“Sheep-proofing” dogs in Norway

There was an article in a Norwegian dog magazine (an English translation is in the comments’ section below) this week, essentially claiming that by delivering an electric shock to a dog a single time as it runs towards a flock of sheep, you can stop it from ever chasing sheep again. Indeed, it gets a diploma to show that it’s “sheep-proofed”. There are no other ways of teaching this behaviour, the article claims, and also, it suggests that it’s a way to increase the overall quality of life of the shocked dogs.

Needless to say, this article has stirred a lot of feelings within the Norwegian R+ community.

And for good reason.

Now, I don’t train dogs myself, but reading through it I found I have four areas of concern:

“Only one shock is needed”

It is implied that 99% of dogs only need one shock, after which they become “sheep-proof”.

To me this number seems suspiciously low, and I’m concerned about potential whitewashing – I’d like to see the data supporting that statement.

To my knowledge, the only published comparison on the efficacy of shock collar training to stop hunting behaviour is from the recent “banana” shock collar paper (which suffered from major design faults as well as a blatant conflict of interest – however I see no reason to question their findings on the actual shock collar training), in which the average number of shocks on day 1 was 10, and even after 6 sessions on three consecutive days, some of the dogs were still getting shocked in the training arena to stop the chasing behaviour – even though the intensity of the shock was high enough to cause all dogs to yelp in pain at some point in their training. So even though that study was made on dogs chasing a moving plastic lure rather than live sheep, I would question that a single shock would immediately have such a laser-targeted, generalizable and long-lasting effect.

In other words, I suspect multiple shocks (over multiple sessions in multiple locations) would be needed to teach a solid avoidance behaviour in novel contexts. I’m not disputing that it can be done, but I’m hesitant to the suggestion that only a single shock in a single location would be needed.

The “Sheep proof” diploma

I would also question whether they actually become “sheep-proof”, implying that they will never again hunt sheep in any other context, after just one shock in one training session in one location.

Although there is often one-trial learning with aversive stimuli, generalization doesn’t always occur that quickly.

For instance, in the same “banana” study mentioned above, only 33% of the shocked dogs who had stopped chasing the lure in the first scenario refrained from chasing another lure in a novel location without first hearing the conditioned punisher (the word “banana”) – and 33% ignored the “banana” warning.

In other words, I would question that 99% of dogs would stop chasing sheep in novel scenarios after such limited training – especially given that there’s apparently no warning sound conditioned in the Norwegian shock collar scenario.

In short, that the dogs stop chasing the sheep after a single shock in the training scenario is not a reliable predictor of whether they will chase sheep two years later in a completely different setting.

Frankly, unless there’s more training than the single shock during a single session as indicated in the magazine, in my book that “sheep proof” diploma wouldn’t be worth the paper it’s printed on.

And now, I realize that some of my readers will be going “wait a minute!! We actually do shock the dogs in multiple locations and do the proofing in novel locations where the animal has no history of being shocked, so that diploma is indeed valid!” or “we actually do the shock training every year and update the diploma every year!!”

That just proves my point. I’m not saying I don’t think it can be done, I’m saying that I suspect that the procedure is being whitewashed in the article, playing down the severity and frequency of the required shocking.

Why not mention the potential side effects?

But let’s give them the benefit of a doubt for a moment.

After all, I’m not a dog trainer – I could be entirely wrong about the first two issues I’ve raised.

Let’s say that the outcomes of training with live sheep is vastly different from running after a plastic lure – let’s say that 99% of dogs actually do become “sheep proof” in novel locations after a single electric shock, even years after their original training.

The only way that you would get 99% compliance would be if fear conditioning occurred to the sheep specifically – and not to any other features of the training environment in which shocks occurred. How do we know that that doesn’t happen?

But even more importantly, would there be any known potential downsides, side effects, to this procedure?

Will the dogs really be happy and outgoing, merrily roaming the mountainside off leash chasing all kinds of other wildlife except sheep, and not develop any irrational fears or phobias?

I am rather sure that some dogs would – and that some wouldn’t.

I’m sure that some dogs take that single shock in stride and will happily ignore sheep forevermore, confidently trotting along beside their guardian into the sunset.

As I’m equally sure that some dogs will be more or less destroyed by that single shock.

