Showing posts with label positive reinforcement training. Show all posts
Showing posts with label positive reinforcement training. Show all posts

Friday, October 5, 2018

Baby got back

Last night, after my first ride back, I was on pins and needles hoping I wouldn't feel pins and needles.  My back did get sore and I was in a panic, but some Alleve and a Salon Pas patch was all I needed.  And it was in a different spot than when I'd hurt my back.  Not riding for three weeks did leave me with a couple of sore spots after going back.  The stabilizers for my ankles were sore as well, since I apparently don't use those when running.

I got up this morning and was relieved to have no pain.  A little tight in the lower back, but that's not a surprise.  I headed to my lesson with a back brace on to see how I ride with that.  I also got to debut Theo's new boots.


Yes, they are glittery.  We found them at a tack shop during an adult ladies outing.  When I saw that they had front, back, AND bell boots that matched a saddle pad I have and were in his size?  They had to come home with me.

Trainer A was . . . well, she's gotten used to me at this point.  There were some sighs, but I've done worse. She still hates my hearts saddle pad more.

My lesson went well.  I definitely struggle with following in the back brace, so I'm hoping it's not something I need going forward.  I posted for most of this lesson, but did sit the canter.  I was guarding myself too much to sit the trot well and we don't want to block his forward.  I only sat for the lateral stuff where I just can't freaking manage all those body parts without my bum in the saddle.

With his hocks straightened out, we're full steam ahead on building a light, uphill canter with self carriage.  Today we started with shoulder in on the quarter line at the trot, making sure that he kept the forward while wrapping around my inside leg.  He likes to suck back, but he's getting the idea that he can stretch in the movement.  It's just a matter of confidence.

Then we swapped to cantering.  Pick up the canter, cut across the diagonal so we are counter cantering, then leg yield off the wall to the quarter line.  Swap leads, repeat.  Poor Theo, there was smoke coming out of his ears.  This is going to be our fight this winter.  We do things in walk and trot.  Canter?  We just canter, we don't actually do things.  He almost tripped over his own feet when he realized I wanted him to give me a real, leg crossing leg yield while cantering.  And all the counter canter work.  His counter canter has gotten very steady and he held it without any real resistance, so that's a big victory.  We don't do square turns yet, but we can go all around the ring without him inverting or trying to break.  He's comfortable with it and starting to push into the land of confirmed. 

He was stressed enough with this new, confusing exercise that we rewarded heavily.  It was straight up clicker training with a big release and a cookie when he stepped off the wall without bracing or dramatics.  It was walk to canter, canter the diagonal, a neck scritch for holding the counter canter on the short side, the leg yield on the long side, verbal click followed by halting for a treat.  A couple iterations of that and he was stepping off with no protests.  Each pass I'd ask for a couple more steps until we were making it to the quarter line.  I'll probably do one more session of rewarding every pass before I start to wean him off.  It's mentally challenging for him right now and I want him to stay light and positive about it.  It's not difficult physically, but he didn't expect it.

I think we'll be doing canter leg yields for awhile as we start getting the shoulder in and half pass at the canter started.  He needs to figure out that laterals are a thing at all three gaits.  It will also help us build a real collected canter without me getting in his face.  He'll sit down because it's the only way to free up his front end and go sideways.  Even our limited work today got his shoulders up more as he realized he had to do more than canter in a straight line.  He also had to halt promptly for those cookies.


Salon Pas patch is on my tight lower back but no sign of the nerve pain or pulled muscle, so I'm mentally feeling better.  Tomorrow his adult rider will have her flat lesson on him and I'll be resting my back.  She's learning about adjusting Theo and managing his tendency to fake a contact while bracing on the underside of his neck, so he'll be doing a lot of long and low while she learns what it feels like when he lifts his topline.  Gradually she'll learn how to bring him up without losing that lift.

Sunday is my half marathon.  Bib pick up is today.


AHHHHHHHHHHH.  *inhale*  AHHHHHHHHHHHHHHHHHHH.

I might be a bit nervous about that.

Monday, September 4, 2017

Release and reset

When you've had multiple negative rides in a row, there's really only one goal worth working toward:  have a positive ride.  I love the fact that I know mi papi well enough at this point that I can make a positive ride happen.  No matter what's going on, no matter the weather, no matter the causes, I can make a ride pleasant.  I have cookies.  So many cookies.

I got on today with about a quarter pound of treats in my pocket.  I would have a pleasant ride, damn it.

