Monday, April 22, 2013

How can you tell you fixed it if if it didn't happen?

There's a great deal of discussion in the air now regarding the evaluation process for public school teachers.  The main issue appears to be whether or not it's fair to judge a teacher on the test scores of her students.  One school of thought (espoused by non-teachers, mostly) believes it's the best possible way--outcome-based criteria are rock solid, they contend--as only good teachers can help their students as a group to improve their skills.  Another school of thought (teachers', mostly) focuses on the idea that kids will be kids, and they're as likely to blow off the test because they just don't care, because they're overly stressed, or because they know their much-despised teacher's job depends on their results.  Still another group (philosophers and economists, I suppose) wonder if it's possible to create improvement in a student whose abilities have reached their peak, and whether it's even possible to determine that they have or do not have the ability to move up.  This becomes most pertinent at the upper levels as there's only so much one can do to improve IQ in teens even if one can pry the smart phones from their numb fingers.

At the same time, there's been testy discourse over possible changes in firearm regulations.  The "Let's all go shooting!" club says if everyone owned a gun, there wouldn't be any gun violence because we'd scare each other too much (or something along those lines), while the "Oh, no!  A GUN!" club says where there are fewer guns there are fewer shootings.  In reality it's hard to prove either side because, well, how can you prove that what didn't happen was in any way related to what did?

An area that's been a little neglected is the most esoteric statistical and the most hypothetical (but possibly the most rational) of world views.  How can one say that the teacher who didn't manage to improve her students' test scores didn't prevent them from dropping?  How can one prove that the home invasion where the homeowner didn't get shot was safer because the homeowner was also a gun owner (unless, of course, he shot the invader, which is a different scenario entirely)?  How can you prove what hasn't happened and lay responsibility for the non-event with any accuracy?

It's obvious that there's a connection to horse training in this last question.  When you've spent massed practice time trying to get Widget to stop fussing in the cross-ties, and after six months he still manages to grab your sleeve or leave tooth marks on your glove, who's to say that he wouldn't have been worse if you hadn't done the work with him?

Ethel running alongside the lawn tractor catching grass and bugs
is a learned behavior...or is it?  And what behavior might she
have displayed if she'd never seen a lawn tractor?
We'll never know.

I took this to my Upper Pasture Advisory Council for debate.  This time all hooves pointed at Dakota, the big-butted Appy, who is not a cribber, but a wood-chewer and surface-licker extraordinaire and whose efforts to dissolve, by dint of the longest tongue I've ever seen, the edges of his buckets have left the plastic festooned with gobs of well-chewed sweet feed.  Every anodized gate has a half-moon of rust in the middle of the second rail from the top courtesy of the highly acidic nature of his spit and the devotion to duty that he has displayed over eight years on the premises.

As I repainted the edges of the front wall of the new shed in the pasture where he and Zip had left their dental autographs, I felt hot breath on my neck.  "You've done a pretty shabby job of breaking Slobber Boy, there, of his bad habit, haven't you?"  Leo pointed out.  "He's still as disgusting as he ever was!"  

"But," I said, wiping snot off my hair, "how do you know he wouldn't be worse if I hadn't kept after him?  How do you know he wouldn't have chewed all the way through the fence boards?  In eighteen years I couldn't stop Pokey from licking the oak board on the side of her stall door, and she licked it almost all the way through.  Who's to say he wouldn't have done the same?"

Leo, having had enough of the discussion, snorted and wandered off, but the thought stuck with me.  How can you tell?

Dakota demonstrates his disdain for couture and
possibly a reason for his grudge against wood, metal, plastic
and me.  

The answer is you can't.  The bad part of dealing with live subjects, be they human or otherwise, is that there are no do-overs.  It's not possible to go back in time and readdress the same issue in the same subject under precisely the same conditions to see what changes might have been wrought over time without your intervention.  And there's certainly no way (as I hope I belabored to the point of browbeating in the last two blog posts) to account for random events intervening.  Variables, Man!  There are just so many of them!  For instance, even without my intervention, he might have been kicked in the mouth (Zip, I'm looking at you!) and lost his front teeth.  Or he might simply have lost his taste for rough-cut lumber or shifted his focus to cedar trees--lumber-on-the-hoof, so to speak.  Or perhaps on the day I chose to record his behavior he wasn't really feeling the whole chewing thing, or a mouse distracted him, or he stopped licking because I bleached his bucket and it tasted bad.  Who's to say?

