Classical conditioning examples of dogs and other animals on YouTube are rare. And there’s a reason. It’s because the process is generally comparable to watching grass grow. The creature being conditioned isn’t necessarily doing anything. The action is often off camera. And you have to do many repetitions to see any results.
You could make a 2 minute video of shaping your dog to blow bubbles in a bowl of water and that is fun and impressive. Lots of action, and look how fast the dog learns! It might go viral. Or you could show your dog doing nothing at all while you feed her because the doorbell is ringing. I’m betting it won’t go viral, but one can always hope. Here’s an exciting picture of classical conditioning in action.
I seem to have a knack for stumbling into these weird little holes in the Internet video world. I like filling them with videos. After I published The Barking Recall, several people asked me to show how I taught Clara to lick her chops, wag her tail, and look for treats instead of joining in when Summer barked. I figured I’d give it a try, boring or not.
The process can be summed up in one sentence. Starting when Clara was a puppy, every time Summer barked, no matter what Clara was doing, I gave her her favorite treat: canned cheese.
The process is pretty straightforward. There are two important factors though: timing and consistency.
The timing is that the treat has to come after the event. That sounds easy, but in the real world it can get tricky. If you are conditioning your dog to some visual event, say the appearance of a silent scary monster, if you see it first and start scrabbling for the food, your dog’s experience can be seriously messed up. To her the food starts coming, then the scary monster appears, instead of the other way around. Food predicts monster. If you do that enough times your dog could end up getting worried whenever you reach for a treat and still be scared of the monster. Not good. But in our case, since I wasn’t likely to know before Clara that Summer was barking, I didn’t have any problems like that.
The other factor, consistency, can be hard if you work outside of the home. To convert a neutral or scary thing to a good thing by associating it with other good things, you have to provide them virtually every time the event happens, at least at first. That’s why I have never tried to condition Summer out of her fear of delivery trucks. For her to learn that the sound of delivery trucks on the street predicts wonderful things, I would have to be there to provide the wonderful things a large majority of the time. Poor Summer; I just can’t prevent the trucks from coming when I am at work, or take enough weeks off to be there every time.
Again, I was lucky with Clara, in that I was home a lot when she was a puppy, and that I could separate her from Summer when I was not home.
A few other things bear mentioning.
First, note that I am referring to classical conditioning, not counter conditioning. Counter conditioning is when you perform this same process on an animal who already has a negative emotional reaction to the event (like Summer and loud trucks, or Zani and the elliptical exercise machine). The process is the same, but even more attention must be paid to consistency as described above, and it is likely to take longer. Here is a very thorough video on Counter Conditioning Leash Reactive Dogs by Drayton Michaels. And for a wonderful seminar about counterconditioning aggression in dogs, there is Kathy Sdao’s Cujo Meets Pavlov: Classical Conditioning for On-Leash Aggression DVD set.
I had the great fortune to start conditioning Clara when she was a puppy, so I had a blank slate. She was certainly attentive when another dog barked, but she was too little to join in a barking frenzy, which was the otherwise inevitable behavior I was aiming to prevent.
Another issue that commonly pops up in discussions of classical conditioning is concerns about accidentally performing operant conditioning instead. Once we humans learn about operant conditioning, it’s hard to understand why classical can trump it. Case in point: what if Clara was chewing on Zani or peeing on the carpet right when Summer was barking and I gave her her canned cheese? Cringe. Who wants to do that? Aren’t we reinforcing the biting or peeing? In general, no. Here’s why. Let’s say Summer barked 50 times during a week and I gave Clara canned cheese each time. During each one of those times she was doing something, even if it was just sleeping. But it was a variety of things. During one of those times she was biting Zani. During another she was peeing on the carpet. So she got 50 pairings of Summer barking and cheese, but only one of biting and cheese and one of peeing and cheese. And the barking/cheese relationship still existed even when she herself was biting or peeing. So she is well on her way to learning that Summer barking predicts cheese, no matter what she herself is doing. That will trump the individual behaviors she is performing.
For fun, here is a link to an entertaining video of a Psych project where a man transforms a neutral event, the sound of a cow mooing, into a predictor that he will turn off the TV that his children are watching. He is conditioning a negative emotional response to the mooing sound. You can watch the short video and realize that what the children are doing when the moo sound happens is irrelevant. Even though something bad happens, what they are doing at the time does not get punished. The consistency of the TV going off after the mooing sound makes that the relevant pairing, and it doesn’t involve the children’s behavior.
Finally, here’s some clarification about what the classical response actually is. Classical conditioning is also called respondent conditioning. The internal changes we are seeking have to do with involuntary, respondent behaviors. Sneezing, startling, drooling, or experiencing an emotion are all things that happen without our volition. So if we are conditioning with food, we are aiming for a gustatory response. That means the body is preparing to eat, which causes hormonal chain reactions and preparation of the relevant organs and body systems. This response is mostly internal and invisible.
So how do we know if it is happening? Pavlov was lucky he had dogs, because many of them visibly drool in anticipation of food. If he had been using, say, lizards, would we know about respondent conditioning today?
What if you have a dog who doesn’t drool? Or perhaps the dog is having the gustatory response but not quite to the level that would cause drool. We can extrapolate that the invisible internal response is occurring from the operant responses that we do see. Clara now runs to me, wagging her tail, when Summer barks. Those are operant behaviors that are extremely unlikely to occur consistently without there having been some trigger other than the barking. They are not a common response to a barking dog. But they are likely caused by the conditioning, and pretty reliable indicators of it. In Clara’s case, even licking her chops is operant, but it is a direct result of the commencement of salivation.
So here’s my video on the method of pairing barking and wonderful treats. I hope it is more interesting than watching grass grow.
SPOILER ALERT: At the end of the video I test Clara’s response on camera when I am not home and someone impersonates the mailman. Both dogs hear the person on the porch. Summer immediately barks and continues to do so. Clara jumps to her feet in her crate and stands there, listening. She does not bark. Neither does she drool and wag her tail. But I am pleased with her response. Someone coming up on the front porch is a event a dog would normally pay attention to and I’m glad she did. She was extremely attentive, but she did not get into a frenzied state or “catch” Summer’s barking. I think this is likely due to the conditioning and its result that she never practices that behavior when I am home. Without the automatic barking and arousal, she was able to make an assessment (anthropomorphically speaking) that the sound of the mailman was the same as it is every day, and it has never predicted anything bad.
And in case you’re wondering: this conditioning did not inhibit Clara’s barking in response to various events that she herself finds alarming or exciting. There are plenty of those episodes. The conditioning just inhibits her from joining in contagious, group barking.
If you didn’t see my post The Barking Recall, you might want to check it out now. It features a video that shows how barking eventually became a trigger for Clara to check in with me, even when I am out of sight, and how it helped create Clara’s strong recall.
Thanks for reading!
- When Premack is not the best choice
- Body language resources
- Est-ce que votre chien veut VRAIMENT être caressé ? (Does your dog REALLY want to be petted?–the French version!)