Clicker training Part 1
| February 25, 2016 | Posted by Melinda under Uncategorized |
I started clicker training to give ML a job while I was pregnant. I figured if either of my horses was going to have success doing clicker training it would be the people oriented one that was eager to please. The other reason ML was a good candidate is because she has a tendency to get mouthy and in my space – 2 things I haven’t been successful at entirely eliminating using the other tools in my tool chest – it’s better, it’s infrequent, but it’s still there AND she will do it with other people. I figured that clicker training would make it better or worse and would be a good test of the most common complaints I hear about clicker training from regular people (mouthy pushy horses), but what the *fundamentalist guru’s of the method insist will get better.
*more on these people later
First we need to make sure we are speaking the same language.
“Clicker training” is a method of positive reinforcement where you ADD something POSITIVE to reward a behavior you want. This can be a click that is paired with a treat, a verbal praise, a scratch, a pet….whatever the animal finds rewarding and motivating. I’m lumping all of this under “clicker training” for the purposes of this post because in the end it’s all the same kind of training with all the same pitfalls and benefits with the individual variation of what reward/marker system you decide to use dependent on the animal.
The key to this tool is the animal earns something it finds rewarding for a behavior so it’s likely to repeat the behavior in the future.
You want to be very careful to CONTROL the REWARD the horse is getting and CONTROL the BEHAVIOR that is being reinforced. You can *think* you are using positive reinforcement to teach a horse something, but because horses are exquisitely sensitive to the *pressure and release type of training it’s easy to inadvertently be rewarding something else completely because you are applying and releasing pressure – even if that is something as simple as loss of eye contact or touch or anything else that the horse considers stressful or “pressure”.
*REMOVING something when the horse does something right is technically called negative reinforcement, which is also a perfectly valid training tool for horses.
That is why I chose to use the *traditional clicker and small treats.
*I like and use this one. In addition to having a really distinctive but not ear deafening click and a nice hand feel, I like the button because I can put it under my foot and click it with my toes if I need to teach a behavior that requires me to have both hands free – which has been important for some of Tess’s tricks.
Clicking lets me “mark” the behavior precisely in a way I just can’t verbally. When I see the behavior it’s a lightning fast gut reaction to press that click. In most cases I’m verbally praising the effort (“Good girl!!!!!”) but there’s no confusion for the animal what precise moment in time is responsible for the praise.
I use treats for 2 reasons.
- They are fast – repetition is important is important to allow the animal to put together pieces of the behavior so I need a reward that only takes a second before we can try again.
- Food is very rewarding – and with the exception of 1 dog, if I evaluate every animal (dog, bird, horse) I own, whether or not they have a history of being trained with treats, treats/food rank WAY higher than affection or praise. In many cases I’m asking the animal to do something that is very hard for it, either physically or mentally requiring lots of willpower. Rewarding that effort sufficiently is key to success. It doesn’t matter whether you think that praise should be more rewarding for the animal – you have to work with what the *animal finds rewarding.
*This can change over time or in different circumstances. When I first started with Tess she only worked for treats. Now she will work for treats, toys, verbal praise, or freedom depending on the circumstance. It’s about using the reward to build a language, and then using the reward to increase the value of the other things.
The biggest weakness of clicker training is the sheer number of fundamentalists out there who are publicizing it.
Fundamentalists are the all or nothing crowd. They insist that clicker training/positive reinforcement is the only tool you will ever need and that anyone can apply this method to all horses/animals and have success.
Here’s my two cents
- You still need to have good to excellent timing.
- You still need to be able to read body language
- You need to be able to think 2 or 3 steps ahead of what you are actually training.
- You need to be exquisitely aware of what you are rewarding.
Most of this can be improved over time and with practice. However the feel needed for training in general cannot always be taught and some people just don’t have what it takes.