The point, then, is that we can't know what behaviors we prevented from occurring.  We can only watch for long-term changes, ignore the minor interim shifts that are nothing but extraneous information, interesting but useless, and hope that the result we see is actually related to the training (or teaching, or protection) method we pursued.

As far as we horse people go, mostly we can stop being so nit-picky and just enjoy the fact that we're interacting on some level with a creature who is more unlike than like us.  That's total coolness and worthy in and of itself.  The guns and teachers (or even teachers with guns) issues will have to roll on unimpeded by logic and without even a cursory nod to synchronicity or randomness while they're beaten to a pulp by soothsayers and pundits.  That's what we do.

Monday, April 15, 2013

I love my life!

Early spring afternoons have their own special flow.  Early spring afternoons when we say, "F*#% the gardening; I'm going riding!" are even more special.

So here I am on a special early spring afternoon, typing with dust-covered fingers, smelling of horse, because I said, "F*#%; I'm going riding!"  But that's not the point of this post.  The point is that, once more, my horses have taught me something, and I feel the need to share, if not the fact of what they taught, at least the fact that we always have more to learn, and that life with animals is a never-ending learning experience.

What I learned today was this:  Sometimes it's not the "out there" that's scary to our horses, it's the "in here"--something in their hearts--that's scaring them.  And more importantly, sometimes (not always) we are the cure.

Dolly grazing "at liberty" on the front lawn this afternoon moments
before she stepped on her lead, got upset, and I looped it around her neck.
There followed a few minutes of her glued to my side as the lead rope was far
too snake-like to be acceptable.  And then...voila!  Freedom!

For the past few days my lovely inherited mare, Dolly, has been keeping the herd from accessing the wonderful grass in the upper pasture.  I opened it as part of my usual rotation, and expected hysteria and lots of quiet munching.  I got the hysteria.  The quiet munching, not so much.  Dolly found something in that familiar field that set her teeth on edge, and she responded by driving the herd out of the area.  She allowed them minimal access, then they were banned and banished from further excursions.  She did this every day for the past three days, sometimes five times a day.

Today I got it in my head in one of those sketchy, "A-HA!" moments, that if I rode her, the rest of the Boyz, sans her personal wild hair, would meander up there and enjoy a much-needed, post-winter snack.  That's not what happened.  It's never what we expect.

What happened--after the ride, after the grazing on the lawn which was its own learning experience for the newly-returned mare, and after the post-ride fussing and grooming--was that Dolly got her bitch on and drove the herd up into the field that had been terrifying her.  Now, this is a mare who, from the get-go, relied heavily on her bond with her rider to get her through the tough parts.  Jess always contended that she's very dependent, and now that she's mine (as opposed to my daughter's, with me as onlooker), I am beginning to see what that means on a daily basis.  It's not just under saddle that she expects her human to give her direction; it's also on the ground.  So our little "I swear you can graze the lawn with the lead looped around  your neck" moments translated into, "Damn!  I'm tough enough to face that pasture with my eyes wide open!" Off they went, and I couldn't help but smile.
Dolly can't help but smile either.

But that's not all I learned today.  I've posted before about horses learning from other horses.  I don't care that researchers deny the existence of such behavior; I've seen it in my barn, and I was pretty sure I had a grip on the concept.  Then I had Pokey put down.

Pokey was Zip's mom, and with the exception of a month when I boarded the big guy out at a dressage barn, they'd been together.  Pokey had her quirks, and one of them involved intently licking the board outside her stall door after she ate.   Every meal.  For 18 years.  Her attention pretty much dissolved the 2x4.  Zip, her babykins, followed suit, and I assumed this was an idiosyncracy I'd have to live with until ...well, forever!

Not so.  Pokey went down on March 13th.  I'm here to attest, exactly a month later, that from the moment his mom disappeared from the barn, Zip has not given the wall more than a cursory lick.  Not once.

So I need to amend my supposition that horses learn from each other.  What they do is imitation, not learning.  As long as Pokey was two stalls down and fully committed to wall-licking, so was Zip.  The minute she was gone, so was the habit.  Not learned, imitated.  What an eye-opener!

Granted this is about as anecdotal as evidence gets, but for me it's huge.  I would love to hear from my readers who have seen similar behavior.  Even more, I'd like my readers to be totally aware of the beginnings and endings of behavioral patterns and what connections they find with things going on around the stable.  This could be the start of a whole new understanding of equine behavior...or not. In any case, it's interesting and worth exploring, don't you think?

Let me know what you find out.  Me?  I'm going riding.

