Guest Essay by Dr. Melissa Starling
How do we know that our training decisions are good ones? I think it is especially important to know this in dog training. Because we are working with animals that can’t tell us directly when we are wrong, we have to be especially confident that we are right. I am a scientist and I work on dog behaviour in the real world where most dogs live. So, typically I am the first to push how important science is, and I am known for an infuriating habit of asking to see scientific evidence when someone makes a big claim. I want to see scientific evidence because I want to be as sure as possible of what is true. I know that it is not good enough to just reckon I know what’s going on or make casual observations. Everyone is biased, and the scientific method is supposed to help us overcome those biases so we can find what is true. Or, at least, most likely true.
So, in the search for the highest truth, scientific literature generally produces high quality evidence. It is therefore unsurprising that there are some discerning trainers out there that are eager to align their training with scientific findings. They may call themselves “science-based” trainers, which sounds ideal on the surface of it. But what does basing your training on science actually mean, and does it make you more right than everyone else?
Science has given us operant conditioning, which is in my opinion a trainer’s bread and butter. It super-charges our ability to change a dog’s behaviour by giving us an enormous amount of detailed information about how reinforcers and punishers can influence behaviour. However, operant conditioning does include suppression of behaviour (punishment) as well as reinforcement of behaviour. While Skinner was a proponent of using predominantly positive reinforcement, I don’t think it can really be said that science doesn’t support the use of punishment. For example, it shows us exactly how punishment can suppress behaviour. I argue that if I wanted to suppress behaviour (and I might, you never know), I would be a science-based trainer if I used scientific findings to apply punishment to achieve that.
It is often argued that science supports the use of positive reinforcement and the avoidance of aversive control (punishment or negative reinforcement). Usually, Murray Sidman’s “Coercion and its Fallout” is cited, which is interesting, because Sidman’s book does not contain a reference list, so there is no opportunity to check up on the experimental evidence Sidman refers to. I think that is one way to interpret the literature, and there is evidence that animals that are regularly punished suffer some global changes in how they approach life that we might not be comfortable with. But the thing about scientific literature is that you can often find some in support of opposing viewpoints. I am often reminded of Shakespeare’s quote “The devil can cite scripture for his purposes.”
Science is impartial and unemotional, and animal behaviour is complex and adaptable. Different conditions produce different results. It’s not that hard to find some support in scientific literature to justify the majority of things you might be doing in your training, or beliefs you may hold, as long as you are not outright abusive. A trainer that uses punishment regularly could call themselves a “science-based” trainer and be pretty accurate. It’s just a matter of what subset of the scientific literature they are going to pay attention to and what they choose to ignore. There is so much scientific literature that everyone will be either intentionally or unintentionally ignoring some of it.
So, inevitably we have come to a curly issue. We want out training decisions to be good ones, and science provides good evidence to base training decisions on, but what can it tell us about being ethical trainers as well as ones that make good training decisions? Science actually can’t really tell us much about ethical treatment of animals in training. I am sure some people are going to disagree with this, and that is their prerogative. I argue that once you introduce the kind of dynamic environment that we usually train dogs in, a lot more is going on to produce behaviour than the simple, standardised lab conditions used in studies of learning theory. We are also dealing with an animal’s ethology, or what they have evolved doing. Life is not food deprivation, levers, and lights.
Whoa, hold on! What am I saying here? Operant conditioning works on EVERY ANIMAL WITH A BRAIN STEM. Where did I get my degrees? A corn flake packet?? Yeah, I hear you, but hear me out. The search for the highest truth might end at science, but we still need to find science relevant to our field. Psychology based on what an animal does in a box is not the highest truth for me out there in a changing, complex environment with competing stimuli, motivation and emotional states constantly in flux, and animals with rich and varied learning histories. Science is a very important consideration, but it won’t tell me how to train all dogs with all problems in all environments. For that, I need some specificity that scientific studies have not produced much data on yet.
So, what is the highest truth for me in dog training, then? Evidence. There may be few specific scientific studies I can draw on to help me make decisions. There are broader studies I can draw on to help me figure out the function of behaviour so I can better treat it. But I am not a psychologist and I don’t need to be to be a good, ethical trainer. I can keep track of objective things like how many times a dog does a behaviour in a set time frame, how quickly they will engage with new things, or how often they do a behaviour, or how quickly they respond. And I love to count success rate – how many times I get the behaviour I want compared to how many times I ask for it.
It is not just using numbers to tweak operant conditioning. It is using numbers to track the effectiveness of what I’m doing. If it’s not effective, that may be addressed with what I know of operant conditioning, but it may also have other causes. It may be that the dog’s arousal is inappropriate, or the behaviour has a function different to the one I am assuming, or the dog is emotional for some reason. There are many reasons why it might not be working so well, but if I am keeping track of the dog’s progress with objective measures, I can ensure that there is good evidence that is relevant to the situation for the decisions I am making. I have the ability to adapt my training so that I am not bound by results someone in a lab has got with rats that may or may not apply to my troubled dog on a leash in the street.
Science is right a lot of the time, but the farther we get from what has been specifically studied, the more likely we are to be surprised or find that things are not working like science said they should. It doesn’t mean that science is wrong and should be shrugged off. It just means we should acknowledge that there is a currently a limit to what science can tell us about our craft and how to perform it, but we can still use the principles of science in pursuit of our highest truth in dog training. We should just maybe consider sometimes that something weird and undocumented could be going on, and we should be documenting it ourselves!