A rational look at punishment in dog training
By Laure-Anne Visele, written Nov 2012.
The issue of punishment is one of the stickiest issues in the dog world today.
There are two extremes, and a flurry in the middle:
Team right: extremely traditional trainers: You get bonus points for hanging your dog up when he does not sit on command.
Team left: extremely positive trainers: You get kicked out for raising your voice to your dog.
My position is this:
- In practice, I am pretty left.
- In theory, I am OK with punishment if done rationally –which it never is.
So, let’s look under the scientific bonnet of punishment, shall we?
Skinner: On punishment and reward
B F Skinner (1904 – 1990) was an American experimental psychologist who bore the flag of ‘Operant Conditioning’. He believed you could get any animal to perform any (physically possible) behaviour by manipulating what follows it, by punishing or rewarding the animal for the preceding behaviour.
Operant Conditioning forms the pillar of dog training today, be it reward-based (clicker training is fundamentally about OC) or traditional.
So, consequences can be:
- Reinforcers: Increase how likely it is the behaviour happens again; or
- Punishers: Decrease how likely it is the behaviour happens again.
And consequences can be:
- Positive: Adding or starting something like petting or a hitting; or
- Negative: Withdrawing or interrupting something like petting or hitting.
And there you have your Operant Conditioning Quadrant:
- Positive Reinforcement (R+): Good behaviour -> Pleasant consequence (e.g. dog gets a treat).
- Negative Reinforcement (R-): Good behaviour -> Unpleasant stimulus stops (e.g. removing hand from dog’s back when it finally sits)
- Positive Punishment (P+): Bad behaviour -> Unpleasant stimulus starts (e.g. shouting, startling the dog, jerking the leash)
- Negative Punishment (P-): Bad behaviour -> Pleasant stimulus stops (e.g. stop walking when dog pulls on the leash).
Here’s a little riddle suggested to me by the author of the Koda Diaries:
Dog sits -> Dog gets a biscuit -> Dog sits less and less.
Question: What is the biscuit in this situation: R+, R-, P+, or P-?
……….. (drum roll)
Answer: P+. The dog in this example finds receiving a biscuit unpleasant, so the biscuit is a positive punishment in this case!
Right, that’s the theory bit out of the way. Now back onto my soapbox:
- I am definitely not against punishment (in OC terms): of course you want to decrease the frequency of some behaviours.
- What I am less in favour of is aversive (something the dog would avoid given a choice) or painful methods. Don’t get me wrong, I use aversives, but I do this with great caution (and reluctance).
Now that the theory is over, let’s look at practice.
1. Timing: The art of quick delivery
If you’re going to shout at your dog, make sure you are shouting for the right behaviour. There is no point in telling off your dog for chewing your furniture hours after the fact. Why is that?
Here’s what Poochie has done after committing the “crime”:
- chewed furniture
- scratched its ear
- had a bit of a sniff in the kitchen
- barked at the neighbour’s cat
- had a nap
- sprung to the door at the sound of your keys
- greeted your return
- got told off
Which behaviour do you think he thinks you’re shouting about? What’s worse, he might link your return with shouting. Suddenly your arrival will be a bit of a mixed blessing.
“But my dog knows it did wrong. He has this guilty look.” I hate to bust anyone’s bubble, but isn’t it more likely that he is trying to appease you because, well, you’ve sometimes been shouting at him when you got home for reasons he cannot fathom?
Dogs do not have moral sense, and definitely don’t know the difference between your Prada shoes as your old slippers. They think it terms of chewiliciousness, not dollars.
So what can you do?
- (if you find the mess too late) Nothing much, really. The damage is done. Make sure you don’t leave these things lying around when the dog is unsupervised
- Catch him in the act and then use a positive interrupter (see video below) to stop it
It’s disheartening, I know: instead of Lassie saving the orphans, you have to spell out to your canine genius that chewing shoes = not OK.
But the sooner you adjust your expectations, the more effective you will be.
2. Bad boy, good boy: how to keep them apart
This is a mistake I make again and again, so you’re not on your own.
If you reprimand your dog (Sometimes I’ll say ‘Hey!’ if I catch him doing something naughty) , try not to praise him seconds afterwards (for stopping the unwanted behaviour). Depending on the intensity of the telling-off (in my case, decidedly half-baked), he might come to see it as the predictor of praise. Perverse, isn’t it?
So, even if the dog complied quickly and willingly after being chastised, you would need to leave it a few seconds before starting the cooing fest again. To avoid that pitfall, how about interrupting unwanted behaviour with a positive interrupter (video above).
3. Root causes: Useful fellows
In some circumstances, you can take a good guess at the cause of a behaviour. Take barking:
- Is the dog guarding?
- demanding something?
- lunging towards another dog?
Often, knowing the root cause can help you find the appropriate response.