Monday, April 01, 2013

Naive Interventionism and the Thinking Horseman

Iatrogenics [eye-at-tro-GEN-iks].  Now there's a word for your Words With Friends game.  It refers to illness (or poor outcomes) created by overzealous medical intervention.  In other words, it's the unintended consequences that result from our desire to manipulate our environment.  I've said it before, and here is comes again:  Just because you can doesn't mean that you should.

Makes you want to knit him an afghan, doesn't it?
But this little guy is far better equipped to handle the cold than  you are,
and he'd probably eat the afghan and die from your efforts.

We humans can't resist fixing things we believe are broken or wrong or just not the way we like them.  When we evolved opposable thumbs, it must have cost us a few of the rationality cells that allowed our brains to see some of the vast array of random events that are laid out around us like a Hubble photo of heavenly bodies.  One would think we'd be able to finger just a few of them, but we seem bereft of foresight in so many arenas.  Just watch society at work for an hour, and you'll find yourself shaking your head in disbelief.

As I type, there are discussions raging all over the Interwebs about what the world should be doing about the North Korean despot, Kim Jong Un's, latest declaration of war against South Korea and the rest of humanity. The desire to intervene and preclude action on the threat is strong in some regions.  That most of the powers involved have no idea what the long-term result of intervention might be doesn't stop pundits from postulating, posturing, and positing their individual brands of problem-solving.  And just last week we heard at great length about a mother suing a pharmaceutical company because she'd administered a common over-the-counter fever reducer to her child, and the child had suffered a bizarre and severe adverse reaction.  The anti-pharma folks screamed foul at a product that endangers users.  The medical profession screamed back that if Mom had just let the child have the fever for a bit, she would probably have recovered without intervention, and warnings to that effect went out around the world.  When the medical community warns against intervention, you know it's got to be serious.

Naive doesn't mean "innocent", it means "without experience".  So "naive interventionism" means attempting corrections when it's not actually proven that there's anything in need of correcting.  Insufficient data, as my psych profs used to tell us, proves only that we're lazy experimenters.

The result of attempting to "correct" a horse's hoof appearance
can result in screwing up fine points like breakover,
resulting in a jumper suddenly unable to get over a rail without clipping it..

Nassim Nicholas Taleb writes in Antifragile that the inability to predict the future isn't, in itself, our downfall. It's our desire to do so combined with our inability to understand that we need to be less distracted by the endless input from our environment that sends us spinning off in a thousand directions at once (and probably accounts for most of the cases of depression and anxiety that are increasingly common in our population).  It's the fact that we consistently try to predict what can't, with any certainty, be predicted and to fix what shouldn't be fixed that causes so much grief species-wide.

We Thinking Horsemen are among the masses who tend to jump the gun and try to "fix" things that probably would be better off left alone.  We watch our horses like surveillance squads, noting every misstep and every uneven breath, and we're on it before Judy can belt Punch with her rolling pin.   Taleb points out that we would gain more accurate information if we assess changes over a period of, say, a year than if we do a continuous daily assessment.  The odds are better that we will see more "signal" and less "noise"  , hence have less likelihood of correcting something that doesn't need to be corrected and creating an unexpected problem as a result.

A random event can throw off data.  Let's say you are worried that Fuzzbutt is never going to learn to piaffe despite his physical prowess and mental ability to master the movement.  You ride him for two hours in one day, and during the tenth repetition of the cues, he perks his ears, calls loudly to a pasture buddy, and spins around.  What probably happened was a movement beyond your field of vision or a sound out of your hearing range that startled him.  A random event.  As a one-in-ten, it's pretty meaningful.  You only got 9 good trials before you got one chaotic whirl-around.  Project forward from that flimsy sample, and you might expect that in the next year--perhaps 200 riding sessions and 10 rounds of cues per session for a total of 2000 events--you will get 1800 good responses and 200 incorrect responses.  Seems logical, right.

But if you waited before making your assessment until you had more events under your belt, you might actually find that the ratio was more like 1 spin out of every 400 trials or even less.  It could have been a one-time event, and it's on that shoddy sample that you'll base your training regimen.  You'll start over with ground work, sacking out, retraining to the leg and seat cues....and possibly wind up with a bored, pissed-off animal who (as we well know) will find myriad ways to get even with you for insulting his intelligence to such a degree.  I can certainly attest to how creatively our animals respond to nonsensical treatment.

So today's lesson rides on not being naive.  Make sure you have enough evidence obtained over a long enough period of time to be of some value before jumping in and fixing something that you'll likely have to un-fix a month down the road.  Patience is never misplaced unless you're faced with an outright threat to life or limb, in which case running and hiding are always good.