R+, R-, P+, P- explained

OK  I am a "purist" for lack of a better term [].
Fishers definitions are very well done.  However, I see them just a bit differently.  Actually it is the "old" way (reframed like I do at work for "non believers"[])

Positive reinforcement:  Likelihood behavior will happen again because the consequence following the behavior met the dog's immediate "need"

Negative reinforcement Likelihood behavior will happen again because the behavior is immediately followed by the removal of an avervise (some thing that does meet the dog's "need" by stopping something from happening)

Punishment Likelihood the behavior will not happen again because the behavior is immediately followed by something that stopped the beahvior.  Stopping the behavior meets the dog's "need".

Extinction:  Likelihood the behavior will not happen again because the action  following the behavior no longer meets the dog's "need".

The problems with these concepts I believe, centers around the typical/common meanings of the words positive, negative, punishment etc.

So to offer some examples to illustrate that point.
You are must speak in front of a group.

Positively reinforced:  applause causes a pleasure reaction;  teacher praises a kid in class for answering and the kid keeps volunteering.

Negatively reinforced:  people are murmuring while you speak.  you stop speaking, murmuring stops.  you speak again.  Kid on playground is teased.  Spits on kids who are doing teasing, teasing stops.

Punishment:  people are murmuring while you speak.  you stop, they keep murmuring, no reaction or minimal reaction at the end.  You swear to never speak to a group again and dont.  Kids in class are talking, teacher walks and stands next to kids.  Kids stop talking.

Extinction:  no one shows up.  Kid is calling out answers during discussion.  Teacher only calls on kid raising hand.  Kid calling out starts to raise hand.  Teacher calls on kid.  Extinction of calling out, followed by positive reinforcement (teacher calls on kid when hand is raised.

One of the things folks need to remember, that you need to manage or control the environment to change behavior. You need to observe and take "data" on MULTIPLE behaviors over MULTIPLE days to determine is something has been reinforced, punished or extinguished.

I agree with the linked reference that -R rarely gets used. An applied irritant is removed when a behavior stops. It would be too easy to call the aversive sound or stimulus +P. How many many people use a NRM with an irritating sound (eh-ehh) and consider it a verbal correction, which would be +P? And is it, in fact, +P? It happens so fast, there's no way to tell if the sequence lends itself to -R, instead.
 
While we're talking about the quadrants of OC, I want people to notice Ixa's Girl's application of it as mod. While she may be quick to correct someone for out of bounds stuff, she also pops in with a kind note if the thread is staying civilized. So many times, all any of us ever hear is when we did something wrong, hardly ever when we did something right. So, I would say that's an example of +P and +R, applied when and where it seemed needed. And, at least with humans, it can be tough to do +R because not every one thinks to reward for good behavior.

ORIGINAL: ron2

I agree with the linked reference that -R rarely gets used.

 
My understanding is that -R is represented in Clicker Training when a treat is withheld.
 
Ron2, another example of this is the removal of the 2 from loginId?  Time will tell if it is effective.

ORIGINAL: DPU
My understanding is that -R is represented in Clicker Training when a treat is withheld.

That's actually negative punishment, -P - withholding something good (a treat) to decrease a behaviour (the behaviour you don't want).

LOL Ron! Well, you know what they say about P+, if it's not sharp enough to change behavior it's just a lot of nagging. []
Anyway, thanks, you're one of many who know, first hand, about my orientation towards R+ [sm=angel.gif].

let me use the "speeding" example:  The cops want you to drive at 55 mph on a road. You prefer to drive at 100 mph cause it's fun-- it is intrinsically motivating to you to go really fast (self-rewarding +R).
Cop applies nagging-+P: when he sees you speeding, he makes you stop and punishes you by giving you a ticket. Your behavior of speeding is temporarily suppressed, but after a few days you're back to speeding. You also don't feel kindly towards cops for the rest of your life. This the method most people use when they train dogs-- the collar correction, the NO.
Cop applies effective +P: he booby traps your car so it explodes if you go over 55 mph. Two explosions later you decide to never speed again. You're also a nervous wreck wheneever you are in a car for the rest of your life. You don't associate cops with the explosion, so you feel fine towards cops.
Cop applies +R: whenever he sees you driving at 55 mph he gives you a hundred dollar bill. You spend a lot more time driving at 55 mph. You love cops.
Cop applies -R: cop booby traps your car so it makes a horrible squealing noise whenever you go over 55 mph. After a few experiences you never drive over 55 mph again. You don't associate cop with the booby trap so you feel fine towards cops.
Cop applies -P: can't really come up with a good way to apply -P to modify self-rewarding behaviors.

Here was my original question:



[blockquote]
lostcoyote:

r+, p-, p+, r-

+R means the dog does something the trainer request and get's reward

-P means the dog does something wrong and gets a punishment.

what are examples of -R and +P ? or is there such a thing in this jargon?

educate me

=============================================[/blockquote]
fisher replies with:

These are specific operant conditioning terms that have fixed meanings. This terminology is designed not to be subjective, and yes, there are examples!

The Logic:

A behavior is labelled + if it is doing or giving something.
A behavior is labelled - if it is withholding something.

It is Reinforcement if it's *increasing* the likelyhood of a behavior in the future.
It's Punishment if it's *decreasing* the likelyhood of a behavior in the future.

+R Positive Reinforcement. Doing/Giving something to increase the chances of a specific behavior happening again.