And here’s the problem: we don’t know how many fall into the first category, and how many fall into the second. To my knowledge, there’s not any data on those proportions, but I will share one case I am familiar with.

Eva-Lynn told me: “When my dog was young, I allowed myself to be persuaded to use shock training in the presence of sheep so that he would be able to run loose on the mountain. The authorized trainer shocked him when he was happily and excitedly running out into the field with the sheep. He became terrified and screamed, followed by whining and trembling for 3 days. He did not develop a fear of sheep, but I drove back to where we had taken this so-called ‘sheep-proof training”, and he was terrified of that particular field in which the training had taken place. Afterwards, it turned out that he likes to play with sheep including their lambs, and has never been a threat to any sheep! Now he is dead, and I have spent 15 years of my life forgiving myself for what I exposed him to in all my ignorance, pressure from others, and stupidity. I’m still ashamed!”

In other words, whether the dogs learn that it’s the chasing of sheep that is connected to the pain, and not some other factor such as the field where training takes place, or the clothes the trainer was wearing on that particular day, or indeed the presence of a specific person, seems entirely arbitrary.

The trainers are assuming that they’re teaching an operant conditioning contingency (if I chase sheep, bad things happen) rather than an association based on classical conditioning (in this particular situation, in this location, with these sheep, with those people present, bad things happen). I would argue that we don’t know whether the conditioning that occurs is predominantly operant – or rather classical; most likely some combination of the two.

Also, we can’t assume that there won’t be any potentially really serious side effects, including generalized fear, aggression, apathy and reduced learning. For instance, the animal might start showing fearful or aggressive behaviour in that location, with those sheep, when the trainer wears those clothes or in the presence of those people.

To me it’s outrageous that the magazine seemingly whitewashes the technique without even mentioning the potential downsides – which may sometimes be extremely serious. The title of the article, “shocking dogs improves animal welfare” is highly inappropriate as it might inspire their readers to order a shock collar over the internet to improvise and try to improve their dog’s welfare all on their own. And remind me, are shock collars even legal in Norway? (rhetorical question, the answer is “no” – their use is regulated to only professionals with specific training, such as the trainer mentioned in the article).

What about alternatives?

Finally, the article dismisses humane ways of training saying “yes, you can teach dogs not to chase sheep with alternative methods but it won’t work for the vast majority of dogs”.

This is simply not true – you can indeed train most dogs to stop chasing wildlife (and not just sheep but any prey species, other dogs, bicycles or joggers) with kind methods that are fun for the dog.

It’s not a “quick fix” the way that shock-collar training is outlined in the article, but to suggest that R+ training is that ineffective is misinformation (assuming that the trainer doesn’t know better) or indeed disinformation (that he actually does knows it can be done with most dogs but wants to control the narrative).

Admittedly, there may not be any scientific papers (yet) showing successful predation substitute training with most dogs, but I know of several R+ trainers who are indeed successfully teaching the majority of their clients’ dogs (including “high-drive” ones) to engage with their trainers rather than chase various species of farm animals, wildlife, other dogs or joggers – I don’t know if they’re collecting data for any scientific study, but it wouldn’t surprise me if they were.  

The absence of scientific evidence is not evidence of absence – this is an evolving field and there’s been a landslide of relevant new publications related to dog behaviour in the last 10 years or so; I expect we’ve not yet reached the peak rate of yearly dog-related publications.

If I would hazard a guess, I’d say that it’s just a matter of time before such a study were published – they’re doing the training, and I would encourage the reader to check out their work. I link to the work of several such trainers here.

To conclude, we know that aversives work to stop behaviour, at least momentarily. In this article they downplay the number of shocks needed, making it sound like using shocks on dogs is a necessarily evil, a small moment of discomfort that opens the door to freedom forevermore.  They’re ignoring the documented risks of using punishment as well as the fact that there are valid pain-free alternatives. As long as reaching for that shock collar is the perceived quick – and only – solution, my guess is that these trainers won’t even bother to learn to use gentle techniques to keep dogs from chasing.

And sadly, now the readers of that magazine won’t either.

***

I write the occasional blog post, participate in webinars, masterclasses and online summits, as well as offer extensive online courses, all on the topic of animal learning, behaviour and welfare. If you want me to keep you posted on what I’m up to, just sign up to the newsletter below!