Why Positive Reinforcement Doesn’t Work

I recent­ly read an arti­cle where the author was pro­fess­ing that he had tried pos­i­tive rein­force­ment and was here to say it didn’t work.  Right away my inter­est was piqued.  This has NOT been my expe­ri­ence at all.  In fact, I have seen the direct oppo­site.  I have seen it work what seems like mir­a­cles.  So what gives?

Let’s take an exam­ple.

John’s horse Smokey was extreme­ly hard to catch.  Any­time he saw John com­ing near his field he would run to the oppo­site end.  When John would actu­al­ly enter the field, Smokey would stay just out of arms reach play­ing games with John until he would give up.  So John reads this arti­cle about pos­i­tive rein­force­ment.  “Now he’s armed” he thinks.  He knows too well how much Smokey likes Stud Muffins.  So this time he heads out into the field to catch Smokey.  Well smart Smokey right away notices the bag of treat in Johns hand and VIOLA! allows him­self to be caught.  “I’m onto to some­thing now” John says as he places the hal­ter over Smokey’s head.

This con­tin­ues for about 2 weeks.  Then one day, Smokey decides, “Na, I’m not falling for that trick today”.  John’s befud­dled.  What went wrong?  He tries again the next day.  Again, Smokey will have noth­ing to do with being caught.  This con­tin­ues for a week.  Mean­while, John has resort­ed to shak­ing the bag of treats and shar­ing them with the oth­er hors­es.  Noth­ing works.  John final­ly gives up.  “This pos­i­tive rein­force­ment idea is crap” he says as he walks back to the barn with­out a horse in tow.

WHAT?  Reinforcement Doesn’t work?

In some instances no, it doesn’t.

What John failed to real­ize is that pos­i­tive rein­force­ment is a lit­tle more com­pli­cat­ed.  It requires a think­ing train­er just like any oth­er method.  So why did Smokey decide that he would pass on the Stud Muffins and stay in the field?  Why did pos­i­tive rein­force­ment fail?There could be mul­ti­ple rea­sons.  Here are just some.

  1. Lur­ing doesn’t teach the horse any­thing.  Dan­gling treats, shak­ing buck­ets, entic­ing with car­rots doesn’t work.  The horse is only fol­low­ing the food.  They aren’t think­ing and they aren’t learn­ing.  What hap­pens the day you for­get the treats or car­rots?  The horse goes back to his pre­vi­ous behav­ior hav­ing not real­ly learned to come and be caught.
  2. The Stud Muf­fin isn’t reward­ing enough.  While the treat may taste real­ly good, maybe the rain gave the spring grass a nice jump of growth and that tastes just as good too.  So John’s reinforcement/reward lost val­ue.
  3. John only comes out to the field to catch Smokey when he wants to ride.  This is again anoth­er case of the reward not being of enough val­ue.  Smokey is a lit­tle sore from last weeks train­ing and decides the treat isn’t worth the pain.
  4. There are no con­se­quences to Smokey not being caught.  In fact, if he avoids John he gets reward­ed by get­ting to stay out with his buds and eat more grass.

So how might we address all these poten­tial prob­lems.  And in this case I would use all these sug­ges­tions to train Smokey to be eas­i­ly caught at any­time by any­one.

  1. Take treats out in the field to catch Smokey but don’t flaunt them.  If he allows him­self to be caught then and only then does he get to see you have a reward.
  2. Try car­ry­ing dif­fer­ent rewards and many of them.  Some­thing real­ly valu­able that Smokey doesn’t get any oth­er time.
  3. Go out and give Smokey a treat in the field with­out catch­ing him.  Or catch him and imme­di­ate­ly release him.  Bet­ter yet, catch him, give him the reward then take him some­place even bet­ter to graze.  Vary this up so that every time you enter the field, Smokey doesn’t know whether you’ve come to ride, take him hand graz­ing, just to say hel­lo or for the far­ri­er.
  4. If and when Smokey refus­es to be caught, employ some good ole horse whis­per­ing tech­niques to ensure con­se­quences.  One of my favorite is to slow­ly (walk­ing, no run­ning or chas­ing) fol­low the horse around so he can’t stop and eat.  At first this tech­nique may take a while because in the past the horse has learned that run­ning away has got­ten the human to give up.  How­ev­er, even­tu­al­ly he will real­ize that it’s eas­i­er to be caught than to keep walk­ing and not be able to graze or hang out with his bud­dies.  A few ses­sions of this and he will get eas­i­er and eas­i­er to catch, reward or not.

See how each of these is not about pos­i­tive rein­force­ment not work­ing but about the train­ers appli­ca­tion of the prin­ci­ples.  The prin­ci­ples are fair­ly easy to under­stand but it takes skill to apply them in each sit­u­a­tion to ensure suc­cess.   It’s kin­da like your first expe­ri­ence rid­ing a horse.  You are taught to pull back to stop and kick to go for­ward.  Once you start tak­ing seri­ous rid­ing lessons, you real­ize it’s much more com­pli­cat­ed than that.  That what you once were told was a gross over sim­pli­fi­ca­tion of the process.  It’s the very same when some­one says use pos­i­tive rein­force­ment and reward your horse for doing what you ask.  It’s a gross over sim­pli­fi­ca­tion of the process.  If train­ing where that easy, every­one would be an expert.