Revised March 2019.
I collaborate with a vet who claims that many of the dogs that visit her facility are so eager to get inside the door that they pull their owners by the leash all the way from the parking lot.
Six syllables. I know, most people tune out beyond four. But I still think you should learn this particular term. Why?
Because it’s one of the most important techniques in animal training (arguably top five).
Simply put, conditioning means learning and counter means opposite.
So, re-learning might be another way of putting it. That’s just three syllables.
Practically speaking, it’s about changing someone’s learned associations.
Let’s say we have a dog, who’s started trembling and panting whenever she arrives to the vet’s office.
She has probably learned to associate the vet’s office with aversive events.
Strange sounds and smells.
Unfamiliar people looming.
A string of events that ends up with something painful – most animals will learn these predictors. This is one type of classical conditioning, learning that certain events predict things the animal would rather avoid (just like Pavlov’s dogs drooled when they learned that bells predicted things they really wanted).
Some animals may start showing behaviours indicative of a fear response at the first stimulus in such a predictive chain, and some even resort to defensive aggression in their attempts to escape the situation – which could be very risky for all parties involved.
Counterconditioning (CC) is about re-learning. Typically, rather than predicting pain, the animal learns that certain events predict outstanding free delicacies, delivered right under your nose.
Let’s go back to our example. Through counterconditioning, the dog now re-learns that:
Strange sounds and smells – but followed by fabulous treats.
Unfamiliar people looming – and then you get something that smells just wonderful!
Needles poking – is that chicken liver? (or, if you’re a cat: whaaaat – Tuna?!)
Studies have shown that CC may reduce the risk of defensive aggression in just a few training sessions.
For example, one dog’s aggressive behaviours towards a stranger at the door (charging, lunging, barking, and biting) diminished from 88% of 30-second intervals before training to 3% after CC training.
How to perform counterconditioning
How do you do it, then?
In theory, it’s easy.
- Start by identifying some fabulous treat – something that the animal really likes.
- Then feed small mouthfuls to the animal as soon as it has been exposed to a potentially scary stimulus.
The fear-inducing stimulus should predict the arrival of something fabuluous.
- After a few pairings (scary thing – fab food) the animal learns that one leads to the other.
She re-learns that the potentially scary thing is actually not scary but a predictor of something spectacularly good.
Take a look at dr Sophia Yin counterconditioning a dog to actually enjoying something he initially thought aversive. She discusses – and dismisses – the common assumption that giving treats to animals while or after they’re aggressive inadvertently rewards, or reinforces, the behaviour.
Note that the animal doesn’t have to DO anything. In classical counterconditioning, the potentially scary thing precedes the wonderful treat – regardless of what the animal is doing.
In theory, counterconditioning is easy.
In practice, unless you know what you’re doing, counterconditioning is quite difficult.
Actually, there are at least three ways in which counterconditioning can go wrong.
Problem one: choosing the wrong procedure
We sometimes mistake hyper arousal and happy-excited emotional states for fear.
In such cases, presenting a stimulus that’s not scary, and then treats, might have little to no effect on the animal’s behaviour – or even increase arousal. Such procedures might inadvertently reinforce whatever behaviour is being shown, and solutions other than counterconditioning are called for.
Second problem: exposure above conflict threshold
Let’s walk through this procedure.
The animal is exposed to something potentially scary (and it might be more scary than you think). Step one.
The animal is then exposed to something nice (and it might be less nice than you think). Step two.
Through repeated pairings (step one-step two), the animal learns that the potentially scary thing is actually not scary at all but a predictor of nice things.
So, two questions:
- How scary is the scary thing?
- And how nice is the nice thing?
The answers to those two questions are likely going to determine your success in counterconditioning.
The contrast between the scariness and fabulousness of the two respective stimuli will determine the outcome.
- If the scary thing isn’t that bad, you can probably get away with less-than-fabulous treats and still countercondition successfully.
- At the other extreme, if the scary thing is scary enough, you won’t be able to countercondition successfully regardless of how faboulous your treats are.
- At some intermediate level, the animal gets into conflict. The scary thing is really scary, but the treats are really fabulous too, so the animal may eat the treats – but remain fearful.
The thing is to ensure that the level of scariness is below the conflict threshold – we don’t want the animal to overcome his fear, we want him to re-learn: the potentially scary thing should become a predictor of great things.
Now, we might overestimate fabulousness, but I’d say that the main problem is that we often grossly underestimate scariness. This boils down to two things:
- We fail to see or interpret the subtle signs of fearful body language in our animals.
- We’re not afraid of whatever the animal is afraid of, so it doesn’t even occur to us that they might be fearful.
Animals are more easily frightened than people: they often react fearfully to novel stimuli or things that we can’t even perceive (smells or sounds, for instance).
Or they might respond fearfully to things we wouldn’t ever consider scary – so we might dismiss their fears.
The thing is, if they are too fearful, they’ll either get into conflict, or they won’t take treats – and CC alone won’t work.
