The behaviour interpretation thread

    • Gold Top Dog

    Thanks to 4IC, I've overcome a technical stumbling block. I was able to convert from AVI to WMV format and finally got our training video uploaded to Photobucket. I entitled it offleash, as we were off-leash in the backyard, with the distractions of the neighbors dogs. One dog is a terrier hound mix and the other is a Mini Schnauzer.

    This is without the clicker, as that has been faded out of this, though I should have used it just to show my sense of timing. Oh well, that could be an excuse to film another video.

    ETA:

    You may have to turn your sound up. The mic on the camera is not the greatest. Also, please bear with me filming one-handed. On the first heel, I got all thumbs going and botched filming it. So, I did it again, focused on my left hand. Watch for the touch.

     

    • Gold Top Dog

    Hee Hee, the sound of the neighbor's dog barking brought all my dogs to the PC.  What's even funnier is Paganini did a sit and a down from your voice commands.

    Shadow is a handsome dog and this video really shows that.  A question for you.  How long did it take to train?  I see you are usingl food treats for basic obedience command and I was wondering if this video is the early stages of training. 

    • Gold Top Dog

    I enjoyed the video.....and Shadow is a beauty......

    That being said, I am a little confused, unless what DPU mentioned , the early stages of training, I thought the food reward gets phased out when all commands are tight.......could you help me with that?

    • Gold Top Dog

    Oh, How pretty is Shadow! I know how hard it is to film one handed! LOL And the heel was very impressive. He looks so happy to sit (and a pretty sit, at that) and down. Very nice. Good boy! He also seems unbothered by the crazy neighbor dogs! LOL  Glad you got it uploaded. Look forward to more!

    • Gold Top Dog

    snownose
    I thought the food reward gets phased out when all commands are tight.......could you help me with that?

     

    Agreed, is Shadow not mastering the sit command yet? 

    • Gold Top Dog

    I will answer everyone's question in one post instead of 3 or 4. I used treats to show my timing. And the proper way to give treats. Toss them so that coming to your hand is not part of the actual cued behavior. And actually, I've done all these behaviors without clicker or treat in hand before I made this video. Shadow knew sit before I got him. But it was the only thing he knew in 2004. That and shake paws.

    Also, the value in my botching the first heel in filming was to be able to show variable time between marker and treat. On the second heel, I didn't have any treats on me but I marked the successful position. When I got back to the shelf on the side of the grill, I was able to give him treats for that.

    I had thought of taking a step back and trying to film with the clicker to show my timing on that but I think you guys would rather see the behavior proofed. Until I get that filmed, you will just have to imagine that it was like this video, but without treats.

    I might have to edit this post if I go back and read something else I forgot.

    ETA: Oh yeah, DPU. When I play this video, Shadow thinks the Mini Schnauzer is out there. That's the one with the high pitched bark. I could be a devil and play this video whenever I want him to go outside.

    ETA II:

    As for how long it took to train, I didn't actually keep track of time, and I should have. That's a very valid question, especially, say, to someone who fosters and doesn't have a lot of time. Me, I've got Shadow's lifetime.

    The initial heel took one evening. 3 clicks for touch, 3 clicks for trying to do a distance touch but Shadow kept following me on my left side where I had the treat. The next night, I mentioned it, thinking I made a mistake and someone pointed out that it sounded like the beginning of a heel. So, I had a lightbulb moment, too. Anyway, the second night, I practiced more of the touch while walking. After about 3 clicks, I started calling it heel and the name stuck, as you can see in the video. Then, I practiced in the yard. Later, I practiced in public, on a regular walk, and in the pet  supply stores, such as Petco, Petsmart, and Canine Comissary. The next dog fair event we can go to, I will practice it there, likely with the clicker. That is because, while I have heel in the house, yard, public walk, stores, we haven't done it at a dog event. Dogs don't always generalize the way we think they will and heel at a dog event is a new and separate behavior. So, out comes the clicker for establishing a new behavior.

