motivation and clicker training

    • Gold Top Dog

    motivation and clicker training

    The general chat thread was getting too confusing for me, so I thought I'd start a new one to discuss motivation and how to use it in clicker (or any other method) training. What motivates dogs to act the way you want them to? Here is my take on motivating and training dogs:

    Dogs do what works. If they do something, and it results in something good for them, they are more likely to do it again. If it results in something unpleasant or neutral, they are less likely to do it again.

    Before training your dog, observe your dog, and make a list of what the dog REALLY likes and what he sort of likes and what he doesn't like. Your own opinions do not matter. Some dogs really dislike being petted at certain times, and it doesn't matter TO THE DOG if you think the dog "should" like being petted at all times. Some dogs really like to eat, some enjoy their work to the point of not needing other motivators, others really like to play tug, others really like squeaky toys, others, well, maybe they seem to only like to bark, or chase squirrels, or ride in the car. Never fear, once you identify ONE thing the dog really likes you can use it to train the dog.

    You can and should build "secondary motivators". The more tools you have, the better. If your dog appears to have no interest in praise or food but loves to chase squirrels, you can LINK praise to squirrels in the dog's mind: go out to where the squirrels are, praise your dog, and release to chase. If you repeat often enough the dog will become classically conditioned to experience pleasurable anticipation when he is praised. Secondary motivators are never as strong as primary motivators, however.

    On to clicker training: clicker training is simple but not easy. In the long run it is the fastest way to train a dog to do anything; however, it may seem slow at first because you have to prepare the dog by making the dog "operant". An operant dog is a dog who happily experiments and "throws behaviors" at you until the dog hits upon one you want today, and then you click (tell the dog which one you want) and offer your motivator (to increase the likelihood of the dog offering that behavior again). If you take a dog out into an area with various things lying around, it's obvious whether your dog is operant or not: the operant dog runs around offering all sorts of behaviors, glancing at you all the time to see which one is going to be rewarded today. The non-operant dog ignores you and runs around sniffing, or sits there waiting for orders. The more behaviors you have rewarded, the more your dog will offer. This may take time and skill on your part- watching the dog and clicking and rewarding for all kinds of apparently useless behaviors like lifting a hind leg, or flicking an ear, or sticking a head into a bucket. It takes time to build a dog to this level.

    It is impossible to use so-called "balanced" training with most (if not all) dogs and still get an operant dog. So-called "balanced training" that incorporates significant amounts of positive punishment into the training suppresses behaviors. Particularly puppies and "soft" dogs and "non-working breed" dogs- any kind of positive punishment wipes out these dogs willingness to experiment, why bother, they might get punished. A lot of people think these suppressed, subdued dogs are "THE IDEAL DOG" because they don't do much of anything unless ordered to do so. Very sad, in my opinion.

    • Gold Top Dog

    mudpuppy

    Before training your dog, observe your dog, and make a list of what the dog REALLY likes and what he sort of likes and what he doesn't like. Your own opinions do not matter. Some dogs really dislike being petted at certain times, and it doesn't matter TO THE DOG if you think the dog "should" like being petted at all times. Some dogs really like to eat, some enjoy their work to the point of not needing other motivators, others really like to play tug, others really like squeaky toys, others, well, maybe they seem to only like to bark, or chase squirrels, or ride in the car. Never fear, once you identify ONE thing the dog really likes you can use it to train the dog.

    Here's my comment on the above, consider it because it is based on me observing my dogs.  One most certainly can create such a list and be very objective at that time and in that moment.  But, once one determine the "ONE thing", then you are actually doing the opposite of what you want to accomplish.  What motivates a dog, be it food, toy object, or just plain old companionship naturally changes over time.  They are all prized but their values change because of the nature of deprivation and satiation.  And time could be hour to hour, day to day, week to week, month to month.  Once the human has selected that "ONE thing" that locks the motivator into the number one spot and uses it frequently, the social activity and affection then get attached also, thus further strengthening the motivator.  I think since there is no way to stop the human influence on motivators, that social rewards (social activity and affection) should be the "ONE thing" and the secondary motivators should be what the dog has selected at the time when training is done.

