Wednesday, 1 May 2013

Describing Dog Training: Weasel words or clear descriptions?

Dog training is an unlicensed profession. Sometimes it surprises people to learn there is a science to training, the origins of which can be traced back to Pavlov and Skinner. When studying how ordinary people train their dogs, scientists have to map between technical terms and everyday language. How do they do this?

Positive reinforcement and punishment in dog training

You’ve probably heard the phrase that “dogs do what works”, as explained by Jean Donaldson in her wonderful book Culture Clash. What this means is, the behaviours that are rewarded get repeated, and the ones that don’t get rewards tend to disappear (this is called extinction).

Our knowledge of operant conditioning has its roots in the work of B.F. Skinner, who coined the phrase in 1937, but applied behaviour analysis is still an active field today; for example, it is used to help children and adults with autism.

Operant conditioning relies on punishments and reinforcements. Reinforcement makes a behaviour increase, whereas punishment makes a behaviour go down in frequency. Either can be positive or negative, and this is where it gets technical, because the terms don’t map intuitively into everyday English. 

Positive reinforcement means that something good (reinforcing) is added, such as when a dog is given chicken. Negative reinforcement means that something is taken away, such as a painful stimulus or an unpleasant vibration that stops when the dog does what you want it to do. Positive punishment is essentially what we mean when we talk about punishment in everyday language; something unpleasant is added, such as when a dog gets a bop on the nose or an electric shock. Finally, negative punishment would be something like a time-out; the dog is losing its freedom to play or explore.

When conducting surveys of dog-owners, scientists can’t just ask ‘do you use negative punishment?’, because this is a technical term that unfortunately even some dog trainers don’t understand. So they have to phrase the questions in everyday language.

One way of doing this is illustrated by Hiby, Rooney and Bradshaw (University of Bristol) in 2004. They decided to ask about the training of specific behaviours. This is helpful because some people might use one method for teaching ‘sit’, and another for teaching walking to heel.  They asked open-ended questions so that participants could write in their own words how they had taught various commands, and how they responded if the dog did something like stealing food or chewing a household object.

The scientists themselves then classified these techniques into the quadrants of operant conditioning. Actually, they categorized them as reward-based, punishment-based, and miscellaneous.

Dog training methods: Positive reinforcement vs punishment
Another approach is to ask participants to rate specific statements on a scale. This is what Arhant et al in Vienna did in a large survey published in 2010. They used a five-point scale ranging from 1 (never) to 5 (very often) for owners to rate how often they used particular techniques, including food, praise, play, scolding, startling with a noise, and a jerk on the leash.

Although participants don’t get to use their own words, the advantage is that better statistics can be applied to this kind of data. 

In fact, when Arhant et al looked statistically to see how the different training approaches related to each other, they didn’t end up with four quadrants. They categorized the approaches as punishment (technically speaking, positive punishment), rewards, and ‘reward-based responses to unwanted behaviour’. This last category included negative punishment such as time-outs, as well as comforting a fearful dog.

We can see that the main distinction drawn in these studies is between reward-based and punishment-based approaches.

So, does it matter which quadrants are used in training? Well, Hiby et al’s study found that although many people use rewards, many of them also use punishment. There was a positive correlation between the frequency of using rewards-based training and ratings of general obedience; in other words, dogs trained using rewards were rated as more obedient. There was no correlation between obedience and punishment. 

Arhant et al found similar results. Again, many owners reported using punishment in training, although they did not use it very often. More frequent use of punishment was linked to higher scores on aggression and excitability. In contrast, more frequent use of rewards was linked to more obedience, less aggression, and less anxiousness.

These results (and others) don’t prove a causal connection, but they do suggest it is more effective to stick to the positive reinforcement and negative punishment quadrants in operant conditioning, and not to use positive punishment. One of the risks of using positive punishment is that it might inadvertently cause aggression. A study by Meghan Herron found many owners report an aggressive response to the use of punishment.

Unfortunately it is not always easy to know what methods trainers use. The language of dog training can be full of weasel words. Instead of talking about punishment, some trainers refer to ‘corrections’ or to a ‘balanced approach’. Shock collars are described as remote collars, as if nothing happens at the other end (but of course it does, or else how would they work?). These seem to be ways of avoiding saying they use punishment. This is why scientists need to be careful in how they phrase questions, and also why owners need to look at the details when choosing a dog trainer.

Luckily, the use of positive punishment in training is old-fashioned, and more and more trainers (and owners) are taking a force-free approach. A force-free dog trainer will only use the positive reinforcement and negative punishment quadrants; sometimes, to avoid confusion, they will only mention positive reinforcement. Ask your dog trainer carefully about the approach they use, so that you know you will be happy with the way they treat your dog.

How do you train your dog? And, if you have had dogs for a while, how has your training changed over time?

