Thursday, 30 October 2014

LIMA in Practice: Efficacy and Welfare

As people who work with dogs, we have usually amassed knowledge of many different training techniques.  Our toolboxes are overflowing with products, techniques, protocols, proprietary systems, theoretical concepts, half-formed ideas - all designed to teach a dog how to do what we want.  With so many tools at our disposal, we have to ask ourselves what the best way is to train a desired behavior.  Where do we start?  We need a principle of help us decide what kind of intervention we should select.  

The principle of LIMA was created for just this purpose.  LIMA encapsulates a set of criteria on which we can choose between interventions - we should choose the Least Invasive, Minimally Aversive tool at our disposal.  LIMA, however, is only a practical principle, it is not a justification.  LIMA by itself cannot explain why we should accept LIMA; that would be tautological.  The reason we care what is the Least Invasive, Minimally Aversive intervention is because we care about the dogs welfare.  Training is supposed to make the dogs world a better place.

So, LIMA is an answer to the question, how do I train with welfare in mind?  Seeing it this way, as part of a larger picture, allows us to fold in other important concepts that are not easily captured by seeing LIMA as a standalone principle.  

One of these concepts is efficacy.  Recently, Eric Brad argued that efficacy is a vital consideration in choosing an intervention, and that LIMA is only part of the decision-making process.  He offers an excellent treatment of the concept of efficacy as it relates to dog training, however, I think he is too quick to relegate LIMA to a great place to start. Rather, I think that a complete conception of LIMA necessarily contains the idea of efficacy.

LIMA is based on welfare.  In many cases, dog training aims at improving the dogs welfare.  Let's look at an example:

A dog is afraid of loud noises, but his owner lives right next to a railroad track, with trains making the house shake every couple of hours.  As a result, this dog is living in a constant state of stress and barks like crazy when a train goes past, annoying his owner and their neighbors.  His owner hires a trainer to try to make the dog stop barking.  The trainer notices that the dog is barking because he is afraid, therefore, his aim is to improve the dog's welfare by making him feel better about the trains, which in turn ought to make it easier to stop the barking.  What kind of intervention should the trainer try first? 

The absolutely Least Invasive, Minimally Aversive option in this case is something like an Adaptil pheromone diffuser.  The trainer could instruct the dog's owner to plug it in and hope for the best.  But there is only patchy scientific evidence that these diffusers work, and no evidence that is specific to this situation.  Choosing this approach, then, has a reasonable chance of not working.  This means, the dog has a reasonable chance of continuing to live in misery for as long as it takes for the trainer to decide to try something else. That can't be right. 

By contrast, behavioral modification by classical conditioning and desensitization has been proven to work for decades, in hundreds of situations.  It can be invasive, as it involves setting up the environment to restrict exposure to the trigger.  It can also produce unpleasant feelings if it is done improperly.  But it is effective.  

So, should considerations of efficacy mean the trainer should skip a step in LIMA at this point?  I would say yes, absolutely.  The risk of the dog continuing to experience fear and pain justifies using something that is more invasive.  Considerations of overall welfare have to be paramount.  Effective solutions should take priority over ineffective solutions.  In real life, this means we have to choose solutions that have evidence behind them.  

This isn't an argument in favor of using aversives as a quick fix - the framework of LIMA still applies; don't use a more invasive or aversive technique than you really need to.  As I've argued elsewhere, if it is true that one never needs to use and aversive technique in training, then LIMA doesn't justify doing so. But what a trainer really needs to do must always be determined by the situation of the dog you see in front of them, not by a rigorous structure of interventions.  

Considerations of efficacy can be brought in through positioning LIMA as one part of a holistic, welfare-focused approach to dog training.   The ethical approach to selecting a technique is to choose something that will lead to the dog spending as little time as possible in a state of physical or psychological pain.