- Teach the dog to bark on demand, and then teach him to be quiet on demand. Then you can teach him to be quite after, say, three barks. But Rome wasn’t built in a day and you can’t expect him to get what ‘hush’ means without being taught first.
- Acknowledge the ‘threat’ the dog has spotted, so the dog’s mission is accomplished.
- If he is obsessing about something outside, try to distract him.
- Barking for attention or demanding something: Ignore him until a few seconds after he stops. If you shout at him, you will reinforce the behaviour: he got the attention he wanted.
- Aggressive barking: Please contact a behaviour therapist if this is frequent, or against a human.
- Excited barking:
- Wait until he has stopped barking before giving the goods (e.g. opening the door to go for a walk). If not, he may create a superstitious link between his barking and going for a walk.
- You may want to desensitise him to what is over-exciting him. You could repeatedly ring the doorbell for example. Without it being systematically followed by guests, or postmen, the doorbell will loose some of its predictive value that something out-of-this-world exciting is about to happen.
- Fear barking:
- Acknowledge the source of the ‘threat’
- Make sure the dog has the opportunity to move at a distance with which he feels comfortable
- Expose him to the source gradually. Do not push him beyond his limits.
- If the fear is a downright phobia, contact a behaviour therapist as do-it-yourself therapy can make matters worse.
- Lunging and barking at another dog: If you shouting, they might interpret that as you barking along. There are specific protocols to deal with dog-dog aggression, check them out.
I’ve used the example of barking here but my point is: think before you shout.
4. Proportionality: Are we overreacting?
The ‘alpha-roll’… That epitome of disproportional punishments. It involves forcibly rolling the dog over onto its back and pinning him down.
So why is it disproportional?
- If you really must look at (gray) wolves as models (see this article if you are interested in the wolf-dog fallacy), the closest thing to the ‘alpha-roll’, is the winner preparing to ritually tear out the throat of the looser (‘ritually’, it would not normally result in actual injury). Hefty stuff, though, you’ll agree. Kind of like the equivalent of pointing a gun to your face for spilling some milk.
- The “alpha-roll” is not observed in dogs. They do body slams, neck grabs, you name it, but not alpha-roll. What does happen is that some very submissive dogs roll over off their own accord. This is of course an entirely different thing.
The leash jerks are another infamous aversive method. I cringe when I see someone do this. I have no doubt that they love their dog, but it’s just plain misguided and barbaric.
The thing is, leash jerking does not seem all that effective. Read on…
- Leash pulling: Try Ian Dunbar’s green light – red light method instead. Stop walking when the leash is taut, and start again when it’s slack. It is a lot more effective (see how many people are still leash jerking after months of ‘training’?), and you won’t look like, well, an animal abuser.
- Lunging: Leash jerks can worsen lunging. It exacerbates the dog’s aversion for whatever made him lunge in the first place. Read this article if you are having issues with dog-dog aggression.
So before you get the H-bomb out, why not try to find a more targeted approach?
5. Generalisation: Yes, they really can be THAT stupid
Dogs are appalling generalisers. Seriously, appalling! They can conceivably understand ‘sit’ in the kitchen and have no clue in the garden. Skills need to be brushed up in different environments, they need to be “generalised”.
So, if your dog is performing poorly in a different location, don’t automatically assume that he’s being rebellious.
Just go back to basics and teach him the behaviour again, but in the new place. It won’t take nearly as long as the first time, and the dog will show a way more reliable response.
I am not suggesting that you teach your dog to sit on sand, gravel, grass and earth; in your kitchen, at your mom’s and at the training school. Teaching the behaviour in a couple of settings should be enough for the dog to ‘get it’ so he can apply it pretty much anywhere.
6. Fluency: How well does he really grasp it?
Dogs can grasp skills at varying levels of fluency. If your dog is not responding that reliably to something, it could be that he’s simply not that fluent in it yet.
How to test your dog’s fluency:
- Stretch several training sessions across a few days during.
- In each session, ask the dog to do the same thing about 10 times.
If he complies 8 times out of 10 consistently across a few days, it looks like he gets it. If not, revisit the previous training step. Sounds fairer than chastising him for performing badly on a skill that he does not quite master, doesn’t it?
This is the Lassie syndrome all over again: we would all love a dog who “gets it” immediately. But the majority of dogs are simply not that bright (especially mine… sorry Roger, Bacchus, Oli, etc.).
7. Extenuating circumstances: motivational state
If your dog is feeling stressed, distracted, or tired, it will affect his performance. Stressed, distracted and tired pretty much sums up the mood of most dogs in evening group training lessons.
So if your dog is facing more challenging circumstances, please cut him some slack and lower your criteria a bit. He is not challenging you. He is just not coping that well in more tricky settings.