I’ll get back to what to do about that below, but first, the final main obstacle:
Problem three – asking for behaviour too soon
As one of my readers pointed out, an additional problem is that people tend to use operant counterconditioning rather than classical counterconditioning – and that’s risky if the animal is too fearful.
Oh no, even more syllables? Hang in there, I’ll explain! 🙂
In classical counterconditioning, the animal doesn’t have to do anything in particular, step one precedes step two (scary thing – treat) regardless of behaviour. As we saw in Sophia Yin’s film, the air-in-the-face stimulus preceded the treat. And meanwhile, what was the animal doing?
He was raising his lip, giving a warning signal. At least initially.
Many people might think: “I don’t want to reward that. I’m going to wait until he stops. When the lip goes down, that’s when I’ll present the food”.
But that’s operant training, folks. That’s requiring the animal to perform a behaviour in order to get reinforcement.
Sophia Yin didn’t do that in her video, though. She just got the treat to him as soon as possible – despite his raised lip.
Counterconditioning is primarily about re-learning predictors, its main purpose is not about re-learning behaviour. The dog isn’t learning that “if-I-behave-politely-despite-air-in-the-face-I-get-treats”, he’s learning that “air-in-the-face-precedes-treats”.
He’s learning a predictor. One stimulus predicting another – regardless of behaviour. Classical conditioning. Or in this case, classical counterconditioning.
Technically, Sophia didn’t use the treat as a reinforcer for desired behaviour, she used it as an unconditioned stimulus (the air-in-the-face being the conditioned stimulus).
The purpose of the exercise was not to get polite behaviour, but to change the emotional state related to air-in-the-face. And once the emotional state changed, the behavioural manifestations changed too. So those warning signals disappeared, since the animal no longer feared air-in-the-face – he’s learned that it predicted treats!
The dog stopped giving warning signals, because his emotional response to air-in-the-face changed.
But what does operant counterconditioning look like then, and why is it risky?
It’s asking the animal to do something, show some behaviour, in the presence of something potentially scary – and get rewarded for it.
It might be “sit” on cue. Touch a target. Perform some trick. In an environment containing stimuli that might be aversive.
This is often done later in the counterconditioning procedure. Once classical counterconditioning is achieved, we may ask the animal to perform a behaviour in the presence of the previously scary thing, to ensure that it is, in fact, no longer scary, and the animal feels confident enough to respond to the trainer’s cues in its presence. Failure to respond to such cues may indicate that there’s some residual anxiety, perhaps.
So, classical counterconditioning procedures are sometimes followed by operant counterconditioning.
The problem occurs when people start doing operant counterconditioning too soon. They might require that the animal *not* balk, rear, vocalize, or lift the lip in warning. So they might present the scary thing, and when the animal balks, they withhold the treat, hoping that he’ll stop soon so they can deliver the treat.
And if the animal keeps balking or showing other unwanted behaviour, the treat doesn’t appear at all.
So, rather than learning that the scary-thing-predicts-treats, the animal learns that in-this-context-scary-things-happen.
And so, he might sensitize, and escalate the undesirable behaviour.
sidestepping the problems
What to do then?
Combine counterconditioning with another technique – systematic desensitization (SD): gradually introducing the scary thing.
The combination of counterconditioning and systematic desensitization is called SD/CC and is potentially the most powerful tool available in reducing or eliminating fear – in people as well as in animals.
If you present the potentially scary thing, and the animal shows some unwanted fear-related behaviour:
- deliver the treat regardless (so that he still learns that scary-things-predict-treats)
- in the next repetition, reduce the exposure of the scary thing below the conflict threshold (so that he doesn’t show any detectable signs of unease)
Note that the procedure of creating a positive association with the veterinary visit is best done even before animals develop fears.
Strictly speaking it might not even be “counter” conditioning or re-learning if the animal isn’t afraid to begin with, just “plain” classical conditioning (Pavlov’s drooling dogs again) – but such preventive learning is very powerful in buffering against later fears developing.
The vet should be able to offer a vast assortment of fabulous treats, in order to bring about this important learning. And there’s no scientific support for the old adage “the animal must fast before sedation, so we can’t feed treats in case we need to sedate the animal” – in fact, it might be more risky not to feed.
Many modern animal trainers actually expand this procedure to include giving the animal the option of giving the go-ahead, effectively communicating “yes I’m ready to be exposed to the scary thing”. Giving the animal control, as it were.
… On a side note, all you nerd trainers out there: which are the five most important words (or expressions/concepts) in animal training, according to YOU? Take a moment to give us your thoughts in the comment section below!
I give online courses about getting happy, reasonably well-behaved animals that thrive with people. Wanna learn more, and get information about when they’re available (and also get info about new blogposts and free webinars)? Sign up below and I’ll keep you posted!
Savage, K. E., 2010. A Comparison of Classical Counterconditioning and Differential Reinforcement of Alternative Behavior on Aggressive Behavior in Dogs.