I've been trying to keep up with this whole VSR (I'll just use VSR, instead of trying to break it down into its subcategories) thread, and it sounds like people are confusing the procedures for successive approximation and variable reinforcement with a variable schedule of reinforcement.
In other words, holding out for a better "version" of a behavior is NOT a variable schedule of reinforcement, which I think is where some of the list readers are getting confused, I know that's where I got lost and had to backtrack. It is just raising the criteria of enforcement. But under that criteria, your dogs are still recieving a reinforcement every time they perform right? Whether or not they recieve the same reinforcement also does not change the schedule of reinforcement.
A true VSR is when the reinforcement is randomly given at a particular and set criteria (either a ratio or interval). My criteria is a sphinx down, in 1 second. So I start those. First, I only go to a VSR or a VIR when I'm done shaping what my perfect vision of the behavior was.
Example of VSR:
(You have to gradually shape the extension of number of reps before rewards just like any other criteria.)
Does that make sense? An animal should be put on a variable schedule when there is a distinct possibility that it may have to perform the behavior for a number of times without reinforcement that could initiate extinction procedures. BUT a lot of people do a couple of things wrong.
First, they put behaviors on VSRs for convenience: they don't want to carry rewards with them, they want to hurry up and start some kind of chaining and they feel a VR is necessary for that, etc. Second, they go to VSRs too early. If you're still shaping, changing criterion, extending durations, etc. you shouldn't go to a VSR. The animal is still learning, extinction shouldn't be an issue and the animal needs as much feedback as possible.
People also seem to think that VS's will increase fluency or decrease fluency. That is not true either way if the schedule is put into place correctly. The only thing a pure and properly placed VS will do to behavior is make it less succeptible to extinction. The decrease in fluency that Gary Wilkes mentions (wobbling of behavior) is actually a result of raising the VS criterion too quickly. He does this on purpose to show people different ways to get what I call "abberations of behavior" which can be used in shaping.
I have a pet rat (daughter's actually) that will throw himself on his back and stick all fours straight up and freeze with his mouth hanging open (just added that last part) on the cue "RATTRAP!!" I put that on a variable ratio schedule very early, but I did it improperly to get the abberation of behavior that is necessary in pure shaping. I got him on his side, but I couldn't get him to roll over on his back. I put the side on a harsh variable ratio schedule early and he became frustrated at no reinforcement and rolled on his back once.....CLICK.
That's how that can be used. But again, it was not a proper well set up schedule and that's what caused the "wobbling" as Wilkes puts it. An appropriately admininstered VS would have done nothing but made the laying on his back harder to extinct.
Copyright of all posts is the property of the original author. Please obtain permission from the original author before copying, quoting, or forwarding.
List and Site Owner: Melissa Alexander, mca @ clickersolutions.com