Patricia McConnell Seminar: Science-Based Training?

 Training with dad. Note the clicker in his hand.

A lot of people call clicker training (or positive-reinforcement training in general) “science-based.” But is it, really? What do we know, scientifically, about training dogs? In this, my last post on Patricia's seminar, I'll discuss some of the studies she shared with us.

I'm always astounded by the number of people who don't train their dogs. These are the people that, when they learn I do rally and obedience with Maisy, or that I teach training classes, always laugh and say, “My dog could use some obedience!” I'm usually then regaled with increasingly horrifying stories of near-death incidents resulting from a lack of training. But then Patricia shared two studies that made me wonder if most people even want a trained dog.

The first study looked at 118 dogs. Roughly half had no training, or only one basic-training class. The other half were highly trained agility, schutzhund, or search and rescue dogs. Each dog was tested on his ability to manipulate a box in order to get food out. Twice as many of the dogs in the trained group were able to get the food, suggesting that higher levels of training is associated with better problem solving skills.

The other study tested dogs' ability to discriminate quantities. The dogs were allowed to choose between small and large piles of food; in general, both groups chose the bigger amounts. However, the difference between highly trained dogs and untrained dogs became apparent in the second stage of the experiment, when the dogs watched their owners choose the smaller piles before being allowed to choose for themselves. The untrained dogs typically followed their person's lead, and also chose the smaller amount- this despite the fact that they earlier chose the bigger piles. The trained dogs, however, chose the larger piles, suggesting that training creates independent thinkers.

Independence? Better able to solve problems? Dare I say it: improved ability to think? I really don't think the average pet owner wants to live with a smart dog. Perhaps it's a good thing that pet dogs don't receive high levels of training!

Once we've made the decision to train our dogs, though, the next question becomes: how often should we train them? When Maisy and I were actively attending training classes, we were advised to train in short sessions, several times a day. At the very least, we should try to get in 5 or 6 sessions a week. As it turns out, though, this may not be the most efficient use of time.

Two separate studies found that training once a week results in “better learning performance.” They discovered that dogs acquired the skill in fewer sessions when trained less frequently than when trained daily. (One of the studies also looked at how well the dogs remembered what they'd been taught, and found that the dogs in both groups retained the task equally well.)

I think Patricia put it best: maybe the dogs learned in fewer sessions, but come on: it took eight weeks to teach a simple targetting exercise. Maybe it takes a couple of extra sessions, but by doing several sessions a day, the same task could be learned in just a few days. Still, she said these studies point out the importance of processing time; dogs need rest periods in order to learn most efficiently, especially for more complicated tasks.

Finally, every trainer has to make decisions about how they will train. Patricia shared that there are a number of studies showing that force-based training has negative effects. For example, one study showed that dogs trained with shock collars exhibited more signs of stress, even when compared to dogs trained with “fairly harsh” methods. Another found that punishment was associated with increased behavior problems, like aggression, distractability, and overall lower obedience levels. And the study I found most interesting discovered that punishment was associated with increased anxiety in fear in small dogs, but not in large ones.

There are also studies showing that reward-based training has good effects. These dogs are more likely to interact with strangers, be more playful, and are generally better at novel training tasks than dogs who are trained with punitive methods.

Patricia felt it was only fair to share a study whose results we may not like: it found that search and rescue dogs were more successful in advanced stages of training when there was “an increased use of compulsive methods.” Generally speaking, though, it seems that science favors reward-based training, which leads us to the clicker conundrum: should we use them?

One researcher trained 20 dogs to target a ball with their noses. Half the dogs were trained with a clicker, and half were trained with the verbal marker “good.” The results showed that the clicker trained dogs learned the task faster than those trained with the verbal marker (about 36 minutes as compared to 59 minutes). Patricia believes this is because the clicker makes a short, abrupt sound with a very clear start and stop. It's also a “broad noise band”- it covers more frequencies than the spoken word. All of these things make it more distinct and easier for the dogs to notice.

The last study that Patricia shared with us looked at the use of clickers and food versus food only in training. Thirty-five basenjis were taught to target a traffic cone, and once they learned the task, were variably reinforced for a maintenance period. The researchers found no difference in the amount of time that it took the dogs to learn the task; despite proponents' claims, the clicker was not found to speed up learning.

Then the researchers did extinction trials in which they quit giving food to both groups of dogs, but continued clicking the dogs in the clicker group. The results showed that the clicker-trained dogs were more resistant to extinction, to which I just have to say: DUH. The clicker is a reinforcer- it's a secondary reinforcer, not a primary one, it's true, but it's still a reinforcer. Of course the behavior didn't extinguish as quickly. They were still being reinforced. (To be fair, the study authors state that this suggests the clicker does, indeed, act as a secondary/conditioned reinforcer, and I guess it's nice to have that scientifically verified.)

So, with all of this in mind, will it train the way we train? Personally, the answer is no. I train because I enjoy it. Yes, I have a smarter dog as a result, and yes, that can make her more difficult to live with sometimes (I often wonder who is training who). But I train for the experience moreso than the end result... which is probably why I play endless shaping games but have pretty much nothing on cue. (Sigh.) And my methods? Well, those are unlikely to change, too. My choices have been made on my personal moral and philosophical beliefs, not science.

What about you? Will you change anything about your training based on these studies?

If You Want to Know More

• Does training make you smarter? The effects of training on dogs' performance (Canis familiaris) in a problem solving task, by Marshall-Pescini, et al
• Is your choice my choice? The owners' effect on pet dogs' (Canis lupus familiaris) performance in a food choice task, by Prato-Previde, et al
• The relationship between number of training sessions per week and learning in dogs, by Meyer and Ladewig
• Patricia's blog post on the Meyer and Laedwig study
• The effect of frequency and duration of training sessions on acquisition and long-term memory in dogs, by Demant et al
• Training dogs with the help of the shock collar – short and long-term behavioural effects, by Schilder and van der Borg
• Training methods and owner–dog interactions: Links with dog behaviour and learning ability, by Rooney and Cowan
• Survey of the use and outcome of confrontational and non-confrontational training methods in client-owned dogs showing undesired behaviors, by Herron et al
• Behaviour of smaller and larger dogs: Effects of training methods, inconsistency of owner behaviour and level of engagement in activities with the dog, by Arhant et al
• Obedience training effects on search dog performance, by Alexander et al
• Clicker Bridging Stimulus Efficacy, by Lindsay Wood
• Clicker increases resistance to extinction but does not decrease training time of a simple operant task in domestic dogs (Canis familiaris), by Smith and Davis

This post has been edited for clarity (see comments). It originally said: "Independence? Better able to solve problems? Dare I say it: improved ability to think? I really don't think the average pet owner wants to live with a smart dog. Maybe instead of training the dogs, we should focus on teaching the people how to manage situations better." I think the new version is a better reflection of the study.