Questions about positive training

Share your favorite training tips, ideas and methods with other Positively members!

Moderators: emmabeth, BoardHost

Post Reply
TheBealge
Posts: 14
Joined: Wed Jun 24, 2015 3:50 pm

Questions about positive training

Post by TheBealge »

Ok, so if you are training a dog to do something, anything, and you use rewards, won't the dog only do it if you have a reward with you? I hear you should try to mix it up so that the dog never knows when he will get a treat, and if he doesn't get it that time, then the chances that he will get
one next time are increased. However, what is going to stop the dog from only doing whatever it is you want him to do when he SMELLS the treat.
Like if I told my dog to come, and he didn't smell, hear, or see the reward, why would he come? I don't dislike this style of training, I just don't fully
understand it.
When I signed up, I typed my username in too quickly. I wanted it to be TheBeagle. :lol:
Corgies, beagles, and bassets rock! 8)
delladooo
Posts: 763
Joined: Sun Apr 13, 2014 4:53 am
Location: Blackpool, UK

Re: Questions about positive training

Post by delladooo »

The simple answer is no. Realistically, the answer is slightly more complicated as if you don't do it properly (ie fade and/or mix up rewards) then you can end up with a dog that will only do something when it knows there will be an immediate reward.

I'm not good at explaining so I'm sure something who can say it a lot better will come along and explain it in a much simpler way but basically, your reward is what the dog finds rewarding, usually food is a good one but not the only option, and you want one of these rewarding things to be you. When you're training you mix rewards of food, human interaction and toys (if the dog likes toys) and vary the level of rewards too (think normal kibble vs fresh chicken) and the variety keeps the dog engaged and looking forward to what's coming next.
jacksdad
Posts: 4887
Joined: Thu Dec 10, 2009 10:48 pm

Re: Questions about positive training

Post by jacksdad »

TheBealge wrote:Ok, so if you are training a dog to do something, anything, and you use rewards, won't the dog only do it if you have a reward with you?
It is possible to run into this IF if you do not use the food correctly. But the issue is one of trainer knowledge and skill, not inherit in the application of learning/training principles or "positive reinforcement training". It is just as possible to end up with the exact problem from a punishment perspective, a dog that only walks in a heel like position only IF you regularly apply leash pops. Again, an issue of trainer knowledge and skill.
TheBealge wrote:I hear you should try to mix it up so that the dog never knows when he will get a treat, and if he doesn't get it that time, then the chances that he will get one next time are increased.
what you are talking about is different schedules of reinforcement, and there is no blanket rule as to which one to use and when as there are LOTS and LOTS of different schedules of reinforcement.

If sit earns reinforcement every time sit is performed, that is known as a continuous reinforcement schedule.

if every time your dog sits reinforcement doesn't happen that is known as a variable schedule. may happen this time, may not, but reinforcement happens fairly frequently still, just not overly predictable.

There are LOTS of different reinforcement schedule options, which one we use depends on the situation, goal, need of the dog. There are times staying with a continuous reinforcement schedule is the right thing to do. nail trimming comes to mind as but one example.

Other times it makes more sense to not only keep your dog guessing when reinforcement happens, but what kind of reinforcement happens. this time food, this time nothing, next time toy, back to food for the next two times, then nothing, but praise the next, and toy after that.

Which schedule to use is both an art and science based on training goals and experience and knowledge.
TheBealge wrote:However, what is going to stop the dog from only doing whatever it is you want him to do when he SMELLS the treat.
Like if I told my dog to come, and he didn't smell, hear, or see the reward, why would he come? I don't dislike this style of training, I just don't fully
understand it.
this is not a problem if sound training principles are applied. If you use food right, then it won't matter if your dog smells it or not on you because the power is NOT in the food on you, but the properly applied reinforcement and built up reinforcement history. In theory, this problem could happen with toys as well, since the more a toy is used the more smell gets attached to it from your dog's mouth, your hands, things the toy laid in etc. Again, properly applied principles negate this as a problem.
User avatar
Nettle
Posts: 10753
Joined: Sun Apr 13, 2008 1:40 pm

Re: Questions about positive training

Post by Nettle »

There is also the issue of bonding - of being valuable to your dog because where you are is safe, where you are is where the best things happen, when you are together you do fun things. This is very much neglected in the average pet dog's life. Most pet dogs don't have a great relationship with their owners, even though their owners "love" them. But if the presence of the owner enriches the dog's time, then the dog is far more likely to stop a thing it wants to do in order to go to its owner and share what the owner wants it to do.