Theo's a positive reinforcement pony, a lot of the progress we've made has been built on pets and cookies.  He has no natural work ethic.  If you punish him, his reaction is basically 'wait until you see what I do to you for that'.  When things get difficult, I can fight with him (and probably lose) or I can go to what I know works.  No trainer I've ever worked with has supported my positive reinforcement training.  Ever.  No One.  They universally roll their eyes and accuse me of spoiling my horse.  So I just do it on my own and ignore the commentary. When they're pleased with the results, I smile and carry on.

After my Saturday ride, we were furious with each other and my left shoulder hurt so much I wasn't sure I was going to be able to get my saddle off.  I said stretch, he said no, I said do it now, he said no, and then I tried to push the matter.  He's stronger than me.  My left shoulder quit and he had me on the ropes.  I was ready to get those draw reins out for the simple reason that he was dragging through the contact, running me into the wall, and getting his rewarding release due to being stronger than me.  I made myself back up, compromise, end in a good spot, and go for a trail ride.  But I was so ready to put a big ol' bit in his mouth and tell him no more!  No more dragging on me until my shoulder fails because you know you can outlast me!  He shouldn't get a reward for pulling until I can't physically stop him!

Today I broke down the negative behavior.  Theo likes to stretch down and release, he's figured that out, but it's not a locked in behavior.  He doesn't really 'get' that I'm looking for that.  It's just an exercise, not a behavior to offer.  I decided to lock that behavior in.  For our warm up, I'd get a big, chewing, to the ankles stretch (with his nose in) in the walk and then he'd get a pet and a cookie.  Repeat about ten times.  Then I sat and waited.  Sure enough, with no action from me, he gave me a big stretch and chew, hanging out there until I pet him and gave him a treat.  And on we went, him offering the behavior and me rewarding it.  I weaned him off of the cookies fairly quickly since it's a behavior he knows and added the cue back in so he did something other than drop his head and chew, but the trick was to get him to do it on his own, to offer the behavior rather than push him into it.  He's the king of offering behaviors when he thinks he'll get a cookie for it.  If he sees it as a thing that I will reward for, he will offer it happily.

I was able to get him walking, trotting, and cantering with a big stretch and a light, non-painful for me contact pretty easily.  He enjoys it, he enjoys being told how smart he is, and he enjoys getting cookies.  It was wonderful to see him chewing away and lots of foam around his bit in both directions.  He was relaxed and happy.  He was also behind my leg.

Now this is the part that's probably going to take me weeks.  I need to add forward without losing the chill.  I played with it today, finding the tipping point between chill and negative tension.  I don't think Theo is good at managing positive tension yet and it all tends to tip over into negative tension.  I start to create positive tension, I start to feel power and forward, and then it snaps and I have a braced, non-existent connection with fast feet.  It's a pretty concrete reaction.  We're good, we're good, we're good, we are DONE.

Okay, I can work with that.  He has a point where he mentally feels like he can't take any more and he quits.  It's ridiculously close to going as slow as molasses in January, but I'm not here to judge.  Okay, yes, I'm totally here to judge, but it's where it is and I need to work with it.  I ditched my whip because that brings in too much negative right now.  I tried gradually building and when he pushed past his previous wall, he got big pats and I removed the pressure.  I wanted him to see that I won't, in fact, press forever.  And with each iteration, I want him to go a little further.  Push a bit more, handle a bit more pressure before I back off again.  And when he just coped?  Pats!  Many pats!  Cookies!  All the good things.

It's going to be slow.  Very slow.  I can get quick feet and power now, but it's braced and pissy or it's due to the influence of another.  Getting him to give me that power without the bracing and negative tension is going to take a lot of time and work.  And cookies and pats.  He doesn't like to work hard, he needs to know that there's something in it for him.  Any behavior can be shaped if you're patient enough and have enough rewards.  I will buy those treats in bulk if that's what it takes.  I did have a few moments of a completely tracking up pony that was also stretching nicely to my hand without my shoulders screaming in pain.  When I got off today, my shoulder felt okay.  And he was like jello again, relaxed and happy.

I'm glad I chose to not get the draw reins out.  This was a better lesson.  I want him to offer the behavior, not be shoved into it.  I can't fall into the trap of taking the mechanical advantage.  If my shoulder is too weak to ride in that heavy of a contact, then I'm just going to have to find another way.  He will always be stronger than me.  I better be smarter than him.  You get more flies with honey.

