Monday, 28 April 2014

The Role of Ethics in Discussions of Dog Training

This blog post was widely shared in the dog training community a couple of weeks ago:

The post was well-written and it raised some great points, but one phrase stood out to me: �my ethics are mine and yours are yours�.  I immediately took issues with this because it appeared to be equating �ethics� with �moral relativism�; as if my ethical beliefs are like my opinions - something I am entitled to whether or not they make any sense.  

Ethical beliefs are not like opinions, they�re not vague statements of �stuff I feel� that nobody has the right to question.  Ethics is a discipline that follows logical rules and, theoretically at least, generates principles that are justifiable to any rational person. There might be disagreement over premises, but this doesn't mean there's no possibility of debate.

In fact, I strongly believe that talking about ethics is important in dog training just because it forces people to confront their own beliefs in a rational way, rather than walling them off behind "well I'm entitled to my beliefs and it's all subjective anyway".  The principles that inform dog training, like the desire to create strong, healthy relationships between dogs and their owners, are too important to be waved away as mere subjectivity.  

Most people become dog trainers because they care about the welfare of dogs.  They therefore have ethical principles about how we ought to care for dogs; giving them the skills they need to live in a human world through training is part of animal welfare. I would contend that almost all dog owners and professionals who work with dogs share the principle that causing a dog unnecessary suffering is wrong, and we have a duty to prevent it. 

But caring alone isn�t a guarantee of morally right action.  Being ill-informed about what dogs are and how they learn can lead to a course of action that is, strictly speaking, ethical insofar as the trainer is trying to do the right thing, but it�s still wrong because the facts it draws on are mistaken.  Not understanding the ways a dog shows us that he is suffering, for example, can lead to misinterpreting being shut down through chronic fear for �calm submission�.  It�s not enough to believe that unnecessary suffering is wrong if you can�t recognize it when you see it.  An ethical dog trainer understands dogs and brings this knowledge to bear when choosing a course of action.  

We can criticize arguments about whether an approach or a tool is morally right without insulting anyone or resorting to something like �if you do this you�re a monster� - a sentiment I�ve run across too many times in discussions with trainers. Discussing ethical principles should not be taken as an instant ad hominem attack.  By paying attention to what the underlying ethical principles in an argument are, the conception of what a dog is that informs those principles, and the logic behind the argument we can judge whether a trainer is making the right decision in the tools he chooses to use.

Sunday, 27 April 2014

What Is �Force-Free� Training? A Conceptual Analysis

There are a lot of explanations of force-free training on reinforcement-based training sites.  Clicker training, for example, is claimed to be �force free� because it theoretically does not use aversives in any form, only positive reinforcement.  However, some aversive-based trainers also describe themselves as �force free�, because they do not physically touch the dog during the training process, only using remote corrections. There is some conceptual confusion over the meaning of the word �force� in dog training. 

In this article, I am going to make the argument that �force-free� does not capture what the shock collar trainer or the clicker trainer wants it to capture. 

First, I will give an example of what force-free training does mean, then show that it is inadequate.  Pushing on a dog�s behind to get him to sit, standing on his leash to get him to lie down, or pinching his ear until he drops a toy are all examples of force in dog training. On this definition, �force-free� training is any technique that refrains from physically manipulating a dog�s body, including the use of shock collars, bark discs and other remote aversives.  This is correct to a point, but it does not do the work that reinforcement-based trainers are using it to do, which is to distinguish reinforcement-based training from aversive-based training.

At this point I�ve talked about what force-free doesn�t mean.  Which leaves is with the question, what does it mean to force someone to do something?  We have to move beyond this basic definition of force and look to a more subtle understanding of the concept. 

An analogy with humans is informative at this point.  When someone says, �I was forced to do it�, they mean �I had no choice�.  But what they really mean is, �I felt like I had no choice�.  As Immanuel Kant put it in his characteristically Puritan way, 

Suppose that some one were to aver of his most passionate desire that it were irresistible if the alluring object and the opportunity to it were at hand; ask him whether he might not be able to master this desire if a gallows were erected before the house where he is to avail himself of this opportunity, in order that he might be hanged thereupon immediately after his savored passion . . . it won't take long to guess his answer.
We judge whether a person �really� has no choice by looking at what the potential consequences of his actions were and whether it would be reasonable to expect him to resist.  In Kant�s example, avoiding death is more important than seeing one�s mistress, to the point where no rational person would choose death.  Kant�s example is one where most people would claim they were �forced� to give up their mistress on pain of death.