For folks that have a good grasp of the above concepts, the biggest mistake I see is not understanding the concept of arousal/threshold and trying to clicker train horses/dogs that are way too distracted or nervous. “Arousal” in behavior means how excited/focused/stressed the animal is in a particular situation or exposed to a particular thing. A little arousal is a good thing – enough to keep the animal interested and engaged. A lot is not. The line between “focused enough to learn” and “over threshold and not able to learn” is a lot narrower then you think. A key concept is over threshold = NOT ABLE TO LEARN AT ALL OR EFFECTIVELY. Or, has a very good chance of learning the wrong lesson. Which I generally try to avoid while training any of my animals but especially those big enough to kill me ;).
What does that threshold line look like? Here’s some examples.
- Focused and appropriate: interested in object and wants to look at it, but with very little encouragement is willing to refocus on you.
- Over threshold, not able to learn: Staring at object, will only glance at you for a second before returning to look at object.
Clicker training is primarily used for desensitization and counter conditioning that means we are working at low arousal and under threshold. This is DIFFERENT from pressure/release/overshadowing (which I’ll talk about in a different post related to more AERC convention behavior lecture stuff).
Clicker training’s strength is it’s ability to not only teach the horse the correct response, it also changes the emotional response of the horse to that stimulus from negative or neutral to positive. Pressure/release (negative reinforcement) where the reward is the release of pressure also does an exquisite job of teaching the horse the right response, but it lacks the ability (most of the time) to change the horse’s emotional response to positive.
**There’s some grey areas but I’m talking generalities OK?
Here’s my bottom line:
Clicker training is a great tool to have in the tool box. Used in conjunction with other training techniques such as pressure/release, overshadowing should give you a really solid partner that also has a lot of fun hanging out with you and enjoys their job.
It is not the end all magical rainbow farting unicorn philosophy for training horses.
I’m not doing a “how to clicker train your horse” tutorial because if you have half a brain and a decent handle on the points above you can wade through the YouTube videos that are out there and get decent results.
Moving along to what I’ve done with my horses!
So far I have to say clicker training is a success.
MerryLegs
I’ve done the simple intro stuff. Nothing fancy, just the very basics of clicker training that you see all over the “how to’s” on the internet.. We worked on the requirements for getting and taking a treat, and some targeting.
With just the basics MerryLegs has become better at taking treats, is in my space less often, and seems to appreciate the structured program of how she can earn treats. She’s less crabby, less pushy, and less likely to engage in risky behaviors that might result in her inadvertently getting a treat or rewarded (rewards aren’t always treats and not always intentional on the human’s part!).
It’s been 6-8 weeks since I’ve done any clicker training with her or fed her treats. When I visited her last week with Paige she continued to be very respectful and have perfect ground manners. The lesson stuck!
Future plans? MerryLegs is very comfortable laying down with me near her and lets me rub all over her while she reclines, something I’ve NEVER had a horse allow me to do. I think I have a good chance of teaching her to lay down on command, so that’s what we are working towards.
Here’s some lessons learned and tips from working with ML
- Little circles of sliced baby carrots work way better than the pellets/feed/treats that most of the videos seem to be using.
- I did more of a rapid fire treat-clicker pairing more similar to dogs then what they showed in most of the horse videos I watched. I made sure I was rewarding her away from me and did a click-treat in such fast procession she didn’t have time to do something naughty between them so I could just focus on pairing that association in her mind.
- I was VERY consistent about NEVER rewarding her while she was in my space. Even if that meant she didn’t get a treat when I clicked because she came into my space anticipating a treat faster than I could provide it in the proper spot. It didn’t happen very often because I have good timing, and I tried to set her up for success early on. Since I had made sure to pair the clicker-treat I didn’t worry about a few clicks without treats very occasionally.
- I used my voice to praise and encourage. If she was out of line (in my space, nibbling) I would “accidentally” let her bump into my elbow as I turned or blocked her with my body language. If she wasn’t doing anything dangerous (getting in my space) and was trying to figure it out, I generally let her natural curiosity and drive for the reward motivate her. If she became unmotivated and needed lots of prompting, it was a sign I moved too fast, didn’t break up the task in small enough pieces, or the session had gone too long.