Monday, January 2, 2017

The positive reinforcement pony

Unlike the dog training world, positive reinforcement isn't seen as much with horses.  Pats on the neck and verbal rewards are pretty much it and it's haphazard, forgotten in the heat of the moment when working on a concept.  Bits, spurs, and crops are mainstream, but giving your horse a cookie while training under saddle is seen as weird.  Leaping off of my horse and loosening his girth to mark a break through?  What on earth am I doing?  I have a lot of experts that think I'm ruining my horse when they see me whip out cookies in the midst of a training session.  You're spoiling him!  Your horse is going to get rude and nippy!  He won't work for you if you don't have a treat for him!  He won't respect you!


Ladies, gentlemen, and those that choose to not use those labels, I am here to tell you what it is actually like to have a horse that is a positive reinforcement pony.

First, some definitions:
Positive reinforcement = adding something positive as a reward, like a cookie
Negative reinforcement = taking away something negative as a reward, like releasing pressure on the bit, but this can also happen when a horse learns to evade something (and we wonder why horses learn to stop, its very rewarding to avoid the effort if the effort is seen as a negative)

Positive punishment = adding something negative as a punishment, such as a smack with a crop
Negative punishment = taking away something positive as a punishment, such as me turning away when Theo's being pushy

Your horse will get rude and pushy!

No more than any other horse.  Any horse can learn to be rude and pushy when they think a treat is coming.  Having the process of earning and receiving a cookie being formalized can actually help with pushy behavior.  Theo is not allowed to push for a treat.  He begs when he thinks there is a treat coming and he's not working (whickers, pricks his ears, arches his neck, acts cute), but he is not allowed to try to take one unless it's offered.  He is not allowed to reach for a hand that's not offered or step into a person's space.  It's important that he not get pushy or nippy.  Trainer A is very appreciative.  At the end of a lesson, she usually shares an apple or Kind bar with him.  He will arch his neck and beg, but he won't touch her, crowd her, or try to take something from her unless it's deliberately offered.  It keeps fingers and toes safe.  Theo is more polite with treats than many horses I've met because he's been taught how he has to behave in order to get one.

He won't work if you don't have a treat!

First, I always have a treat.  Always.  I keep them in my pocket at shows, when trailering, when I bring him in from the field.  No matter what situation, there is a cookie in my pocket.  And all of my breeches have pockets.  It's just how I do things.  It helps with a lot of situations, even ones with horses that aren't mine.  More than once I've had someone ask me for a cookie while loading a horse in a trailer or handling a situation like clipping.  I always have something in my pocket.  Theo assumes that I have a a cookie at all times.  It's been very helpful at shows, since he assumes he could get a reward at any time.  Lots of horses get 'ring smart' and realize they won't get disciplined in front of the judge.  I sure don't want to use my whip while in competition.  It's not a good impression for the judge.  But judges, in my experience, think handing a treat to your horse is cute or totally not worth noting.  He's not 'ring smart' because I have rewarded him in the ring at competitions.  He associates the judge's booth with cookies because I always have them just in case he does something amazing like go past without broncing. 

Second, he works regardless of me handing him a cookie.  He's on a random reward schedule, so if he doesn't get one, he assumes he needs to do more to get his cookie.  He doesn't have xray vision, he doesn't know if a treat is in my pocket.  He assumes I have one available, and if he does something good, he'll get a reward.  Random reward schedule is very important for this.  It keeps him trying even if he doesn't get a cookie immediately for any given action.  There's a very specific verbal marker that represents a cookie being delivered.  So long as I don't use that verbal marker when I don't actually have a reward, he will work with pets and cooing until it's time for his treat.  Pets and cooing serve as information for him, he's on the right track.  At the end of a string of correctly executed maneuvers, he'll get a treat.  He's just not sure where the end of that string of maneuvers is.

He's working for the cookie, not you!

One, check your ego at the door.  Who cares?  Horses need more of a reward than some mythical bond with a strange, bipedal omnivore that doesn't speak their language and controls every part of their lives.  Your horse is probably not as in love with you as you think.  Your approval is not what he lies awake at night and wishes for.  Two, I'm the one providing the cookie.  He's begging me, not the cookie.  It's a solid, concrete reward.  Petting and verbal rewards are still a huge part of working with him.  I coo to him, talk to him, scratch along his crest, and give him quick pats with a release all the time.  I use cookies to very specifically mark a behavior when I'm developing it or reward him for overall effort.