We would never get that without pleasant reinforcement. Dogs that work out of fear of punishment (punishment does not have to be harsh to create fear of itself) are not going to bond with the bringer of that punishment. That is why there is so much wastage with the old methods of training, because some dogs will not bend to punishment, and some others become so shut down they don't dare to do anything. Many breeds/types of dog will not be trained by punishment - these are the ones that often get labelled 'stupid' or 'stubborn'. They all train well with reward, but turning a punishment-trained dog round to trust enough to work for reward can be a long task.
A dog is never bad or naughty - it is simply being a dog

SET YOURSELF UP FOR SUCCESS
jacksdad
Posts: 4887
Joined: Thu Dec 10, 2009 10:48 pm

Re: Questions about positive training

Post by jacksdad »

Nettle wrote:There is also the issue of bonding - of being valuable to your dog because where you are is safe
this cannot be over stated. I am constantly amazed at how powerful simply being safe to be around when working with dogs is.
Nettle wrote:(punishment does not have to be harsh to create fear of itself)
since most people are not students of the various signals dogs give off to communicate they often do NOT see their dogs reactions to their own body movements and tone of voice.

when we talk about punishment in training terms, it's not about good/bad or right/wrong (at least it shouldn't be) rather it is about something being added into the "mix" that STOPS a behavior. we tend to think of punishment only in terms of hitting a dog, kicking a dog, using a shock collar, or maybe a prong collar and such. BUT it can be something much more innocent. Just like the learner (in this case the dog) controls what is reinforcing, the learner also controls what is punishing.

Knowing that punishment is ANYTHING added into the "moment" that stops a behavior and seeing it in action even without something harsh happening is a real eye opener. here is a real world example I observed.....

recall training, dog is running happily to it's owner....owner bends over in excitement and love and joy welcoming their pup with open arms and praise.... but to the dog the leaning over is uncomfortable (and can be fear inducing in some) and so stops the behavior of returning to their owner just short of where they really should.... their owner had no idea they had punished the behavior of returning to them.

*edited to clarify a point about punishment. There are other options to stop a unwanted behavior that do not involve punishment.
Last edited by jacksdad on Wed Sep 16, 2015 11:30 am, edited 1 time in total.
JudyN
Posts: 7018
Joined: Tue Jul 26, 2011 1:20 pm
Location: Dorset, UK
Contact:

Re: Questions about positive training

Post by JudyN »

jacksdad wrote:Knowing that punishment is ANYTHING that stops a behavior and seeing it in action even without something harsh happening is a real eye opener. here is a real world example I observed.....

recall training, dog is running happily to it's owner....owner bends over in excitement and love and joy welcoming their pup with open arms and praise.... but to the dog the leaning over is uncomfortable (and can be fear inducing in some) and so stops the behavior of returning to their owner just short of where they really should.... their owner had no idea they had punished the behavior of returning to them.
I'm not an expert but do we have to distinguish between positive and negative punishment here, as we wouldn't want anyone to think that positive trainers can't stop a dog from doing something. Turning your back on a dog when he jumps up decreases the behaviour, and is presumably a 'punishment', but is appropriate. Also, is stopping a behaviour by asking for an incompatible behaviour punishment, e.g. asking the dog to 'leave' a cowpat and the dog then being rewarded with a treat, or to 'wait' when walking towards a road? Just about any behaviour involves stopping a different behaviour - for instance teaching a dog to stop barking by training 'quiet' can be the equivalent of teaching a normally quiet dog to bark on command.
Jasper, lurcher, born December 2009
jacksdad
Posts: 4887
Joined: Thu Dec 10, 2009 10:48 pm

Re: Questions about positive training

Post by jacksdad »

JudyN wrote:I'm not an expert but do we have to distinguish between positive and negative punishment here, as we wouldn't want anyone to think that positive trainers can't stop a dog from doing something.
Yes, lets be clear positive trainers are NOT without options. punishment isn't the only option for addressing unwanted behavior.

Quick review -
positive punishment = add something to stop a behavior.
negative punishment = remove something that is reinforcing a behavior to stop a behavior.

by definition IF a behavior continues it is is somehow someway being reinforced or it would not continue. Teaching an incompatible behavior is NOT punishment. it plays more with extinction of a unwanted behavior by making something else MORE reinforcing.

