Pay for Performance: Setting criteria in training
and how not to stuff up your training in real life.
Your dog knows how to Sit and Stay. At home, with familiar people he will easily Sit and Stay for about 30 seconds before rewarding.
On a walk, you meet a friend you rarely see. You ask your dog to Sit (so you can have some control over the greeting) and he behaves as if he's never even heard of such a thing.
What happened?
Your friend is a big distraction (as well as a big reward if your dog likes people). You are probably quite close to this distraction. You are on a walk and you might never have asked your dog to Sit on a walk (or maybe only at the road).
In short, you are setting the bar too high - your dog is unable to perform what you want him to, because you have raised a number of "criteria" (such as distraction and distance) in a location your dog has never performed this behaviour in before.
On a walk, you meet a friend you rarely see. You ask your dog to Sit (so you can have some control over the greeting) and he behaves as if he's never even heard of such a thing.
What happened?
Your friend is a big distraction (as well as a big reward if your dog likes people). You are probably quite close to this distraction. You are on a walk and you might never have asked your dog to Sit on a walk (or maybe only at the road).
In short, you are setting the bar too high - your dog is unable to perform what you want him to, because you have raised a number of "criteria" (such as distraction and distance) in a location your dog has never performed this behaviour in before.
In rewards focused training, we reward the behaviour we want, and withhold the reward for the behaviour we don't want. The dog will perform the rewarded behaviour more often, and the non-rewarded behaviour less often.
Fair training is about setting the dog up for success most of the time, with some failures to gain the reward as extra information about what it is we want the dog to do (and not to do).
The art of fair training lies in changing the criteria so we gradually work towards the behaviour we want - by slowly asking for more before the dog gets "paid".
Payment can be in whatever is rewarding in that situation - it can be food, or it can be the permission to say hello to someone they would like to go see.
Fair training is about setting the dog up for success most of the time, with some failures to gain the reward as extra information about what it is we want the dog to do (and not to do).
The art of fair training lies in changing the criteria so we gradually work towards the behaviour we want - by slowly asking for more before the dog gets "paid".
Payment can be in whatever is rewarding in that situation - it can be food, or it can be the permission to say hello to someone they would like to go see.
The planned training approach
So, what would a planned approach to the above scenario (Sit to Greet on a walk) look like?
First, you have to go back and lower the bar significantly - to the point where the dog can perform a first approximation to what you want in the end.
Only once your dog can Sit without distractions, can you ask for a Sit and Stay (i.e. you are increasing the "duration" of the behaviour).
Next, practice your dog holding a Sit and Stay with distractions. Of course, this needs to work at home first before you try it outside :-)
When you are ready with all this, you can introduce a person to greet ... remembering to initially pay well and very often, and only ask for a brief Stay before releasing you dog to his reward.
Over time, ask for more: Ask for longer Stays, at a closer distance, with fewer rewards. BUT: Not all at the same time! You dog will likely fail. Make one thing harder, and keep the rest the same.
First, you have to go back and lower the bar significantly - to the point where the dog can perform a first approximation to what you want in the end.
- Ask for random Sits on a walk away from any obvious distractions. (Training a behaviour in a variety of locations is called "Generalisation").
- Reward any Sit really well (i.e. up the value / amount of reinforcement).
- If your dog struggles to Sit, go back to re-training the Sit on a walk, possibly all the way back to luring a Sit (i.e. you are increasing the prompts for this behaviour.)
Only once your dog can Sit without distractions, can you ask for a Sit and Stay (i.e. you are increasing the "duration" of the behaviour).
Next, practice your dog holding a Sit and Stay with distractions. Of course, this needs to work at home first before you try it outside :-)
When you are ready with all this, you can introduce a person to greet ... remembering to initially pay well and very often, and only ask for a brief Stay before releasing you dog to his reward.
Over time, ask for more: Ask for longer Stays, at a closer distance, with fewer rewards. BUT: Not all at the same time! You dog will likely fail. Make one thing harder, and keep the rest the same.
But what if you're caught? (a.k.a Real Life)
Your dog learns all the time, whether you are in training mode or not.
So if you just let your dog pull you over to your friend, and then let him jump up on them, you have just rewarded (and will get more of) pulling on the lead, ignoring you completely, and jumping on people.
Because you'll get more of what works for the dog. Whether you intend to or not, that's the law of behavioural science.
But you haven't had time to train this properly. Yet here you are. What to do?
You have various options:
- Try to lure the dog or use a large hand signal to get him to Sit. Then release to Say Hi [ a DeltaDogz method for controlled people and dog meetings]. You can do this at quite a distance and run the rest of the way to greet your friend!
- Or: Ask the friend to meet another time if your dog doesn't cope.
- But if you really need to talk to this person, and your dog is struggling, try Kay Laurence's Parking protocol. Note that I do not recommend falling back on this frequently ... I suggest you do some training, and you should be able to do better than this, but it can be helpful if all else fails. Better than letting your dog jump all over people.
So if you just let your dog pull you over to your friend, and then let him jump up on them, you have just rewarded (and will get more of) pulling on the lead, ignoring you completely, and jumping on people.
Because you'll get more of what works for the dog. Whether you intend to or not, that's the law of behavioural science.
But you haven't had time to train this properly. Yet here you are. What to do?
You have various options:
- Try to lure the dog or use a large hand signal to get him to Sit. Then release to Say Hi [ a DeltaDogz method for controlled people and dog meetings]. You can do this at quite a distance and run the rest of the way to greet your friend!
- Or: Ask the friend to meet another time if your dog doesn't cope.
- But if you really need to talk to this person, and your dog is struggling, try Kay Laurence's Parking protocol. Note that I do not recommend falling back on this frequently ... I suggest you do some training, and you should be able to do better than this, but it can be helpful if all else fails. Better than letting your dog jump all over people.
In summary
There are many ways to help your dog succeed and create a win-win situation when teaching your dog what you want him to do.
You are the one in charge of making sure your dog can learn, and is learning the right things.
If your dog isn't doing what you want, adjust ...
If your dog is going well ( about five successful repetitions):
Which one of the above you choose depends on what you want to achieve.
Just make sure not to push more than one aspect at at time.
You are the one in charge of making sure your dog can learn, and is learning the right things.
If your dog isn't doing what you want, adjust ...
- Move away from the distraction.
- Reward more frequently / reduce duration: After fewer steps when heeling, after a shorter stay, ...
- Pay better: Choose higher value reinforcers.
- Ask for less behaviour: Go back a few training steps of the behaviour you have in mind. Eg go back to luring every step of Heeling, rather than using a hand signal.
If your dog is going well ( about five successful repetitions):
- Work closer to the distractions, or:
- Ask for more duration / reward less often, or:
- Reduce the value of reinforcers (but keep some higher value ones for really good performances), or:
- Ask for more behaviour: more precise, or without luring, or a faded hand signal ....
Which one of the above you choose depends on what you want to achieve.
Just make sure not to push more than one aspect at at time.