    • Gold Top Dog

    Excellent post, mp, and I couldn't agree more.

     I'm wondering what your thoughts are regarding a primary motivator being needed at first to teach a new behavior partially because for the dog thinking and working like that is hard work and needs a highly valued reward. Later, once the behavior is learned, it's really not very difficult (for most behaviors--recalling off prey and the like will always be difficult for most dogs) anymore so a lower level reinforcer (for a relatively aloof dog like Marlowe, praise and attention is a pretty low level motivator, but works for him if the behavior is already learned and easy to perform) is sufficient to keep the dog motivated.

    I'm also interested in if you think you can build something up to be a strong primary motivator if the dog does not come out of the box with that particular drive. A drive to play with toys is something that I've been working on and off at developing in both dogs and neither of them seem to really "get it" yet. Even very intriguing toys at first may seem interesting and get a lot of play, but pretty quickly they get boring again and are ignored, no matter how much I shake them around and run around with them and toss them about.  I'm intellectually interested in Natural Dog Training and play training but I can't actually try many of those techniques out because I have dogs who don't give a hoot about toys.

    • Gold Top Dog

    Nice post mudpuppy. I have a few comments, though. You knew I would. Smile

    mudpuppy
    Dogs do what works.

     

    I always find this amusing because... don't we all? Of course dogs do what works. Everyone does what works.  

    mudpuppy
    you have to prepare the dog by making the dog "operant". An operant dog is a dog who happily experiments and "throws behaviors" at you until the dog hits upon one you want

     

    I'm certainly not arguing with the fact that it takes preparation, but the term "operant dog" seems inaccurate. I have never heard this particular term before, but it seems to me that all dogs are operant... In other words, all dogs (as well as other animals AND people) respond to operant conditioning. Using this phrase "operant dog" is a bit misleading in this context, in my opinion. Wouldn't it be more accurate to call them a conditioned dog? Or even more accurately, a "clicker-conditioned dog"? Learning through consequences makes an operant dog. And they come to us like that. I think the use of the term is confusing. It attempts to make operant conditioning synonymous with +R.

    mudpuppy
    It is impossible to use so-called "balanced" training with most (if not all) dogs and still get an operant dog. So-called "balanced training" that incorporates significant amounts of positive punishment into the training suppresses behaviors.

     

    Balanced training does not necessarily incorporate "significant amounts" +P. I agree that "significant amounts" +P would tend to suppress the dog trying new behaviors, but connecting balanced training to "significant amounts" +P is a mistake.

    houndlove
    I'm intellectually interested in Natural Dog Training and play training but I can't actually try many of those techniques out because I have dogs who don't give a hoot about toys.

    B'asia wasn't interested in toys until I tried a Flirt Pole. She LOVES it!   

    • Gold Top Dog
    mudpuppy

    On to clicker training: clicker training is simple but not easy. In the long run it is the fastest way to train a dog to do anything

    Can you provide evidence to support that statement? I have achieved very fast results with a number of training methods.
    mudpuppy

    It is impossible to use so-called "balanced" training with most (if not all) dogs and still get an operant dog. So-called "balanced training" that incorporates significant amounts of positive punishment into the training suppresses behaviors. Particularly puppies and "soft" dogs and "non-working breed" dogs- any kind of positive punishment wipes out these dogs willingness to experiment, why bother, they might get punished. A lot of people think these suppressed, subdued dogs are "THE IDEAL DOG" because they don't do much of anything unless ordered to do so. Very sad, in my opinion.

    Not all balanced training incorporates significant amounts of P+. Some incorporates very little, some clicker trainers incorporate more P+ than balanced trainers, some P+ suppresses the offering of behaviors, some simply are used to extinguish undesirable behaviors faster than would be done with P-. I have had several dogs that were trained with balanced methods, who had no problems offering a multitude of behaviors when they had no "orders". I have had a number of dogs who would be down right dangerous if they hadn't received P+.
    • Gold Top Dog

     Thanks for the flirt pole idea, Carla. I think that it really be good for Conrad as he is really really stimulated by movement. Marlowe, well, we'll see! I'll report back!

