How to set up your learner for success! 😊
How and where we deliver reinforcement makes a difference in the learning process and can help us set up our learner, whether they're a dog or a horse, for successful repetitions of the behaviour - or set them up to struggle....
I wanted to use some clips of training Jedi to demonstrate this, as with a horse everything is generally a bit bigger and slower, so it's very easy to see the process in action and spot mistakes.
In this video I have put together clips from a couple of sessions where I am working on teaching movement from one cone to another. It's useful for teaching horses to "lunge" without the need for a lead reign (they just move around the targets on their own) and to eventually turn left and right on cue. You may have seen previous videos of me teaching Jedi to target a cone in my hand, then transitioning to on the ground and finally moving to the cone from a short distance away, so this is a continuation of a lot of initial work.
In this video you can see me starting to teach him to touch two cones, alternately, very close together. This was the first step in teaching the idea that there might be more than one cone in play. The problem was that Jedi would understandably keep touching the same one, because obviously that worked i.e. it earned a click and treat. So I put my thinking cap on and remembered that reward placement can be used to help the learner on the way to the next behaviour required and started feeding at the opposite cone so Jedi was set up to touch it far more easily. It worked perfectly and our rate of success went from 20% to 100%!!
The video then goes on to show Jedi working for the first time on moving between 4 targets in a square. Again, when I feed at the target he has just touched, he will just stay there and keep touching it and I have to call him away. So I am basically setting him up for errors and making the training process more frustrating for both of us. By clicking when he touches the target and then moving on a little to deliver the reinforcement, I manage to help him get the idea of walking really smoothly between the cones with virtually no errors, to a point where I can actually have him walk two or three cones (with touches) before clicking and treating.
What is interesting from an emotional point of view is that Jedi started the 4 cone session very excited. He kept wanting to knock the cones repeatedly until they fell over and he was "dropping" (if you're not sure what that is... he had a 5th leg😉). But as soon as I changed the reinforcement delivery so that he kept moving on and succeeding at getting predictable rewards, he relaxed totally and while still highly motivated, the 5th leg vanished and he was able to think more clearly and learn better.
In the video I left a short bit in where I fiddled in the treat pouch to get the food ready in anticipation of Jedi touching the cone and managed to distract him from actually doing it. 🙈 This is a silly error that I frequently tell clients to stop doing! Just another example of how careful we need to be about reinforcement delivery!
Positive Reinforcement training means that the more our learners get right, the more we can reinforce and therefore, the quicker they learn! How we deliver reinforcement has a big impact on that!