References
Arhant, C., Bubna-Littitz, H., Bartels, A., Futschik, A., & Troxler, J. (2010). Behaviour of smaller and larger dogs: Effects of training methods, inconsistency of owner behaviour and level of engagement in activities with the dog Applied Animal Behaviour Science, 123 (3-4), 131-142 DOI: 10.1016/j.applanim.2010.01.003
Donaldson, Jean (2012/1996) The Culture Clash. Dogwise Publishing.
Hiby, E.F., Rooney, N.J., & Bradshaw, J.W.S. (2004). Dog training methods: Their use, effectiveness, and interaction with behaviour and welfare Animal Welfare, 13, 63-69
Photos: Jeffrey B. Banke (top) & Lukas Gojda (both Shutterstock.com)
P.S. Make your dog happy: Train force free

6 comments:

  1. Very interesting! Thanks for the comparison of studies.

    I mostly use positive reinforcement. I use positive punishment rarely, but sometimes use it in a precise way to 'nip behaviours in the bud'. I use punishment as one-two trial learning experiences for dogs. If it doesn't reduce the behaviour in one-two trials, it's not punishment, just abuse, and not an effective training method.

    ReplyDelete
  2. Great article. I actually don't subscribe to use punishment while training dogs, I believe it delays training and causes confusion.
    http://howtotrainanewdog.com

    ReplyDelete
  3. I adopted my first dog around 8 months ago, and was lucky to have Ian Dunbar and Jean Donaldson recommended to me before she even arrived with us. I read both, and was instantly attracted to positive reinforcement training as a method. I enrolled in a puppy class run by a fantastic positive trainer. I've also read and enjoyed Karen Pryor and Sarah Whitehead and we use clicker training quite a bit.

    I'm incredibly glad now that we used all these methods as our rescue dog has had some behavioural issues, and continues to do so. By FAR the best success we've had with training are those behaviours we actively worked to shape through positive reinforcement. Negative punishment follows next, with slower success. The behaviours we've had the least success in shaping are those where we've shouted "No!" or "Stop!" or clapped our hands (i.e. positive punishment). We try to avoid this as much as possible because (a) we know it is counter productive and (b) we are very much against using punishment. (We generally only use this method for one thing - trying to get our dog to stop chasing our cat!)

    I find this infographic, illustrated by the fantastic Lili Chin, to be a great summary of these four types of training: http://farm6.staticflickr.com/5012/5498007787_4dcd6f7264_b.jpg

    Thanks for this great article :)

    ReplyDelete
  4. Positive reinforcement is a much better solution than using negative reinforcement. Adding a reinforcing stimulus like a treat for example is definitely a better solution then using punishment when your pet is doing something its not supposed to, try giving it a reinforcer for when its behaving good. For example instead of punishing your dog for climbing on the table to get food during dinner, give it a treat when it is remaining calm and laying on the floor. In my experience of having 3 dogs, using positive reinforcement also seems to work a lot faster and the rate of extinction decreases.

    ReplyDelete
  5. "Ask your dog trainer carefully about the approach they use, so that you know you will be happy with the way they treat your dog." Good advice! I suggest Jean Donaldson’s 3 questions to dog trainers:
    •What will happen to my dog if he gets it right?
    •What will happen to my dog if he gets it wrong?
    •Are there any less invasive alternatives to what you propose?
    http://wp.me/p3cQWk-CX


    ReplyDelete
  6. I knew nothing about dogs when I rescued my girl. I started reading some older books, then watched Milan's shows and even his videos and books. I asked a dog park friend what she did for her dog, who had a great recall. She lent me her shock collar and I looked up a video on using it. I tried it on her when she was chasing a squirrel and she ignored it. I turned it up, she still didn't react at all. The dog park friend adjusted the intensity and used it. She leapt in the air with a cry and looked around. Then she slowly returned to me, shaking and crying. It upset me so much I started crying too and refused to use it again. When I started looking up behavioral research on dogs and what studies show is most effective, I expected to learn a better way to use the choke and/or e-collar. But I came across a load of information that contradicted much of what the 'trainers' had said. Some information was blatantly opposite to well-researched information. I was amazed that there aren't legal restrictions on 'trainers' and that these products are sold to the public. If corrective tools must be used, shouldn't they be regulated and only accessible via prescription by a behaviorist, given the potential for harmful side effects? I took Karen Pryor's foundation course to 'retrain' myself how to work with my dog. She is now confident, relaxed, and more reliable than most dogs, no equipment necessary (or food). I want to see training education based on research, just like we do with most everything else- math, psychology, economics, etc.

    ReplyDelete

Companion Animal Psychology is a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for sites to earn advertising fees by advertising and linking to Amazon.com and Amazon.ca. (privacy policy)