    I recall reading somewhere or other that saying a dog is "operant" may be kind of a misnomer. I can't remember where I read that though but it was some really respected clicker training source (maybe one of the articles or blog posts on Karen Pryor's site?). What people mean when they say that I think is that the animal is a fully engaged and willing participant in their own training, and that they are coming up with their own behaviors in an active, experimental manner. So the dog who just sits and stares at you waiting for a command is waiting to be given orders, an "operant dog" is a dog who is actively soliciting the trainer for feedback by trying different things and seeing what happens. The sitting and staring dog has been operantly conditioned in the past to sit and stare until an order is given. I'd maybe make the distinction between the two dogs as one being non-creative and the other being creative. The former has been conditioned to believe that creativity will be punished and thus gives being creative up as a strategy. The creative dog has been conditioned to believe that being creative will be rewarded, and actually gets more and more creative the longer creativity continues to be rewarded. Karen Pryor studied this process in porpoises in the 1960's in an expeirment she called "The Creative Porpoise."

    • Gold Top Dog

    mudpuppy

    An operant dog is a dog who happily experiments and "throws behaviors" at you until the dog hits upon one you want today, and then you click (tell the dog which one you want) and offer your motivator (to increase the likelihood of the dog offering that behavior again). If you take a dog out into an area with various things lying around, it's obvious whether your dog is operant or not: the operant dog runs around offering all sorts of behaviors, glancing at you all the time to see which one is going to be rewarded today. The non-operant dog ignores you and runs around sniffing, or sits there waiting for orders.

    I don't want a dog who always throws behaviors at me. I'm not that high maintenance, and I think it is healthy and good for a dog to be able to explore its environment if not under any commands. I want a dog who's default it relaxed and doggy, with certain behavior restrictions (no peeing in the house or eating the furniture). You may call it a shut down dog. Call it whatever you want. There is a reason I'm not interested in owning a border collie.  

    I

    It is impossible to use so-called "balanced" training with most (if not all) dogs and still get an operant dog. So-called "balanced training" that incorporates significant amounts of positive punishment into the training suppresses behaviors. Particularly puppies and "soft" dogs and "non-working breed" dogs- any kind of positive punishment wipes out these dogs willingness to experiment, why bother, they might get punished. A lot of people think these suppressed, subdued dogs are "THE IDEAL DOG" because they don't do much of anything unless ordered to do so. Very sad, in my opinion.
     

    I'm not so sure. I don't do much clicker training, but if I pull out a clicker Sasha switches into offer-behavior mode. That kind of thinking, planning, anticipatory stuff. She enjoys it, and we use it as a kind of game. But she is also independent, and enjoys plenty of time to do her own thing. I think it is sad for a dog to be so entirely consumed by getting rewarded from a person that the dog can't ever relax.  If I put Sasha in a room full of toys, she'd sniff things but not maul the toys unless one was handed to her. She's been "shut down" from chewing man made objects that aren't presented. Put her in that room and pull out a clicker, and she'll start hopping about looking for the behavior that will cue a click and treat. I don't want to misunderstand what you're saying, mudpuppy. Do you desire a dog who automatically looks for behavior/click/treat, all of the time?

    I use pretty mild positive punishment. No's and eh's. Nothing scary and bad. I communicate "don't do that." If positive punishment were so poisonous, no dog would be able to be "operant" since mama dogs use P+ to stop unwanted puppy behaviors. If you scare the unholy crap out of a dog, you might kill its desire to do much, but mild punishments don't ruin dogs.

    Eko is 9 weeks old, and gets a lot of eh!'s. Mostly for chewing or mouthing the million things he shouldn't. I swear he's half shark.  Stick out tongue But if I pull out one piece of kibble, he immediately starts thinking, and going through the admittedly limited number of behaviors that have been rewarded. He will also come in to the kitchen and randomly throw good behavior at me in the hopes of getting a snack. I haven't used the clicker with him, but I've held a piece of kibble until he did something I liked, and then rewarded with a good boy and the kibble. 