- In some ways it’s very similar to clicker training dog, except I think your game have to be a little bit better. Your timing has to be impeccable and you have to be really careful about what your rewarding how when your rewarding it. If you’ve never clicker train something I’m not sure a horse is the best thing to start with.
- It’s an excellent thing to do with the baby in the pack.
Farley
I was bored and motivated and somehow found myself with Farley on the end of the lead line and thought “why not try and see what happens?” Farley doesn’t isn’t mouthy, doesn’t get in my space, and maintains a professional relationship that has a deep affection as a result of being partners for so long. She’s not a problem horse and she’s definitely not a Merrylegs.
Since she doesn’t have ML’s drive and curiosity I was surprised it didn’t take her long to figure out and was motivated to do the right thing for my treats (turn her head away). She was WAY more engaged in the game more quickly then I thought was possible. Perhaps I had underestimated my little piglet’s drive for goodies? Or willingness to play stupid games with me? Or clicker training’s ability to get horses like Farley to engage with their humans more?
I’ve done similar activities with Farley as I have with ML – targeting, how to earn and receive treats. It has not made her mouthy and she is not mugging me. Future? There’s a couple of things such as her girthiness that I’m going to try to fix using clicker training.
Bottom line?
It took me exactly two 20 minute sessions to completely train the mouthiness out of MerryLegs, after nothing else worked. I have a feeling that it’s going to take a similarly short time to get rid of Farley’s girthiness in a similar set of sessions. Those are two chronic issues that I’ve never been able to totally get rid of in “traditional training” – either pressure and release or adding something negative like a vocal correction or a physical one. I’m also looking forward to seeing whether I can train some behaviors using positive reinforcement (reward training) that I’ve never quite figured out how to do using negative reinforcement (pressure/release).
When I’ve done something besides the basics I’ll give my Dear Reader an update. Please let me know if you have any questions and I’ll do my best to answer them. If I can get some good videos of me clicker training either the dogs or the horses I’ll post them.
Interesting analysis/report — thank you! I’m pretty committed to clicker/marker training with the dog, but never felt much need or desire to try it on the horse; it’s fascinating to see how it translates to other species.
(I like the Star Mark clicker, too, but find it a liiiiittle slow in comparison to the standard box clickers.)
I totally agree with your two-cents-worth of four points and would add that timing can be a variable thing from method to method even for a single person. I had uncannily good timing with the clicker from pretty much word one…which was really weird, because I was crap with, for example, collar corrections. I’ve watched other folks have the opposite experience: really skileld Koehler method trainers who struggled to mark the desired behaviors (independent, even, from being able to spot it). And some folks are equally (um)comfortable either way, of course! But that was a real surprise to me.
I echo your sentiments exactly – really enjoy seeing how far I can take the training with the dogs, but never felt the desire to do it with the horses. Finally did it out of curiosity but I suspect/know it will never be as big a part of my horse training as it is with my dogs.
Clicker training had been a total failure with my bird by the way. Mostly because I can’t find the right reward that can be enjoyed quickly and is motivating.
The most interesting exercise I’ve done that really illustrates timing was something I did in vet school. We were given a hand full of m and ms and a clicker and were told to clicker train each other for a particular behavior. It was HILARIOUS. No talking allowed and I have to admit that the hand gestures my fellow students made were not that useful. It was SO HARD to figure out what I was getting rewarded for unless the timing was perfect on the clicker.
I always teach “look away” as the very first thing for a horse, no matter how mouthy (or not).
Recently I’ve been working to teach Fiddle to fetch a frisbee (the flexible dog-kind) and she is SO reluctant to take the thing in her mouth, even though she will pick up the rubber duckie. I finally figured out: the flex-frisbee “feels” like clothing in her mouth! And touching clothing with her mouth is BAD (I had to teach that, she was a biter when I got her) so now we are fine-tuning old training, which is good practice for both of us.
Okay, that vet school exercise sounds AMAZING! And really useful. Now I kinda want to try it.
If I ever do a lecture or a clinic where I can possibly integrate it I’m totally going to do it!