    Each time I worked on heel, it was only maybe 3 or 4 repetitions at the time and training sessions never lasted more than 10, maybe 15 minutes. I guess the ability to have heel in other circumstances depends on the availability of time and opportunity. But if it were elapsed time on one strong clicker behavior, such as heel, it might be averaged to about 42 minutes, including public outings. But it was spread out over 6 to 7 months. ( the secret to my math. Estimating 2 minutes of heel in a session. Worked on that twice a week, sometimes in public. At about 7 months when I quit dragging the clicker with me. And not always having treats. I may have overestimated the amount of time. What was more important to him was jackpots. To this day, without a command, treat, or clicker, when we walk into Petsmart, Shadow assumes heel position.) And really, I only needed the clicker the first time in a new setting to mark the success in that new setting. Each new setting can be a new behavior to a dog. If I can think of something else, I will eta again, unless someone responds with a new post and then try to answer then.

    ETA III:

    And I don't mind the questions. If Carla can take the heat, so can I.

     

    And thanks, Carla.

     

    • Gold Top Dog

    Seeing as this is the topic with no particular topic, I thought this would be a good place to illustrate a concept seen before my very own eyes.

    Some people here, such as in the prong thread, have dogs that seemingly are impervious to pain or what some of us think are punishments. To the dog, it might simply feel like a touch cue that gets their attention better than anything. Other dogs might find it painful. I think the same dog can have different reactions to physical stimulus.

    Case in point. A little while ago, I went outside to feed Shadow. If it's anywhere below 70 F, he wants to eat outside. And he likes to chase the kong once or twice before eating. Sibe appetite is geared up by exertion. So, anyway, I step outside and he starts to take off and I throw the kong, aiming ahead of him. And nail him squarely on the left side of the muzzle. No muss, no fuss, no harm , no foul. He keeps right on chasing and gets it on a bounce and comes back for his meal. Now, it could also be context. This is play, as opposed to me walking up and deliberately bonking him on the snout for something. Nor did I shower him with sorry and affection and poor baby. It was an oops. He just saw it as part of the hard style of play that he likes. Also, as I have pointed out before, with him, most types of physical punishment we could even think of doing usually feel like play.

    Anyway, I can see how a dog that responds to something physical is helped by that, as opposed to something cerebral. But, in the same breath, it may not be punishment to the dog, no matter how many times we call it a correction. In fact, with some of the descriptions of others' here in their use of correction, it sounds more like physical cues. Which doesn't stop me from trying to train operantly.

    IOW, I'm willing to bet I executed that forward pass to the muzzle with more impact force than most of you apply with your leash, or fingers, or whathaveyou. It was an accident and I wouldn't be able to do that again if I tried.

     

    • Gold Top Dog

    Ron, the reason for my asking about the food reward was in reference to a finished training protocol.

    I have made the mistake before and jumped to conclusion by judging a certain poster's videos as finished, as was not the case, it was training in progress. I would really love to see the finished training session where everything is tight and the dog does not need food reinforcement.

    I know, that would mean you would have to go through all the hoopla of recording and uploading, but I am sure I am not the only one who would love to see it. When I was teaching Ronin the recall outside a fenced area I was not taping in the beginning.... that was actually funny.....he found better things to look at than come running to me.....lol

    • Gold Top Dog

    ron2
    Anyway, I can see how a dog that responds to something physical is helped by that, as opposed to something cerebral. But, in the same breath, it may not be punishment to the dog, no matter how many times we call it a correction. In fact, with some of the descriptions of others' here in their use of correction, it sounds more like physical cues. Which doesn't stop me from trying to train operantly.

     

    We all train operantly. Whether we're very good at it or not is another question.

    Secondly, I do not agree that a punishment has to be harsh enough to extinguish the unwanted behavior in 2 or 3 applications. That's just something somebody said and many in the dog-training world have bought into it. I have not. A punishment decreases the likelihood of the behavior in the future. That's all. It may take 100 applications before the behavior is gone, but each application decreases the likelihood.

    So, you may not call a particular action a "punishment or correction" and I do. It's because we mean something different by the word. I personally tend to think that the +R community insists that a punishment has to be impactful enough to extinguish the behavior in 2 or 3 applications just so people will be less inclined to use punishment at all and start looking for other ways to distract, redirect or otherwise divert their dogs from performing unwanted behavior.  