If you really want him to perform 100% in these circumstances, then teach the fundamentals of the skill again, but in these settings.
8. Crime scene: Was there even a crime?
We are taught to vilify some of our dog’s behaviours when, really, the behaviours may be quite innocent.
Take growling, for example. As DogWilling’s Leah Roberts puts it: “Is it ok for you to say, “I’m afraid?” Then it’s ok for your dog to growl.”
Fair enough, growling is not always motivated by fear, but if it is, chastising your dog will make matters worse. Same for aggressive growling. If your dog growls frequently and you’re at a loss as to why, it might be time to contact a behaviour therapist, rather than attempting to correct the issue yourself.
Many other behaviours are seen as sinister when, really they are quite innocent. This is great parts due to the popularity of the fallacious pack theory. Couple of examples:
- Going through a door before us
- Nudging for attention
- Sleeping in an elevated place like the couch
Punishing a dog for doing this is akin to “bleeding out your humours” to cure a cold. It is so outdated and unfounded, and has no place in our society.
9. Gradation: Let’s go with a bang
If you’re going to use an aversive, go out with a bang.
Gradually shifting the intensity from a near-whisper to a full-blown shout does not make much sense. If the dog is being naughty (am thinking eating horse poo), it’s fine to start with one soft ‘no’, but don’t then gradually escalate if he doesn’t listen. Instead, go straight up to way higher volumes.
Seems counter-intuitive, as we are all striving for dogs who will stop with a whisper of a no, but you have to work on that. It won’t happen in a day. In the meantime, if he doesn’t listen that well on a particular occasion, avoid giving him fifty ineffective warnings.
“Going gradual” would result in the dog:
- Ignoring the gazillion warnings until it gets unpleasantly loud (learned irrelevance for us geeks). That ends up with you going: “no, No, Noooooo, Nooooo!, NOOOOOOOOOOOO!”. Instead of: “No!”
- If you are into physical punishment, consider this: you stand a real chance of having to use harsher and harsher punishment as the dog gradually gets used to increasingly rough treatment (habituation for us geeks). I say don’t put your finger in that vicious circle and just don’t get physical in the first place.
10. Violence: A touchy subject
Call me a New Age hippy if you will, but I cannot stomach violence.
Here’s what falls in that category as far as I am concerned:
- Leash jerks
- Electric collars
- Strangling collars
- Prong collars
- Electric fences
Don’t be fooled by benign-sounding products: if it uses pain to manage a behaviour, it’s painful.
A little side point on electric collars: they go off even when your dog is not barking. A loud truck or the door bell are enough to set it off most times. When it comes to excellent timing, I can think of better mechanisms…
Regardless of my personal views, here is a list of disadvantages of using aversive-based methods in training:
- Frequent violence leads many dogs to become fearful of their owners. A little study of their body language will clearly attest to that. Some dogs become “hand-shy” as a result of violence: they shrink away from approaching hands.
- Violent punishments exacerbate behaviour problems with most phobic, anxious, nervous or aggressive dogs.
- A dog will stop an unwanted behaviour quicker in response to aversive methods than through reward-based techniques, granted. But in the long-term, a behaviour taught through rewards is more resistant to extinction.
- Many dogs who are frequently trained with aversive methods will ‘shut’ down. They won’t be showing bad behaviours, but they will be the shadow of themselves, showing less curiosity, lust for life, or initiative. Do you want a robot, or a dog?
Before you worry about spoiling the dog, note that there is a flurry of positive-only training methods which are producing armies of good canine citizens.
But what it boils down to, for me, is this: If there is an alternative method with equally acceptable results, why would I choose the unpleasant one?
11. Be sneaky
But you said 10 points!!?? Sorry guys, I lied. I was reminded of this point by the author of the Koda Diaries, and it’s just too important to pass.
When it comes to punishing your dog, he could come to associate you with the deed. Try to booby trap the bin with something loud, or put plastic bags on your couch when you’re away.
So why sneaky?
- Best for your relationship: you don’t get so much of the “owner = sometimes nasty” thing going on, and
- The dog is less likely to indulge only in your absence, as many do.
For reviews on some of these books, go to Canis bonus dog book reviews.
- Linda Case – “the Dog, Its Behavior, Nutrition and Health”
- Ian Dunbar – “How to Teach a new Dog old Tricks”
- Steven R Lindsay – “Handbook of Applied Dog Behavior and Training”
- Patricia McConnell – “The Other End of the Leash”
- Pamela Reid – “Excel-erated Learning – Explaining How Dogs Learn and How Best to Teach them “
As ever, I value your comments. I would particularly like to hear your view if you:
- Are you a dog professional having to frequently position yourself in that minefield of a subject?
- Are you confused about, firmly against, or have reservations about gentle training methods?
- Are you, like me, a dedicated advocate for gentle methods?