Barbara asks some questions that are hot buttons for a lot of people coming from the school of “clicker training”. She is inquiring about the use of the bridge signal(clicker) and not following every single bridge with food. In the marine mammal industry we don’t call it clicker training…in fact most people don’t use clickers. The term was derived and introduced into the dog world some years ago. With it came a set of rules that aren’t used in the marine mammal training industry. That has lead to some confusion as to what is “allowed” when using positive reinforcement. At Sea World we studied and utilized behavioral principles and applied learning theory. The marine mammal industry is on the cutting edge of positive reinforcement training. I know from a previous conversation that some of the rules of clicker training, as it applies to dogs, were created for the neophyte dog owner/trainer. I understand why they were implemented but I also realize that we have come a long way since then. People are now much further along in their education and understanding so they can easily handle some of the more “advanced” concepts of training. I address the questions and give more info in this video clip, but I wanted to touch on some of the points here in the post as well.
One is the concept of not feeding every bridge signal (clicker). The bridge signal/clicker is a conditioned reinforcer also known as a secondary reinforcer. This means we have given the clicker value through classic conditioning (think Pavlov’s dogs). Once that happens it then serves as a reinforcer, all by itself. Occasionally reinforcing with a scratch, game, toy, turn out, activity, another signal or any other conditioned/secondary reinforcer does not diminish the value of the clicker or any other bridge signal. If the ratio became out of balance then that would be a different story, but as long as the ratio is skewed toward most of the bridges being followed by food we keep the value quite high. There are plenty of benefits to using secondary reinforcers and it definitely enhances your relationship.
Next Barbara brings up cues. There is a “rule” that you can’t add a cue until the behavior is almost completely trained. That is not necessarily true. In fact in most cases we are instituting some sort of cue (discriminative stimulus) as we start to train a new behavior…otherwise how would they have a clue what to do? So you have a cue from the get go. Often times I will slightly modify the original cue or simply make it more and more subtle. I give the example of my spin cue in the video. At first it was a point to the target, pretty quickly I started adding two fingers pointing to the target. I kept the same signal that I used from day one but it got smaller and smaller so that it is subtle, this helps to keep them more watchful. If I need to take a step back to remind them, I am still using the same working signal that was a part of the initial lessons. It is easy to slowly get bigger to remind them. If I want a different signal altogether, I will start to pair it with the working cue as soon as they start getting successful approximations. With a target to guide them through it starts pretty darn quick. Free shaping and capturing will be a different process. Since there will likely be more guesswork on the horse’s part, I will add the cue later in these cases.
The third thing that Barbara asks about is the use of a magnitude reinforcement…AKA jackpot feed. She wanted to know if I use a jackpot on Mint’s back up (video: “Now that’s a backup!!” on my YouTube Channel) I definitely use jackpots during the training process, especially for the smallest improvements when working on a new behavior. This seems to keep them highly motivated to work through the rough spots. I actually use the magnitude feed quite a bit. I like to make a big impact and allow latent learning to take effect. I believe quite strongly in having short and sweet sessions and I end every session with a magnitude reinforcement. Well, let me clarify, I end every successful session with a magnitude reinforcement.
Okay, that may about cover it, the rest is in the video…If you have any questions or comments please let me know!