    • Gold Top Dog

    FourIsCompany
    Secondly, I do not agree that a punishment has to be harsh enough to extinguish the unwanted behavior in 2 or 3 applications. That's just something somebody said and many in the dog-training world have bought into it. I have not. A punishment decreases the likelihood of the behavior in the future. That's all. It may take 100 applications before the behavior is gone, but each application decreases the likelihood.

    Interesting opinion.

    Here's a source about the how effective positive punishment should be.

    http://www.wagntrain.com/OC/Part2.htm

     Now, your saying that the idea that a punishment should stop a behavior in 2 to 3 applications is wrong or inaccurate. I think one to two, and the wagon train source says it should stop the behavior dead in its tracks and the aversive must be greater than the reward. IMHO, if you are doing 100 corrections and the behavior finally stops at the 101st chance, it is a -R principle. That is, the dog is changing behavior to avoid the aversive.

    Let me try it this way. It is not uncommon for a dog to grasp a concept in about 3 clicks. If we apply the same  limit of applications of effectiveness to punishment, then it should be no more than three applications of the punishment to stop it. Anything after that is a nuisance, nothing at all, or eventually, a -R aversive.

    So, how would you define how many applications of punishment are necessary to stop a behavior. And you are right, I may view what you do as more of a touch cue while you view it as a correction. And neither one of us matters, in our opinion. It's the dog's "opinion" that counts because it is the dog that decides what is aversive enough to stop the first time, what is aversive enough to stop after a while, and what is ineffective at any time. Or what is a "pay attention to me" cue, whether that is the phrase "watch me" or a tap on the neck.

    Another good example is say that the dog is chasing the cat. As the dog passes by, you pat him on the head, thinking that your touch is calming. The dog stops chasing the cat after a few times. It is likely that your pat on the head had the effect of a positive punishment. It stopped the behavior. Or it was a cue to watch you. You might poke the dog, thinking that you are correcting, when the dog simply interprets it as an interruptor or "watch me" cue.

     

    Case in point. My supervisor was talking about getting in trouble with other kids in school and getting beat up. And the punishment of getting beat up did not stop him from being a smart-aleck. That is, while getting beat up hurt and would certainly qualify as an aversive in anyone's book, it didn't stop him from doing whatever it was, and therefore, was not a punishment.

    Let me be a twerp and offer your solution of the studded collar as an environmental +P. As the behavior decreased, I think it became -R. The dog is reinforced to not neck bite the other dog by the absence of feeling the studded collar. And did it take a 100 tries?

    FourIsCompany
    I personally tend to think that the +R community insists that a punishment has to be impactful enough to extinguish the behavior in 2 or 3 applications just so people will be less inclined to use punishment at all and start looking for other ways to distract, redirect or otherwise divert their dogs from performing unwanted behavior.  

    I would also disagree with this notion. I think a lot of positive trainers are simply about what is most effective. And I don't think it's wrong or deceitful or wishy-washy to seek the most gentle and motivating technique first. It's not necessary to start out punishing and "dominating."

    • Gold Top Dog

    ron2
    Here's a source about the how effective positive punishment should be.

     

    Yeah, according to whom? The fact is that positive punishment is defined as reducing the likelihood of the behavior. It doesn't say by how far. And from that source:


    If you find you have to apply a punishment more than three times for one behavior, without any decrease in the behavior, you are not "reducing the behavior", you are harassing (or abusing) the trainee.

     

    That supports my suspicion that a punishment that isn't effective after 3 times is considered abuse (by some). Also, it says without any decrease in the behavior.

    Don't get me wrong, if I use a punishment that I believe causes discomfort or pain, I'm not going to use it for long. If it doesn't work, I'm going to change my strategy to something that does work. But I think putting a number on it and calling it "abuse" is propaganda. Because I might use a punishment for 2 weeks or 12 times and it finally works. And I don't buy that I've been abusing my dog.

    ron2
    I think a lot of positive trainers are simply about what is most effective. And I don't think it's wrong or deceitful or wishy-washy to seek the most gentle and motivating technique first. It's not necessary to start out punishing and "dominating."

     

    I said absolutely nothing that contradict this so I'm not sure what you're disagreeing with...  

    • Gold Top Dog

    Whoo, do I get to play devil's advocate here?

    I actually agree in part with both of you, 4IC and Ron, if that's possible. So perhaps I'm playing devil's advocate for both....muaha.