For example jumping. If you were to teach your dog to sit to greet vs jump to greet you would due this by managing the situation so the dog isn't likely to jump, then HEAVILY reinforcing sit prior to greeting. This is "ignoring" the jumping, focusing on the sit. when I say ignoring the jumping I don't mean as the dog is jumping you ignore the dog. if the dog is jumping you still have not setup your training to prevent reinforcement from jumping. for example you are allowing the dog to get to close to the person prior to asking for a sit, so add more distance between dog and human being greeted. this is using management, breaking down the behavior into small steps that allow success and create MANY opportunities for reinforcement of the wanted behavior.

Turning your back to address a jumping dog is an attempt at negative punishment to address jumping. but this option will only work if jumping on your back is NOT reinforcing. some dogs don't care if it is your back, your front, your side jumping they get reinforcement out of simply jumping on you. any response by you is reinforcing so they don't give a darn what part of you they can jump on. this is just one reason turning your back on a jumping dog is highly over rated as a fix. Another, it doesn't teach what the dog should do.

teaching incompatible or alternate behaviors and creating a reinforcement history so strong that these new behaviors become the new default really is a much more powerful solution to unwanted behavior.
TheBealge
Posts: 14
Joined: Wed Jun 24, 2015 3:50 pm

Re: Questions about positive training

Post by TheBealge »

jacksdad wrote:
this is not a problem if sound training principles are applied. If you use food right, then it won't matter if your dog smells it or not on you because the power is NOT in the food on you, but the properly applied reinforcement and built up reinforcement history. In theory, this problem could happen with toys as well, since the more a toy is used the more smell gets attached to it from your dog's mouth, your hands, things the toy laid in etc. Again, properly applied principles negate this as a problem.
How exactly would you avoid this issue; how could I get my dog to come to me if there is no reward? What are the sound principles?
When I signed up, I typed my username in too quickly. I wanted it to be TheBeagle. :lol:
Corgies, beagles, and bassets rock! 8)
katej215
Posts: 302
Joined: Wed Jan 11, 2012 6:05 am

Re: Questions about positive training

Post by katej215 »

TheBealge wrote:jacksdad wrote:

this is not a problem if sound training principles are applied. If you use food right, then it won't matter if your dog smells it or not on you because the power is NOT in the food on you, but the properly applied reinforcement and built up reinforcement history. In theory, this problem could happen with toys as well, since the more a toy is used the more smell gets attached to it from your dog's mouth, your hands, things the toy laid in etc. Again, properly applied principles negate this as a problem.

How exactly would you avoid this issue; how could I get my dog to come to me if there is no reward? What are the sound principles?
In simple terms my understanding is...the reinforcement (or reward) MUST come AFTER the behaviour....so you get a behaviuor you want, you mark it with a clicker or a marker word like 'yes' and it is that alone that signals the reinforcement is coming.

I think where people run into problems is either they do not reinforce at a high enough rate at the beginning of teaching a new behaviuor (ie they havent built enough of a reinforcement history ...the analogy i like is they have not put enough money in the bank before making withdrawals, so reserves are soon drained) OR

they have the reinforcement visible to the animal BEFORE the behaviour occurs which then means you are getting into the realms of bribery ie: the dog will soon be saying, 'hang on, i'm not doing that till i have seen the reward'

If that is happens I would go back to mega easy behaviours and reinforce highly for them, once you have fluency of response then start to mix things up: ie your reinforcement could be a piece of kibble or a bit of sausage or you could jackpot for a really good response..get fluency at this level, then start to alter your reinforcement rate, ie: you can ask for more for the same amount of reinforcement...

Behaviour responses always vary, that the law of behaviuor....your job as a trainer is to be quick enough to capitalise on that when you get an above average response.

Have you read any Karen Pryor? Don't shoot the dog is the classic, but I would also recommend her collection of essays and lads before the wind...really helps make sense of the mechanics with illustrative stories rather than being too text book like. Jean Donaldson also explains technocalities well in Culture Clash :D
jacksdad
Posts: 4887
Joined: Thu Dec 10, 2009 10:48 pm

Re: Questions about positive training

Post by jacksdad »

TheBealge wrote:
jacksdad wrote:
this is not a problem if sound training principles are applied. If you use food right, then it won't matter if your dog smells it or not on you because the power is NOT in the food on you, but the properly applied reinforcement and built up reinforcement history. In theory, this problem could happen with toys as well, since the more a toy is used the more smell gets attached to it from your dog's mouth, your hands, things the toy laid in etc. Again, properly applied principles negate this as a problem.
How exactly would you avoid this issue; how could I get my dog to come to me if there is no reward? What are the sound principles?