    • Gold Top Dog

    houndlove
    I recall reading somewhere or other that saying a dog is "operant" may be kind of a misnomer.

     

    Oh, that's good. I won't be seen as being "difficult" then. LOL It's just really misleading, I think. Does anyone know what Karen Pryor calls a dog who is "clicker conditioned"?

    houndlove
    The former has been conditioned to believe that creativity will be punished and thus gives being creative up as a strategy.

     

    Again, even though I do rarely use the clicker, most of my training is without it and my dogs are WAY creative. In fact, this morning, B'asia offered to bring me one of my house shoes... And even though I let her know that that was unacceptable, she created a way to bring me the other one. Confused

     

    • Gold Top Dog

    Carla, I'd like to respond to your comment on a dog being an "operant dog".  My understanding is that not all dogs are truly operant.  In fact, most dogs I've seen aren't really that operant at all.  They will do behaviors that they know, but they won't try anything new.  Kenya is like this.  If I stand in front of her with something she wants, she will sit, lay down, sit again, give paw, do a left finish, lay down....but those are all things I've trained her to do.  I have trouble getting her to try something totally new, like spin in a circle, pick something up in her mouth, etc.  The only really operant thing I've had success free shaping/capturing was that she would paw at certain objects on her own and we turned that into a "touch" command.  Generally though, she will offer a few behaviors she knows and then give me a blank stare or leave.

    Coke is slightly more operant in that he will shuffle his positions.  For example, if I say "sit" and he does a front sit but it is skewed, I withhold the reward.  He will shuffle his butt into various places until he finds for himself that the front sit means in line with me and only a foot away from me.  If I ask Kenya front a front sit and hers is skewed, she will never think to shuffle around until she gets the threat.  We have to start over with the behavior and see if there's better luck next time.

    A truly operant dog will never stop offering behaviors until the reward they want is earned, that's the way I understand it.  If your dog is operant, you can stand there with the highest value treat and think up a behavior, even something new, not command the dog, and they will eventually offer behaviors until they figure it out.  

    Neither of mine will do this, though Coke is closer.  However, he is far more food motivated.  I do think that how "operant" the dog is can also be a function of how badly the dog wants the reward! 

    • Gold Top Dog

    Thanks, Liesje. My confusion isn't in what it means. I'm clear on what it means. I just think using the term "operant dog" is a misnomer. "Operant conditioning" is a very specific thing and making a direct relationship between a dog who offers new behaviors and operant conditioning seems to me to be inaccurate. Yes, these dogs are responding to Operant Conditioning, but so is a dog who cowers in fear of punishment.

    I'd like to know what other professionals like Karen Pryor call a dog like this...

    • Gold Top Dog

    You don't need a clicker to have a creative dog, Carla. What you do need is to not punish for creativity, but to reward for it in whatever way you typically reward. And you can use a clicker and wind up with a dog who isn't really all that creative, if you don't reward for creativity and allow that to go extinct. If you only ever shape and lure and have a target behavior in mind every time you get your clicker out, you may have a very happy, enthusiastic, well-trained dog, but the dog may not be particularly creative. I have to keep reminding myself of this because I don't have that much free time to train and I always seem to have something that I'm trying to make happen and will shape and lure to get it, and that does have an impact on creativity in the end. Now that I have both the right tool (a clicker+ that makes separate sounds for each animal) and right behavior out of at least one of my two (a solid "go to your mat and stay there until released, no matter what is happening around you" from Marlowe--we're still working on it with Conrad but making progress), I think it will be a lot easier for me to just do some creativity work with the dogs because now I don't really have to find a way to separate them and keep them entertained while I work with the other.

    I keep an eye on the creativity angle because I am always at the back of my mind concerned about my dogs personal development. I know it sounds totally hokey and I may get laughed off this thread, but I control so much of their lives just by default in order to keep them safe in the human environment they live in. I want them to be fully realized, self-actualized beings, I guess. Lord. I swear I'm not this flakey in real life. But the Creative Porpoise experiments, when I first read about them (here) I felt like here is a way to help animals that are in our care develop themselves, just for the sake of giving them the tools to stretch their intellects.