A positive example is, for example, if I offered you a thousand dollars for the last candy in your bag.  You might have been looking forward to it, but really, you�d be crazy to refuse.  Although we don�t typically talk in terms of being �forced� to part with the candy, the mechanics are the same - in both cases we are being made an offer we can�t rationally refuse.  

Anyway, the point here is that on this definition, me �forcing� you to do something means making all options other than the one I want you to do seem like they are not really options at all.  This can be by making the other options so terrible that only one seems possible, or making this option so good that the others seem terrible by comparison.  

In dog training, we call this �setting the dog up for success�.  Why?  Because behavior is lawful. Dogs perform the behavior that they believe, at the time, will lead to the best consequences for them. Either they�re most likely to get food, or a toy, or praise or, they�re most likely to avoid a shock or a startle or a verbal correction.  We manipulate the antecedents and the consequences to make the behavior we want the one we get. 

When we give a dog positive reinforcement for sitting, we�re creating a history of positive associations between the command �sit� and sitting.  Create a strong enough history, the theory goes, and the dog will sit even in the presence of a distracting squirrel, because to them, sitting on command is like taking a thousand dollars for a piece of candy. 

When we give a dog positive aversive for breaking a sit before he�s told, we�re creating a history of negative associations between the cue �sit� and any action other than sitting. For them, even if there�s a squirrel, maintaining the sit seems irresistible, even if there�s nothing physically stopping him.  If the shock is aversive enough, the squirrel is like the mistress in Kant�s example - pleasurable, but not worth it.  In both cases, reliability of the behavior is dependent on creating the conditions where a dog would not rationally choose to do anything else. 

In human terms, if the examples I gave above are convincing, we would call this �forcing� because it would undermine a human�s sense of responsibility for her actions.  In dogs, it�s called �training�; making the choice we want the only one the dog would choose no matter what the situation.*

So once again �force� doesn�t really cut across reinforcement and aversive - the difference between these two is not in what�s happening in terms of creating choice-conditions for the dog.  In both cases we�re funneling the dog�s desires into the behaviors we want him to perform; because behavior is lawful, he effectively doesn�t have a choice.  

Other than the initial distinction between use of physical manipulation, the concept of �force free� training does not do the work that dog trainers use it for.  It cannot distinguish between reinforcement-based and aversive-based methods.  This is one reason why the phrase has been a source of contention in the dog training world. 

Is the concept of �force free� training worth anything, then?  I would argue that it is, for two reasons.  One is to make that initial distinction between physical force and shaping choices remotely, which is useful in some circumstances.  Secondly, much more importantly is that, for dogs, the illusion of choice is important for their wellbeing.  Arguably, the same is true of humans (although I�m not touching the free will debate with a bargepole!).  �Force free� makes us confront the fact that in shaping behaviors we are making certain actions irresistible for dogs, whether as part of a training program or accidentally, and the fact that dogs are so easily shaped and manipulated in this way gives us a certain responsibility to them.  

Even if we�re in the background doing the funneling, dogs want to feel like they�re a partner in their training, and they should be allowed to express their creativity where possible and not be controlled at all hours of the day.  We shouldn�t see dogs as robots that we strive to master the programming of - �force free� training to my mind is letting the dog decide what it wants without always telling it first.  It�s trying to accommodate a wide variety of reinforcers and developing a relationship so that both learner and teacher work out the best way to build a bond. Too much control over behavior can lead to compulsion, not obedience.  

*Of course, this isn�t always possible - a dog that reliably sits and stays in the presence of other dogs may not do so in the presence of a running chicken, even though he knows you have a pocket full of hot dogs.  A dog that reliably avoids the shock from an invisible fence might choose to run right through it if the neighbor passes by with a cat on a harness.