This is my opinion, based on my experiences with dogs over the past 20 years and my understanding of behavioral science (which is pretty damn good!)  So if you strongly disagree with me, write your own blog about it.

Among dog trainers we seem to have been divided into camps based on our underlying philosophy and the techniques we typically choose.  One camp, the one that I’m proud to be part of, is focused on the use of positive reinforcement as a primary training tool.  Note that I did not say purely positive.  I am not that and never claimed to be.  I would label myself as primarily positive reinforcement.  Another camp has claimed the name of “balanced” trainers, implying that those focused primarily on positively reinforcement are apparently “unbalanced”.  If you didn’t know any better you would probably guess that balanced is good and unbalanced is bad, but you would be very very wrong.  It’s a problem of definition.  Then there are those referred to as “traditional” trainers that use a variety of 0ld-fashioned methods, often not at all supported by behavior science.  But I’ll leave them out of this for now and focus on the distinctions between “primarily positive reinforcement” and “balanced”.

The term “unbalanced” certainly sounds bad.  Maybe I have a mental illness or am unstable?  I imagine that too many pomegranate martinis or too many glasses of pinot noir might have the effect of making me “unbalanced”!  But not my training method.  I certainly don’t feel lopsided when I train.  I feel like I have a full arsenal of methods and techniques at my disposal and am very confident that my lean towards positive reinforcement gives me everything I need in training.

So what is the opposite of primarily positive reinforcement?  Before we can discuss that we all need to be using the same language, the language of learning theory.  Here’s a short and simple tutorial on the proper terminology of learning theory.  There are 5 possible consequences to a behavior.  Positive reinforcement, negative reinforcement, positive punishment, negative punishment, and nothing (no consequence one way or the other).  Positive means adding something and negative means taking something away.  They have nothing at all to do with value judgments of good and bad.  These are used as if they were mathematical terms (addition and subtraction).  Reinforcement is intended to increase the behavior it follows and punishment is intended to decrease the behavior it follows.  Note this says “intended”.  Sometimes it works; sometimes it doesn’t.  By combining these terms we get the 4 quadrants that people often talk about:  positive reinforcement, negative reinforcement, positive punishment, negative punishment.  If you put the definitions together then you get the following.  Positive reinforcement means adding something with the intention of increasing behavior.  Negative reinforcement means taking something away with the intention of increasing behavior.  Positive punishment means adding something to decrease behavior.  Negative punishment means taking something away to decrease behavior.  And no consequence would lead to extinction, the decrease and eventual loss of a behavior.  People often leave that out because it doesn’t fit nicely into the 4 quadrant model.

So if I train by primarily positive reinforcement then what are the other options?  Negative reinforcement, positive punishment, and negative punishment are left.  Negative reinforcement is always a tricky one.  Most trainers don’t really understand it and even my students, after an entire semester of study, don’t always get it right.  I’ve seen it defined and described incorrectly in many books, blogs, and posts.  Another way to think is negative reinforcement is “escape/avoidance training”.  Your dog can escape from or avoid something unpleasant if he does what you want.  The problem with negative reinforcement is that first there must be an aversive applied so that it can then be taken away.  This is the process behind using an ear pinch to train a retrieve.  The ear pinch is applied, and when the dog does what the trainer wants (opens his mouth and takes the dumbbell) the pinch is released.  The dog learns that taking the dumbbell makes the pain go away.  So in the future he is more likely to take the dumbbell when it is presented to escape or avoid the ear pinch.  But he also learns that his trainer is willing to use pain to get what he wants.  If I were a dog, I would not ever totally trust someone willing to hurt me like that.  Plus, the dumbbell is now paired with a painful stimulus (the ear pinch).  In classical conditioning we have just created an unpleasant emotional response to the dumbbell.  So negative reinforcement can lead to two big problems.  First, the trainer must do something unpleasant to the dog and second, the dog associates the behavior with the unpleasant thing.

Go ahead!  Try an ear pinch on  your cat and let me know how your retrieve training works.  Dogs are too tolerant sometimes.

Then we move to punishment.  Remember, punishment is designed to decrease behavior, so if you are focused on things you want to decrease or stop, and you are primarily approaching training from that perspective, you would indeed be primarily punishment.  Positive punishment is adding something unpleasant to decrease a behavior.  So if a dog is giving another dog a warning growl and you pop on the leash (let’s say attached to a pinch collar) you have just applied positive punishment.  Again, there are going to be problems.  First, you just tried to decrease an important bit of information that your dog gave you (he felt threatened or worried enough to growl).  By punishing the growl you may well eliminate it, and your dog will now go directly to physical attack in the future.  Great.  Plus, you again classically conditioned an aversive with a behavior.  Seeing other dogs leads your dog to be corrected.  Since he was already leery of other dogs, or else he would not have growled, your added correction convinced him that other dogs really are bad.  They cause him to get in trouble.  So you may have punished out the warning signs, but you just made the problem much, much worse through the use of positive punishment.  If there were a sarcasm font I would use it here when I say “Great job!”  The rule in training is to never make things worse, and in this scenario, you did.