A full answer is honestly larger than a single post. But it would include learning when to give food, understanding how to use food a lure to get something close to the behavior you want. how to shape for a behavior, how to capture, understanding that for example sit in your living room does not mean your dog is ready to sit in a park with dogs, squirrels, bikes and joggers all out and about offering distractions. these things have to be factored into your training plan. you need to build experience with breaking down a behavior into small parts, then putting them back together in a way that lets your dog be successful. you also need to understand that mistakes are part of learning, for both you and your dog. a dog that isn't doing exactly what you want is not stubborn, blowing you off, trying to be in charge etc, this is very, very likely a dog that is confused or not feeling well, may simply need a break, is tired, is stressed, OR is simply being asked to do something you only think you trained, but in reality didn't. Recall is a common one for this list issue.

In my opinion the two best books to start out with for learning sound training principles and concepts are

Don't shoot the dog
http://www.amazon.com/Dont-Shoot-Dog-Te ... ot+the+dog

train your dog like a pro
http://www.amazon.com/Train-Your-Dog-Li ... n+like+pro

if you really want to dive into the geeky details of the science behind learning and behavior, a very easy to read and understand book for that is
Learning and Behavior
http://www.amazon.com/Learning-Behavior ... th+edition

So how would you start using sound principles for training a recall that does not require food every single time you need to use it? The reality is, to keep a behavior strong you do have to do refresher training. NO behavior be it having your dog return to you on cue, swinging the golf club just right for a long drive, shooting from the free throw line in basketball, skiing, running for fun and exercise all perishable skills needed to "do it right" that if not kept up..perish and need to be retrained. So while it is possible to have a dog NOT need food or reinforcement for every single time you call them to you, if you want to keep that level of proficiency, you will have to build in training time and have some kind of reinforcement going on for life. this is NOT a failure of "positive training" this is simply reality of working with a living, thinking being be it human or dog.

Note - technically when we "reward" our dogs, it is reinforcement. Reinforcement means to make stronger. when our dogs do what we want, we give them something they value to make that behavior stronger aks reinforced, aka, more likely to happen than something we don't want. it is ok to say reward, but in the strictest sense we are NOT rewarding, we are reinforcing. It is an important principle to understand.

Where to start....

step one. build a default behavior for checking in with you. you want your dog to on their own learn to "keep tabs" on you. one way is every time your dog looks at you, they earn reinforcement.

step two. a cue is meaning less to our dogs until we teach them there is value in it it. until our dogs learn that sound we make when we say "come" has meaning it is just noise to them and as such it should be no surprise they don't respond to it until trained. one way to do that, in a VERY quite and non distracting place in your home, twice a day for 1 week get 10 pieces of the MOST YUMMMY food your dog likes. We are talking SUPER YUMMY steak or hot dogs or chicken etc. something your dog considers high value. The key is not what you think your dog should like, but what your dog actually likes. then in your quite place, with your dog, say the cue "come", wait 1 second, give the food. repeat 10 times. this classically conditions the word "come" to have meaning.

Step three. in a zero distracting place, like say a bathroom or short hallway, have your dog with you, take one step away...say "come". does your dog move to you? if so, reinforce. work at this level for a few days until your dog is able to come from one end of the room or hall to you when you say "come" ONCE.

At this point we are NOT worked about variable schedules of reinforcement. we are still building the behavior and continuous reinforcement is what you want. one behavior = one reinforcement.... each "come" earns reinforcement.

you repeat step 3 in different ways introducing more and more distraction. go from bathroom to hall to living room to back yard, to front yard, to quite soccer field, to slightly busier park etc. all in small enough steps so that your dog can succeed easily. you can with careful thought add in distractions such as passing food, toys etc. but again you do this in ways that make it easy at first, you build a strong history of reinforcement and success and make it harder and harder over time.

lots of ifs, what if, what about, it depends etc to this, this is not a complete "recipe" for how to teach recall. but it would get you started.

not having to use or have food on you at all times is dependent on reinforcement history and did you train for X situation. you may have trained coming to you with nothing going on...dog is at one end of field, you at the other, but no birds, bikers, runners, squirrels, bunnies etc....and maybe highly successful in this situation. add a bunny, don't be surprised when this all falls apart. it may not, but it may...and why might it fall apart...you didn't train for coming away from a bunny.

hopefully this helps.
Post Reply