Wednesday, March 12, 2008

Schedules of Reinforcement or When and Why Does the Good Stuff Happen

Schedules of Reinforcement...I hate these guys. Just give a dog a treat and be done with it! But it's not that simple and different schedules are more effective for certain things, so here goes my take on them.

Schedule of Reinforcement – “a program or rule that determines how and when a response will be followed by a reward.” The schedule has an effect on how the response is learned and how it is maintained. Use a different schedule for learning than for maintaining.

Continuous Reinforcement Schedule or CRF - there is a reinforcement each time the response is observed. This is used when the student is first learning the skill.

Havana has had a hard time bringing the ball to me from the flyball box. At this point she always gets her tug if she brings me the ball, no matter what else happened on the run.

Bug is in the early stages of potty training. Every time she uses the potty she gets lavishly praised and choses some kind of reward (mint, tell daddy, call Nana, etc.).

Partial Reinforcement Schedule or PRF – also called an intermittent reinforcement schedule. Certain responses are reinforced, not all. The reinforcement is offered on a ratio or at intervals. Good for maintaining all-or-nothing behaviors.

  • fixed ratio FR – a set number between the number of responses and the number or rewards

    I actually don't use this one very much for Bug or my dogs. It doesn't work as well for me as other schedules. Hypothetically:

Cousteau is on a FR-4 schedule (4 responses before his reward). If he gives me four good sits I
will reward him after the 4
th sit.

A silly game Bug and I could play is where she'd hit my palm and I'd wait for 3 slaps before I
grab her hand. This would be a FR – 3.

  • variable ratio VR – the number of responses between each reinforcement changes from one time to the next. Also known as the “slot machine schedule” since this is what makes those one armed bandits so reinforcing for some people. Good for maintaining all-or-nothing behaviors.

When Havana works on heeling she may be reinforced after four steps one time, nine steps the next time, and five steps the time after that. This would be a VR – 6 because on average she has to give six responses before getting reinforced.

I try not to fall into this schedule, but I'm sure I do sometimes and I just can't think of a time. What I see with other children, especially in the grocery store, is a child asking, whining, demanding something over and over again. Sometimes it takes 5 requests and the parent gives permission just for some peace. Other times it might take 20 requests or on a rough day the parent may give in after one or two requests. This would be a VR – 9 or on average, this imaginary child gets what s/he wants after every ninth request.

  • random ratio RR – there is no correlation between the the behavior and the reinforcement. It just happens, like Fate.

      Cousteau walks by the pop corn popper and a piece falls out as his feet.

      Bug sits on the couch and Havana comes over to sniff her face for no apparent reason (this is very reinforcing for Bug who just wants the animals by her.

  • fixed interval FI – the response is reinforced only after a certain amount of time has passed.

    When working on proofing Cousteau's sit stay, he is only reinforced for holding position for at least 60 seconds. He may have to hold for longer than 60 seconds, but he will not be reinforced before 60 seconds has passed.

Bug LOVES her vitamins. (Or Bite-A-MUNS) She asks for them many times a day, but can only
have them in the morning. It does no good for her to ask at lunch time, after her nap, and during
dinner because she only gets that vitamin after she says please in the mornings.

  • variable interval VI -­ reinforcement occurs after a varied amount of time passes.

      Cousteau likes to run agility, but he has to wait quietly for his turn. He may have to wait quietly for three fast dogs to run, for one dog needing a lot of coaching to run, or he may be on the course back to back if we don't want to reset jump heights.

Bug likes to look at the pictures on the LOLCats website. If she asks 10 minutes after we first
looked at the pictures, there may not be any new ones. If she asks 20 minutes after we first
looked there may be two new ones.

Differential Reinforcement Schedule – the quickness of the response determines if the reinforcement.

Havana is learning impulse control as well as stimulus control. I will walk around a room with a toy and ask her to “sit” or “down” at random. If she slowly goes into position I tell her “uh uh” and walk on. If she gets into position quickly I tell her “get it” and we tug.

Even though Bug is capable of walking up and down stairs on her own, she'd rather have me with her. If I say I'm going upstairs and she tells me to wait, but doesn't show any indication of coming by me, I go upstairs on my own. If instead she puts down her toy and runs over to me we walk upstairs together.

Differential Reinforcement of Incompatible Behaviors or DRI – rewarding responses that cannot be done at the same time as an unwanted behavior. Also called alternative response training or countercommanding.

Havana was the most mature dog in her beginning agility class. We had a problem with loose dogs so I taught her if a loose dog came to her that she should look at me. If she was looking at me she wasn't face to face with a hyper, adolescent dog.

When I have a lot of dishes to do I bring a stool for Bug to stand on at the sink. She gets bubbly water and a few plastic dishes to wash while I do the rest of the dishes. If she's washing dishes with me she's not whining about wanting my attention or doing things to get my attention.

Differential Reinforcement of Other Behaviors or DRO – is what I call “hey, at least it's not anything bad.” You pick one target problem behavior in the student and reward anything that isn't that behavior.

My first dog walking client was a nightmare on leash. One day I got smart and as long as she wasn't pulling, I clicked and treated her. She may have been sniffing, peeing on something, looking around, whatever, it didn't matter because she wasn't pulling me.

I do the same thing with Bug. When she's having a tough day and she's been whining and getting into lots of trouble I make sure to tell her how much I like everything that isn't driving me insane. If she's been screaming in the car at the top of her lungs and she starts talking to her giraffe, I'll tell her how much I like hearing her talk with Geti. Then if she starts singing I'll tell her I enjoy that. I don't care what she's doing just so long as it's not screaming in my ear from the backseat.

No comments: