Clicker training horses is a very interesting topic. Lots of people have very strong opinions for or against it. Personally, I was interested in it having seen some neat things that people had trained for, however I wasn't that motivated to try it since traditional methods worked for me and it's what I'd always done. Two bloggers that I know of use clicker training though, first is Matt the Cowpony who I've seen has a clicker-trained spin on the ground (really neat to watch). Secondly, just recently Bad Eventer posted about her success in using clicker training with her monster pony. Neat stuff!
Enter my animal learning class where our term project is to clicker train an animal. I am actually really enjoying this class contrary to how I thought it would be. I just find training so fascinating and I've loved training animals since I was little. Our family dog knows at least 15 different commands as a result of having to hang out at my family's store every day when I was between 8 and 10 years old. Boredom often has great results, haha.
First, some terminology. I didn't know the distinction between these terms before this class, though we so often use them in general life incorrectly.
Positive = adding something to the environment. Not good or bad, just the addition of a stimuli or object.
Negative = removing something from the environment.
Punishment = anything that makes an animal perform the specified action less
Reinforcement = anything that makes an animal perform the specified action more
So therefore
positive reinforcement is adding something to the environment (such as treats or praise) to make the animal perform the behavior more often,
negative reinforcement is removing something from the environment (such as pressure or an annoying sound) to get an animal to perform a behavior more often.
Negative punishment is removing something (such as a child's favorite toy) to make them perform a behavior less often and
positive punishment is adding something (an aversive stimuli such as a shock or slap) to make them perform the behavior less often.
The majority of horse training is done through negative reinforcement. Put pressure on the reins to get the horse to give to the bit, pressure on the sides with your calf to get the horse to move forward or turn, etc. There's some positive reinforcement through praise, but mostly it's negative. The reward is removing the pressure. For the most part I don't see too much punishment in training horses and it's something that is generally avoided in my experience, but the most common I've experienced is smacking a horse with a crop for refusing a fence.
Clicker training is all about positive reinforcement. You don't have to use a clicker, but the reason it is used is to mark a positive behavior as soon as it happens. I think many people have experienced training a dog to sit and as soon as they do, you reach down to give them a treat and they immediately jump up to get the food, thus getting rewarded for standing up. Usually in training dogs (without a clicker) you still mark the good behavior as soon as it happens with a "good dog!" so they know exactly what is getting them that reward.
The most common arguments I've heard against clicker training horses are that the horse will become pushy and rude to get the treats (common argument against hand-feeding treats), that the horse will perform the trained behavior even when it's not asked for, and that the horse will only perform for treats, so if you don't have the treat forget it.
My opinion (notice it's an opinion!) is that these are both very untrue if done correctly.
You can read another completely different opinion here. (Note: this was my first introduction to a view on clicker training horses several years ago. I'd never really considered using clicker training on horses before. Somehow I ran across this woman's blog. I really hope that this piece is intended to be more sarcastic than educational).
Anyways. The first argument against clicker training:
the horse will become pushy and rude to get treats. They really shouldn't actually. I was slightly worried about this because Jetta is, well, apt to become just that: rude and pushy. But, the thing is that after the first session, where she sniffed me up and down and rifled through my clothes pockets to get the treats (which she got reprimanded for "positive punishment" because that is rude and not allowed) she quickly figured out the ONLY way to get a treat was to perform the behavior I wanted. In this case, she only got a treat if she touched the traffic cone and heard a click. And she's not going to "take off my hand" trying to get the treat, because as soon as I click to mark the behavior I basically shoved the treat in her face where she politely took it. There's no opportunity to bite even if she thought that was appropriate behavior, which she doesn't. If I had to struggle to get the treat out of my pocket, Jetta waited somewhat patiently because she knew that she's not allowed to beg for a treat. That doesn't get her anywhere.
Second argument,
the horse will perform the trained behavior even when it's not asked for. If that is the case, then by definition, this behavior is not correctly or completely trained. A behavior is only completely trained when the animal performs it immediately upon command, it never occurs when not commanded, it never occurs in response to a different command (such as telling a dog to lie down and they sit instead), and no other behavior occurs in response to this command (so telling a dog to sit wouldn't result in them jumping up). If a behavior is completely trained, then the animal knows that they will receive no reinforcement for a behavior that is unasked for. In the beginning of training, yes the animal must offer behaviors that you will choose to reinforce, but by the end of training you can attach a verbal, visual or tactile cue and then animal will learn to ONLY perform that behavior when that cue is given.
Third argument,
the horse will only perform for treats so if you don't have treats, forget it. Also a very common thought. What I never realized is that there are two basic schedules of reinforcement, meaning how often you reinforce for something. There is fixed ratio (FR) and variable ratio (VR). Initially, you'll use an FR1 schedule of reinforcement, meaning that every time the behavior is performed, the animal will be reinforced. Other types of FR schedules could be FR5, meaning that the behavior has to be performed five times to get one reinforcement. A VR schedule of reinforcement means that the number of times the behavior must be performed to get reinforced varies each time. A VR5 schedule means that the behavior must be performed an AVERAGE of 5 times. This could mean they do it once to get a treat or do it ten times to get a treat. They never know how many times it will take to get the treat. The great thing about this is that once a behavior is established, you can change from an FR1 schedule to a variable schedule. VR schedules are great because they create behaviors that are also really resistant to
extinction, meaning if you forget treats, that behavior will still be performed when asked because eventually the animal knows that they will get reinforced.
Extinction is the decrease or disappearance of a behavior that was previously reinforced once reinforcement stops. So if a horse is reinforced for, let's say, laying down. All of the sudden if you stop giving them treats whenever they lay down then eventually that behavior will decrease or stop altogether. However, it's not something that happens immediately. Commonly there is an "extinction burst" where the animal will increase the behavior at first to try to get reinforced before slowly decreasing the behavior. This is something that I think a lot of people don't realize, but makes a lot of sense if you think about it. I think a great example is with little kids. If you're on a road trip and they ask "Are we there yet?" and you respond, giving them attention (reinforcement) then they will continue to ask this question throughout the trip. However, if you all of the sudden stop responding to them and start ignoring the question to try and get them to stop bugging you, at first they will increase their rate of asking, often escalating their obnoxiousness levels trying to get that attention back. However, if you wait it out long enough, they'll lose interest and eventually learn to stop asking because they won't get attention that way.
So that's my basic overview of clicker training. I'm still having a blast doing it with Jetta and plan to continue to use it after the class is over. Next thing on the list is teaching her to stand still in the wash stall cross ties. Still not something we've mastered over the past six years...
If you're interested in reading more, one of the required books for the class I'm in is Karen Pryor's "Don't Shoot the Dog". Easy to understand and an easy read.
