We’ve been having a good discussion recently on one of the clicker training facebook groups about how to reduce or fade out the clicks and treats when clicker training.
When people start clicker training (What is clicker training?), many are afraid that they are doomed to always carry around a clicker and a bag of treats, clicking for every single simple behavior.
This, of course, is certainly not the case, as many people who clicker train are able to reach high levels of performance, such as in competitive sports where it is impossible to click and treat at every step.
But how does this work?
Once a behavior is well learned and the animal enjoys doing it, you can start to string together longer and longer sequences of behaviors. It’s not that you’re eliminating the clicks, but, instead, you can start asking for more behavior before each click. The glue that will hold multiple behaviors together will be your cues between the behaviors.
This works because cues for behaviors taught with positive reinforcement that animal enjoys can come to serve as not only cues for the next behavior, but as conditioned reinforcers for the previous behavior. Does that make sense?
Here’s an example– pretend you taught your horse to back up on cue using clicker training and he really likes doing this to earn a click. If backing up was taught using positive reinforcement and the horse enjoys doing it, the horse will be “looking” for ways to get you to give the back up cue.
So, you can use your cue for back up to reinforce other behavior(s) that you want. If the horse lowers his head — rather than click and treat, you could give the cue for back up, and once the horse backs, then click/treat. This creates a chain of two behaviors. For more information about how this works, check out my post on loopy training.
This is a really important discussion because many of the clicker training books gloss over this part and end up with a confused animal who thinks you're not C/T-ing because they didn't do the behavior right. Bob Bailey said, “Variable reinforcement creates variable behavior”. Most people try to go way too fast with what you're describing above, and for the average pet owner the nuances of this may be too much to bother with. For the average pet owner, the idea of continuing to mark and reward behaviors as a way to say “Yes, thank you, that was it” is not unreasonable and reduces the risk of their messing up this Premack principal stuff you describe above.
Hi Diane,
Thanks so much for the thoughtful comment. You make some great points.
One other thing that I think pet owners can do is find ways to incorporate the behaviors they are teaching into their daily routines so that the owner can use “life rewards” to reward the behavior, rather than a click and treat. For example, asking a dog to sit before opening a door to go outside or asking a dog to sit before putting the dog’s food bowl on the floor. Finding these opportunities is one way to still reward the behavior, but at the same time start reducing the clicks and treats.
I really like the Bob Bailey quote that you included in your comment. 🙂
cheers,
Mary
Yes, you make sense! Thanks for an important reminder about clicker training. Joy