Photo by Lisa Fotios
Yeah, I know – dog trainer advice can be confusing sometimes.
I mean, I’m always going on and on about how you need to reward behavior you like, and make sure you don’t accidentally reward unwanted behavior. Whatever gets rewarded gets repeated, blah blah blah.
But then in the next breath I’ll tell you something like: “if your puppy growls at a stranger and then tries to hide behind you, you should let her hide behind you. AND give her treats while you’re at it.”
Huh? Doesn’t that mean you’re rewarding unwanted behavior?
Well, no, but I totally get why someone would think that. Let’s talk about this issue real quick and clear up the confusion.
There are two separate learning systems at play:
1. Operant conditioning: the realm of voluntary behavior
Operant conditioning is where an animal voluntarily adjusts their behavior based on the consequences of that behavior.
Examples of operant conditioning:
- The puppy sits. Consequence: you give him a treat. Behavior change: the puppy is likely to sit again.
- The dog begs at the table. Consequence: You can’t resist those big brown eyes and give him a tidbit. Behavior change: the dog is likely to beg again.
- The dog begs at the table. Consequence: you ignore him, wait for him to go to his bed, then toss him a treat. Behavior change: the dog is more likely to go to his bed during future dinners.
- The toddler puts their hand on a hot burner. Consequence: they get burned. Behavior change: the child is less likely to put their hand on a burner again.
- Teenage me goes to karate class. Consequence: I break my foot. Behavior change: no more karate for me.
Reinforcement and punishment fall under the category of operant conditioning. No matter what methods a trainer uses to train behaviors -whether they rely on treats or prong collars- they’re using operant conditioning.
2. Classical conditioning: the realm of emotions and other involuntary responses
This is where an animal develops an involuntary response to a stimulus. This is the Pavlovian response we’re all familiar with, named for Ivan Pavlov’s experiments where he found that his laboratory dogs would salivate at the sound of a bell that predicted food.
Examples of classical conditioning:
- When Sparky goes to the vet, unpleasant things happen. He gets poked and prodded, maybe even abducted by a stranger and taken into the back room. Now every time he is brought into the clinic, he panics at the door. In severe cases, this can extend to panicking at the sight of the car, because getting in the car means going to the vet.
- You grab the leash off its hook when it’s time to take Sparky for a walk. Eventually Sparky learns that you grabbing the leash means he’s going out, so he develops a SUPER EXCITED reaction to seeing you grab the leash.
- Your wonderful grandmother used to bake blueberry pies when you visited her. Now when you smell blueberry pie, you feel happy.
- Your HORRIBLE grandmother used to bake blueberry pies when you visited her. Now when you smell blueberry pie, you feel sick.
This is why it’s not a great idea to set your favorite song as your alarm to wake you up in the morning to go to the job you hate. You might be trying to make the experience of waking up more pleasant, but what you end up doing is making yourself hate your favorite song.
Desensitization and counter-conditioning, the top methods of handling fear and aggression in dogs, fall under the umbrella of classical conditioning.
How do you know which one to use?
When you’re trying to change behaviors that are annoying to humans but innocuous to the dog, we use operant conditioning to control the consequences: make sure unwanted behavior doesn’t get accidentally rewarded, and train and reward the crap out of behavior you’d prefer. Behaviors we handle this way are stuff that is normal for dogs, but that humans consider bad manners. Stuff that pretty much all dogs will do unless taught otherwise.
- Destructive chewing
- Peeing on the carpet
- Digging up the garden
- Jumping on visitors
- Pulling on leash
- Stealing food off the table
- Puppy biting
For more on this approach, see: How to Solve Practically Any Annoying Dog Behavior Problem
But when changing behavior that results from a dog being distressed, we have to address the emotion first. A dog who is trembling at the vet’s office is in a very different state from the dog who is begging for food at the dinner table, and needs a very different training approach – one based in classical conditioning.
Can you accidentally reinforce fear?
Emotions are the territory of classical conditioning. You can’t “reinforce” anxiety, fear, anger, or happiness the way you reinforce a behavior.
Sure, technically, if my dog growls at children, I could smack him on the nose in punishment every time until he no longer exhibits the behavior of growling. But I’ve done nothing to address the distress that caused the behavior – and in fact, I’ve probably made it worse, because now he’s been classically conditioned to associate children with being smacked on the nose. Now I just have a miserable, silent dog who has been trained not to give warning before he bites. Oops.
Going back to our example from the introduction, let’s say I’m out walking my shy puppy. We encounter a nice stranger who makes delighted “ooooh puppyyy!” noises and wants to meet her. But the puppy is spooked. She growls and runs to hide behind me. Should I let her hide behind me, or will that just reinforce the growling and hiding?
The growling and hiding are symptoms of the fear she’s feeling. To make the growling and hiding stop, I need to make the fear stop first. So I let her hide, and give her a reassuring pat. Voilà, she’s gained the comfort she sought, and is feeling less fear! After a moment, she might feel brave enough to come out of hiding and approach the stranger. If she doesn’t come out of hiding, I bid farewell to the stranger and move on with my life, secure in the knowledge that I haven’t further traumatized my puppy by betraying her trust and forcing her into an interaction that terrifies her.
(This isn’t a hypothetical scenario, BTW. The shy puppy in question is my Belgian malinois, River, and this is how we got her over her puppyhood fears. See: How I Survived the First Three Months with a New Puppy – and You Can, Too!)
Emotion and behavior run on separate, but adjacent, tracks
To use another train metaphor, you can’t have operant conditioning without classical conditioning hitching a ride.
The toddler who touches the hot burner is operantly conditioned not to touch hot burners. And you can bet he’s also been classically conditioned to have at least a little bit of a fearful reaction to stove tops.
When training your dog to come when called, to make the behavior as rock solid as possible, you want to develop a strong classically conditioned emotional response of “YAAAAY!” when the dog hears the recall cue. (See: Training Your Dog to Reliably Come When Called is Hard. Here’s How to Make it Easier)
The puppy whose new owner teaches her to do all her basic commands for treats will also develop a classically conditioned happy response to her owner; she associates her new owner with the fun reward-based training exercises they’ve been doing.
So whatever operant conditioning training methods you choose for your dog, ask yourself: what classically conditioned emotional response is going to come along for the ride?
- When a behavior is not based in distress, use operant conditioning to change it
- When a behavior is based in distress, use classical conditioning to stop the distress
- Emotion and behavior run on separate, but adjacent, tracks
- You can’t reinforce fear or other emotions
- Any time you operantly condition a behavior, you can be certain a classically conditioned response is also being “installed”
Things can get a little more complex and nuanced than this, but as long as you understand the fundamental difference between classical and operant conditioning, you’ll be better equipped to make good decisions for your dog’s training.