Negative punishment isn’t quite as bad, though it can certainly be misused as well.  In negative punishment you take away something to decrease behavior.  The ultimate negative punishment for humans is jail.  Taking away something (freedom) in order to decrease behavior (crime).  Sarcasm font again “We all know how well THAT works!”  But sometimes very short time-outs can be useful in training.  These are not, however, the same as using environmental control and management to prevent unwanted behaviors.  For example, crating a puppy when you’re not home and can’t watch him so he doesn’t have accidents and chew things up.  These techniques come first as a way to avoid the situation that leads to the unwanted behavior.  Punishment comes after the behavior (remember, it’s a consequence not an antecedent).

Bear sleeping

In the above photo the crate is being used as management, not negative punishment.

Being “balanced” actually suggests that you are equally likely to use both reinforcement and punishment in their positive and negative forms.  And therein lies a big part of the problem.  Punishment has a host of serious side effects, even when done “correctly”.  It has a much stronger likelihood of unintended unpleasant fallout than does reinforcement.  That’s not to say that you can’t go horribly wrong with reinforcement based training, people can and do.  But their mistakes are typically less stressful to the dog and less serious in nature.  Too many cookies might lead to a fat dog.  Too many leash corrections might lead to a collapsed trachea.  The damage to the relationship between human and dog is much more likely with punishment than reinforcement.  I am definitely not saying we should let our dogs run wild and free so they can enjoy their lives without human intervention.  Dogs need to be trained to get along in human society.  I am saying that it can be done with minimal to no use of negative reinforcement, positive punishment, and negative punishment.

My goal is to be a good enough trainer that I minimize the need for aversives in training.  If I say that I need to use aversives then I am advertising my failing as a trainer.  If I am educated, aware, and proactive, my dog should not “need” me to introduce unpleasant events in training.  Trainers often punish because they don’t know the other options or how to use them effectively.  For example, if someone says “so what should I do, throw cookies at the dog?” that indicates a clear lack of understanding of the sophisticated and complex use of reinforcement.  People punish because that is how they learned and it has worked well enough for them in the past.

Even if it works, which punishment can, that doesn’t mean it’s the most effective or efficient method, and it definitely doesn’t mean that, ethically, it is the right thing to do.  The means-end argument suggests that “the ends justify the means”.  But that can lead you down a bad road very very quickly.  So saying “I used punishment and it worked” should not be justification for the punishment.  Shooting the dog (ala Karen Pryor) would work too, but we don’t recommend it.

I also hear people say that punishment used “correctly” is more effective and humane than reinforcement used “incorrectly”.  I disagree quite strongly.  If a father physically abuses his child and the child now obeys, would you say that the abuse was justified and appropriate?  It must have been correct because it worked, right?  Of course not!  We know that’s ridiculous because abuse can never be justified by the outcome.  While I’m not saying all punishment (or negative reinforcement) is abusive, there is quite a bit that crosses the line.  Granted, the line between abuse and punishment is a fuzzy one, and that is a big part of this problem.  One person’s punishment is another’s abuse.  In his book Willpower (a great book that I highly recommend) Roy Baumeister talks about a “bright line”.  A clear and obvious rule to follow.  We don’t have a “bright line” to distinguish punishment from abuse.

I must be very sensitive.  It bothers me at a gut level to see dogs that are stressed, shut down, anxious, or fearful.  But this clearly does not bother everyone.  Some argue it is simply part of the training process, or it is just what the dog must learn.  Or that life is going to be unpleasant sometimes, so the dogs need to learn to deal with it.  Poor dogs.  I can’t help them and I really want to.  It sometimes makes me really hate people.  Especially people who have information and knowledge about modern training, but choose to use old-fashioned harsh methods anyway.  If you really don’t know better, but you learn, grow and change, that’s fabulous.  But if you simply argue that anything different from the way you’ve always done it is wrong, that’s a problem.

So back to “balance”.  If my dog could, at any moment, be subjected to reinforcement or punishment, I’m pretty sure I know which he would choose.  I worry about the inconsistency that underlies the concept of “balance”.  My dog never knows what might be next.  There could be cookies and toys and games or  there could be leash pops, ear pinches, or a shock collar.  Imagine spending your life in an uncertain state.  Today might be a good day where you get reinforced with things you love.  Or it could be the day you get punished.  You just never know which is coming next.  To me, that sounds like the perfect recipe for stress, instability and lack of “balance” in the dog.  Lots of uncertainty and conflict.

I want my dogs to be stable, happy, and confident.  I want them to find me a source of consistency and security.  If I alternate between reinforcement and punishment they can never be 100% trusting of me.  So “balance” in training can actually have a very bad outcome for the dog.  Don’t be fooled because it sounds reasonable!  When people feel the need to make up new definitions of words so they sound more acceptable, you can guess there’s an underlying problem with what they’re doing, and they’re trying to find a way to make it sound better.

Advertisements