+R Examples: Dog barks, dog gets attention. Dog barks again (for more attention). Dog poops outside. Dog gets lavish praise and treats. Dog poops outside in the future.

+P Positive Punishment. Doing/Giving something to decrease the chances of a specific behavior happening again.

+P Examples: Dog growls, dog gets alpha roll. Dog less likely to growl in future. Dog pulls on leash, dog gets collar correction, dog less likely to pull on leash in the future.

-R Negative Reinforcement. Rarely used in a dog training environment. Witholding an aversive in order to increase the chances of a specific behavior happening again in the future.

-R Examples: Dog is in room with loud squealing sound. Dog sits, squealing sound goes away. Dog more likely to sit in the future. (See? Not so practical for everyday use)

-P Negative Punishment. Withholding something the dog wants in order to decrease the chances of a specific behavior happening in the future.

-P Examples: Dog jumps up on human, human walks away and shuts door behind dog. Dog less likely to jump up in the future. Dog doesn't come when called. Human goes and gets dog and puts dog on a leash. Dog less likely to blow off "come" in the future.

The Catch:

Any behavior you, the trainer give, is labelled +P or -P or whatever based not on your intentions, but on what it actually makes the dog do. Example: I can think that yelling "No!" at my dog when he barks is +P, when in fact it is +R, because he doesn't understand that he's being "yelled at," he just likes getting any attention.

=====================================================

coyote has additional questions:[align=right] [align=right] The Logic:
A behavior is labeled + if it is doing or giving something.
A behavior is labeled - if it is witholding something.
 
okay, so we are referring to the dogs behavior, the humans behavior, or both?

example below:





[blockquote]quote:

-P Negative Punishment. Withholding something the dog wants in order to decrease the chances of a specific behavior happening in the future.

-P Examples: Dog jumps up on human, human walks away and shuts door behind dog. Dog less likely to jump up in the future. Dog doesn't come when called. Human goes and gets dog and puts dog on a leash. Dog less likely to blow off "come" in the future.[/blockquote]



[blockquote][/blockquote]

but the dog jumped so the dog did something(+).
and in this case, the (-) indicates behavior on part of the human.




[blockquote]quote:

+R Positive Reinforcement. Doing/Giving something to increase the chances of a specific behavior happening again.

+R Examples: Dog barks, dog gets attention. Dog barks again (for more attention). Dog poops outside. Dog gets lavish praise and treats. Dog poops outside in the future.[/blockquote]

in this case, the dog did something (+) and the human also gave something (+) so both human and dog get a (+)


see my confusion here?
so do the + and - marks refer only to the dogs behavior or to the humans reaction to that behavior?

The + and the - refer to the dog's behavior. This is all from the standpoint of the subject (in this case, the dog) not the trainer.

The + and the - refer to the dog's behavior. This is all from the standpoint of the subject (in this case, the dog) not the trainer.

okay, so the previous quote would be incorrect??

 
A behavior is labeled + if it is doing or giving something.
A behavior is labeled - if it is witholding something.

-P Negative Punishment. Withholding something the dog wants in order to decrease the chances of a specific behavior happening in the future.

-P Examples: Dog jumps up on human, human walks away and shuts door behind dog. Dog less likely to jump up in the future. Dog doesn't come when called. Human goes and gets dog and puts dog on a leash. Dog less likely to blow off "come" in the future.

 
the example gives a (-) but pre previous definition, (-) implies the dog is witholding something... but the dog jumped which would make that a (+ = doing something, in this case, jumping)), wouldn't it?

Oh wait I see what you are saying. Think of the + as something added to the dog's environment and the - as something removed from the dogs environment. Get the human out of it (though of course in real life it's usually the human doing the adding or removing) and that might help you. The + and - don't refer to behavior, they refer to things in the subject's environment. The R or P refer to the behavior of the subject (decreasing in likelihood or increasing in liklihood).

so then giving a treat for something you want to reinforce (such as a "sit" command) is +R


while giving a CM "shhhhht" would be a +P  (adding the "shhht" to reduce the behavior)


right?

When I give a treat after Marlowe jumps up to hit his target stick, that's +R becuase I am adding something to his environment that will increase the liklihood of that particular behavior to be repeated. Turning away from him and withdrawing attention when he jumps up at other times would be -P (removing my attention in order to decrease the liklihood of the behavior happeneing again). If I were to knee him in the ribs when he jumped, that would be +P becuase I would be adding something to his environment (pain) in order to decrease the liklihood of the behavior being repeated. If when teaching Marlowe to jump up and target his stick, I pinched his ear until he finally jumped up and performed his target at which point I released his ear, that would be -R because I would be removing something (pain) from his environment in order to increase the liklihood of that behavior happeneing again (-R is defiantely the least practical of the quadrents).


Edited to reflect corrected typeos above!

ORIGINAL: lostcoyote

so then giving a treat for something you want to reinforce (such as a "sit" command) is +P
while giving a CM "shhhhht" would be a +R  (adding the "shhht" to reduce the behavior)
right?

Hey, I always gets these quandrants confused and when I think I get it straight then someone new corrects me.  You are having fun but you are confusing me.

i did, that was a typo - i was thinking when i typed that -  that p = positive - hahahahahah - will go back and edit it... 
 
 
p.s.  i think i'll just stick with using the eglish language - all this tech stuff is making my head hurt (and i'm employed to work with tech stuff - go figure)