    I've seen a lot about the "punishment must be severe enough the first time, and no more than 3 times, or else it's not effective/is abuse. I've likely even said it a few times, but I'm beginning to question it myself as I grow and begin to think about it. Not that I'm any more apt to use it, but because I'm a critical thinker and like to question everything I read *G*

    Firstly, the overall basis of the behavioural definition, as I've learned it in university in learning theory, is pretty much as 4IC states. There is nothing in the definition that states that behaviour will ever actually *disappear* (that's just the hope of dog trainers and dog parents!), or that it has to do so quickly in order to be considered punishment. Technically, any decrease in behaviour is considered punishment.

    Then I look at the realism of using either reinforcement or punishment. Both are required throughout life in order to maintain behaviour. Behaviours that are not reinforced somehow, and that are not self-reinforcing, will eventually become extinct. In other words, even intermittent reinforcement is required through the life of the dog, in order to maintain behaviour.

    I think the same goes for punishment, to be honest. Dogs who are trained using mostly punishments (I'm not pointing anyone out, but we all know somebody who has taught this way), or even just some punishments, tend to periodically need punishments throughout their lives as well to maintain the behaviours that they know (which includes "not doing something";).

    For instance, the dog taught to heel, by being collar popped with a choke chain every time it is out of position (I'm using this example because I happen to know someone who teaches it this very way, so it comes to mind quickly), and the dog learns to avoid punishment, it stays in the heel position. Anyhow, that dog, too, will periodically require punishment throughout its life, in order to maintain the behaviour of staying in heel. It may be rare, but it's there, just as reinforcement of some sort is required to maintain behaviour. Without some contingency, the behaviour will not continue and the dog's heel would become more sloppy.

    I also think that just like with reinforcement, once the context of a situation changes, the dog's understanding of the situation changes as well. So a dog who has been punished for jumping on people in the home, might not bring that out doors to say, a park, and thus will need to experience the punishment there as well. When we say dogs tend to be poor generalizers, I think it applies as much to punishment as it does to reinforcement. Dogs need to be "proofed" with punishments just as they do with reinforcements, so that in effect would require more than 2 punishments, as the context shifts.

    I do think that the author of this article is correct in saying, though, that there is an inherent downside to using escalating punishments, and I think that is of more importance that whether it stops it in 2 or 3 tries. In other words, if you are going to do something to lessen behaviour, make sure the dog actually cares about what you are doing, rather than starting out small and going harder, and harder, and harder. And I do agree that via a cost/benefit analysis, if you are to decide to use a punisher, it does have to outweight the value of the reward. That might seem obvious to some, but in order for a punishment to be effective, it has to have a higher cost/benefit ratio than the reward does, because if not then the punishment certainly won't be effective long-term.

    I have a sneaky suspicion that along with a variation in natural pain sensitivities, a lot of these dogs you see lunging in prongs or pulling in chokes, are dogs who have experienced either of the following:
    1) Slowly escalating punishments such that the dog becomes habituated to even some of the most forceful punishment attempts. In fact, I believe at one point this is precisely how they trained protection/police dogs, so that they would not be bothered when people beat the dogs or hit the dogs, or try to hurt the dogs by force - the dogs had been habituated to force, and as such didn't take it as a punishment. It's also how a lot of dog fighters train, purposely, dog fighting dogs so that they will continue on in the most painful situations.

    2) A dog for whom the cost/benefit ratio is higher for the reward than it is for the punishing effects of the prong/choke (these are just examples).

    Even with my example with Shimmer and pawing, for her the cost/benefit ratio of pawing was greater than for the temporary removal of attention, so it was very slow to change. But in ignoring the pawing, and then heavily reinforcing the "sitting without pawing", the cost/benefit analysis of "sitting without pawing" become higher than the pawing, so it became the overall behaviour.

    It's like I've seen with one lady working with her dog.....the punishments she was actually using was rather benign, such that it's not the intention of the woman that resulted in the ending of a behaviour. What she thought she was doing, was not seen as a punisher to the dog. Rather, it was the constant nagging, the sheer annoyance of the woman, that actually ended up acting as the punisher....lol. The woman didn't do anything inherently aversive, but the fact that she continually nagged over a certain behaviour, caused the dog to lessen the behaviour overall, and the cost of doing the behaviour became higher than the benefit from doing it.

    It's like a child learning to stop leaving things around  *G*. The nagging mother tends to win out in the end, not by being inherently aversive, or doing anything painful/fearful to the child, but simply by becoming an annoyance to the child such that the child learns to stop leaving things around to avoid the nagging of the mother. Some children (and some dogs) handle nagging better/worse than others, which is why it is not usually an effective strategy to recommend as a whole!

    I think the issue with "2 or 3 punishments" is that if you have to give more than 2 or 3 decent punishments (decent, to mean effective) in a row, for a given behaviour, you have a problem with the type of punishment that is being used. Or if you have to use the punishment every day, in and out, there's a problem with the punishment. I'd even wager a guess, to stretch it, to say that if you don't notice some sort of meaningful decrease in behaviour after 2 or 3 tries, then it is not a very effective punishment. That might be a better way to put it (or it might not, I'm still deciding.....hah)

    It's the same way with if I had to give a cookie every time my dog sat, and the dog wouldn't sit without a cookie, I'd have a problem with my level of reinforcement and the teaching. And that I can understand. There is no reason that anybody should be giving 5, 10, 20 punishments for one behaviour during a given situation, no more than there is no reason that once a behaviour is learned, I should be giving treats every time my dog performs a behaviour that I ask (although I rarely ask for a behaviour without it having some contingency, usually a life reward, - it's a way I live my life with dogs, so that they always have the option to make choices). Perhaps in 3 hours if the behaviour crops up again, punishment may be required then, but again at that given time, it should be effective enough within a few uses, to have an effect without becoming simply an unnecessary aversive.

    • Gold Top Dog

    Great post, Kim. I really don't have anything else to say. Oh, except that B'asia paws, too! LOL  

    • Gold Top Dog

    I'll try a different way, then. There may be times when a punishment is deemed necessary. -P, +P, even some -R.

    I simply don't think there is an actual ulterior motive amongst positive trainers to, as you might put, place such a heavy burden of proof on punishment so that people would rather try +R instead. I could be wrong, but I don't see it, as yet. And it seems like a handy bit of conspiracy theory just to keep justifying the use of punishment. IMO, and I could be wrong.

    So, you've rejected the one source I have listed so far in how fast and effective a punishment should be, supplanting your own number, for instance, 100 corrections.

    I will agree with Kim's notion that just as variable R will be necessary for behaviors that are not self-rewarding, variable P may also be necessary for behaviors that are not self-extinguishing. And I thought that was rather fair and in line with my supposition that if a behavior can be reinforced in 3 clicks, it should be able to be stopped in 3 clicks. I haven't read that somewhere, it's just my sense of "balance." I figure if you can reject the source I quoted and come up with your own number, so can I. Fair is fair, even if we always disagree on this point. Different styles and viewpoints.

    • Gold Top Dog

    ron2
    I simply don't think there is an actual ulterior motive amongst positive trainers to, as you might put, place such a heavy burden of proof on punishment so that people would rather try +R instead.

     

    I'll give you that. I could be just over-sensitive to it all. I am a conspiracy theorist, after all. Wink Really.

    ron2
    So, you've rejected the one source I have listed so far in how fast and effective a punishment should be, supplanting your own number, for instance, 100 corrections.

     

    Ron, I didn't reject your source. In fact I've used that source before myself. But please read carefully what it says.

    If you find you have to apply a punishment more than three times for one behavior, without any decrease in the behavior, you are not "reducing the behavior", you are harassing (or abusing) the trainee.

    Without ANY decrease in the behavior. So I agree with you. If you do something 3 times or 50 times without ANY decrease in the behavior, it's not punishment because punishment requires a decrease in behavior. It doesn't matter how many times you do it. If you're not decreasing the likelihood of the behavior, it's not punishment. I agree with your source.

    I just think that if the behavior is decreased after 3 times, then it's effective punishment. If it's decreased after 12 times, it's ALSO effective punishment. If 3 is your number, that's fine. I don't have a solid number, but I know that if I use punishment and it doesn't seem to be working, I find another way to deal with it.

    I don't think we disagree much at all.