     

    Edited: I hear the term "clicker savvy" used a lot to describe a dog who is wise to the ways of shaping and who offers novel behaviors when they know "the game is on".
     

    • Gold Top Dog
    mudpuppy

    A lot of people think these suppressed, subdued dogs are "THE IDEAL DOG" because they don't do much of anything unless ordered to do so. Very sad, in my opinion.

    The dog I have that doesn't do much of anything unless he is ordered to do so is far from suppressed. He simply has a temperament that leads him to worry, and that leads him to want to please me more than to want to experiment. For him, doing what's asked is one of his biggest motivators. Eventually I reward his sit and stare act, because he begins to get anxious about why he's not being rewarded, and then he figures "AHA!!! That's what you wanted -- a sit stay with focused attention." It's not because he's received corrections ~ he hardly ever receives corrections, and when he does, it's a "eh-eh". He simply is very handler sensitive. Used to be that I couldn't praise him while he did bitework. He loves doing bitework, but the least bit of a "good boy" from me would have him off the sleeve and at my side.
    • Gold Top Dog

     Great OP, mudpuppy.  And, Liesje, nice succinct response regarding the fact that many dogs are not truly operant.  When a dog is even intermittently punished with +P, the rate at which he is willing to take a chance on offering a new behavior is reduced.  After all, from his perspective, what if it's wrong?  Will he get punished?  That's one reason why severely force trained dogs seem so obedient - they only offer what they have been taught, and pretty much don't move otherwise.  They are the ones that don't respond as well to clicker training in the beginning - and it's why many clicker trainers suggest that crossover dogs get really strong food rewards associated with the clicker (tripe, roast beef, garlic chicken, cheese) so that the dog begins to realize that only the greatest things happen during a clicker session.  Once he realizes that there is no more punishment, he starts to offer.  (It's also a key reason why, if you are going to try clicker training a crossover dog, you should eliminate ALL correction for quite a while.)
     

    I always find this amusing because... don't we all? Of course dogs do what works. Everyone does what works. 

     

    Carla, I find it amusing, too, and then I usually start scratching my head because if people understand the concept of doing what works, why do they then assume that the dog simply does things to please us, or for praise, when, in fact that is not always what the dog wants.  Dogs are no different than we are in the sense that their motivators can change from day to day, or minute to minute.  If the dog has to pee, then he wants you to open the door.  At that moment, if I want to work on proofing a good "sit and wait by the door until the human releases you to go out", then I will do it at that moment.  Who is really training who?  I am training the dog to sit/wait, but he is training me to open the door when he sits.  Symbiotic and effective LOL.  Trainers get in trouble when they assume that the ONLY motivator should be either praise or food.  A motivator is anything that the dog wants.  Now, that doesn't infer that we always give the dog what he wants (perhaps to chew your best pair of leather loafers), but it does suggest that we need to "think outside the box" when it comes to training, and not bore the dog by overusing one reward versus another exclusively.  That keeps us on the scientifically proven track of getting a dog to repeat the behaviors we want through variable, or intermittent, reinforcement.

    • Gold Top Dog

    Houndlove, I didn't read your entire link about the porpoise experiments, but I wonder if that's the one that Patricia McConnell cites in her book For the Love of a Dog. Have you read it? I loaned it my sister, (who has 6 cats, and NO DOG, but is interested in getting one when she retires - she LOVED the book BTW!) so I can't look it up. There was a study where a porpoise had to come up with a new behavior, teach it to another porpoise, and then the porpoises would perform it in tandem. Awesome!

    • Gold Top Dog

    I have read For the Love of a Dog (library copy so I don't have one around either) but I don't remember the specific experiment cited in it. The original Creative Porpoise experiments didn't include any cooperative learning between the porpoises, but those original experiments set the foundation for a lot of future replication and extension of the same ideas, which is probably where the cooperative learning came in.