh…..now I get it! I was not following your train of thought, I think. So, really, what you are asking is what behaviors would signify no stress, simply learning from the application of extinction? I think it is a great point, since extinction, by definition, means that you are trying to measure “absence of behavior” (which in itself can be difficult, to be sure). So, walking away, if not associated with other signs of stress, would signify a successful use of extinction and how does one differentiate between that and what might be frustration or stress. (Sorry, I am a bit slow on the uptake, sometime! :)).
Your post made me (personally) revisit extinction as a method once again, more in terms of my own comfort level than anything else. Warning – this is my personal opinion as a trainer, not data: Given that so many dogs who we as trainers and behaviorists see are living in homes in which they are under-stimulated, under-exercised and untrained, I think that I have evolved to a place that using any method that gives owners permission, if you will, to further ignore their dog is difficult for me (personally) to use. Unlike what is often reported in dog training books and even in some professional materials, many “attention-seeking behaviors” that are classified as problematic may actually be legitimate bids for attention and affection by a dog whose needs are not being met. While this is certainly not always the case, in general, I tilt towards methods that encourage more, not less, interaction between owners and their dogs, so this is probably why I personally do not promote the use of extinction. Still, your point about “walking away as simple disinterest” is a good one. Thanks for sharing and I look forward to continued discussions with you!
Just wanted to say quickly that I fully agree with that assessment of the state of many dog-inhabited households today. I think that is where the art of the science of dog training comes into play. It would be a lot easier if we were training dogs absent all these other circumstances that have to be taken into consideration such as the owner’s motivation and skill level, etc!