I have a degree in psychology.  I taught rats to do a lot of things using nothing but operant conditioning as part of my college work.  I have a fair bit of training for this skill between dog training and rat work.  Timing is everything.  When I first introduce something difficult to Theo, I'll shape his behavior.  He gives me an honest try, I mark the behavior with a specific word ('good boy'), followed by a cookie.  I do this a couple times to get him rolling.  Then I expect more from him to get the same reward.  For shoulder in, he got marked for the first offer of shoulder fore to start.  Then I asked for more angle before I'd give him a reward.  Then I weaned him off of the reward, asking for several reps to get a reward.  Now it's old hat and he just does the movement.  He'll get a reward if he really blows my socks off, like doing it at the canter, but for the most part he understands and gives me the behavior without needing a cookie.  He gets a pat on the neck, a verbal reward, and we keep working.

He won't respect you!

Make no mistake, Theo respects me.  Just because I formally mark and reward behaviors doesn't mean that's the only thing I use.  He's a thousand pound animal.  He tries to push me or bite me, I carry a dressage whip for a reason.  However, because of his personality, after he's been sharply corrected, I will give him a chance to earn a reward very quickly.  He gets rude during a hand walk.  I snap the chain, back him up, and give him a verbal correction.  Then I turn, act like nothing happened, and find something to play the 'touch it' game.  Lesson learned, but I also end the fight.  My horse respects me.  I'm going to guess this myth comes from the natural horsemanship world.  Yes, I've seen a lot of horses with Parelli levels that have no respect for humans.  This isn't natural horsemanship.  It's conditioning, pure and simple.

There's a lot more myths and variations, including concerns about what I'm doing to my bits and bridles.  But these are the ones I hear the most.

It's proven that positive reinforcement is more effective than positive punishment when learning.  Positive punishment is associated with increased stress in animals (and people).  If I have an option to use a technique that will lower stress, increase motivation, and increase the likelihood of the behavior being offered, I'll take it.

I don't recommend every single horse and rider pair jump into positive reinforcement with both feet.  It takes some study for the rider to really get it.  It takes careful timing and practice.  It also takes practice to manage the cookies, especially with gloves.  You have to consider your wardrobe in terms of pocket space.  My poor washer and dryer have processed a lot of forgotten horse cookies.  I don't use a treat pouch since that gives him something to visually cue on for the presence of cookies.  Cookies should be a source of mystery, magically appearing in the rider's hand from hammerspace or some other dimension.

Some horses get obsessed with the reward.  Same as dogs, some can't work with something that they want so badly.  It becomes a distraction rather than a help.  But I think positive reinforcement has a place with horses and is often overlooked or dismissed when it should be a common tool in the toolbox.  Theo is highly food motivated and used to have a terrible work ethic.  Really, what was in it for him?  He got kicked either way and work was hard.  He hated work.  Adding something he loved improved his opinion.  He now associates work with rewards to be earned, rewards that are more motivating than the random pets of someone that's also making him do the nonsense.  I could have tried kicking him more, getting bigger spurs, used the whip, any number of negative reinforcements, but we're all happier with the positive reinforcements. 

In the average ride, Theo gets about three to five rewards.  Despite the jokes, I'm not a pez dispenser.  The number sky rockets when we're adding something new that needs heavy reinforcement (turn on the haunches for the first couple weeks, teaching him 'touch it', right now it's his canter to walk transition and flying changes), but he typically gets one when I mount (he never moves a foot, even when I'm mounting on trails or at hunter paces), one after warm up, one after his hard work is done, and one when I dismount.  Two of those cookies are to reinforce his behavior when I'm mounting and dismounting.  It has saved my bacon several times that he stands completely still for both maneuvers no matter where we are, waiting for his cookie.  When we're jumping, he'll get one when he's completed a full course or when he's on break between sets with the grid. 

Dog training manuals give excellent advice on positive reinforcement, for anyone that wants to consider adding a couple of sugar cubes into their regime.  Horses, as a rule, I don't consider to be candidates for strictly positive reinforcement training because they're so big and sometimes, just sometimes, I need immediate obedience.  Like when standing on my foot.  Horses also do an awful lot of communicating in the herd with shoving and biting.  But the equestrian culture leans too far toward negative reinforcement.  I think horses need a reward to work toward, something more concrete than our approval.  The most powerful reward I have is to vault out of the saddle and loosen the girth.  That should say something in regards to how horses view their work.

I do see dressage riders presenting their horses with a sugar cube at the end of a test and it makes me feel less odd.  Maybe we'll find a spot for equestrian culture that's not so dependent on negative reinforcement or positive punishment.  I can hope so.  Some horses just need a bit more positive motivation than others.