Last Spring I wrote this post about how you can’t reinforce fear.
I wanted to revisit the topic today, though, because I still see this idea everywhere. Also, a conversation I had yesterday on Twitter with a blog-less reader made me want to add a few thoughts to my original post.
“Don’t give the dog a cookie while she’s scared! You’ll make her worse!”
Reinforcement increases behavior. Fear is not a behavior; it is an emotion. We don’t choose to have or not have our emotions, and dogs don’t have our ability to talk themselves out of their feelings. Technically, you could reinforce a fear-based behavior, like cowering, but that assumes a level of active awareness on the dog’s part that I personally haven’t seen. Could you exactly replicate the way you jump when someone slams a door behind you?
One of the reasons that the “reinforcing fear” myth is so pervasive is that there is a distantly related scientific truth. You will, absolutely and without a doubt, have the best results from counterconditioning/desensitization if the dog is what behavioral scientists call “under threshold.” That is, the ideal time to deliver your reward is to do it before the dog starts reacting to something. So, yes, once your dog is already cowering or barking, you’re behind the curve. In a perfect world, you would deliver every reward while the dog is calm or happy.
Let’s get this clear, though:
The absolutely worst thing you can do for a fearful dog is to do nothing.
So, you make a timing mistake. Your dog sees or hears something that you weren’t prepared for. Maybe you have a situation like thunder phobia where there is no “milder version” or “greater distance” to work with. If you believe that you will reinforce fear by delivering a cookie, petting, praise, or even by getting the dog out of the situation, your dog’s behavior will deteriorate. This is science.
Once your dog is reacting fearfully, he is over threshold. Not only is being over threshold bad for your dog physiologically, it will sensitize your dog to future encounters. His threshold for future fear reactions will lower. Sensitization and the lowering of thresholds is bad enough for problems like leash reactivity, but for conditions like separation anxiety or thunder phobia it can be disastrous. For any fear, once that threshold gets low enough, your only choice will be to medicate, because it is no longer possible for your dog to be safely counter-conditioned otherwise.
Now that your dog is over threshold, you have two choices: you can either do nothing, because it “reinforces fear” (bad idea), or you can deliver an admittedly sub-optimal reward. What your poorly-timed cookie/praise/petting/escape will get you is the chance that your dog will go back under that threshold. You may or may not get long-term learning out of it (whatever the books say, counter conditioning in the real world is hard, imprecise work), but at the very least you are stopping the damage.
Your frightened dog is not making a choice that you can validate or (heaven help me) punish. You are helping your dog–a creature who completely depends on you–handle a bad situation, whether he’s coping in an ideal way or not.
4 thoughts on “You Can’t Reinforce Fear, Continued”
Yes yes yes times infinity.
I mentioned this yesterday, and I hope you don’t mind if I leave it here: “How Can She Possibly Be Eating” by Eileenanddogs offers some practical discussion of those difficult situations.
As you say, we can’t get it right all the time. But if we have a strong foundation of behaviors and a trusting relationship, we can at least prevent further damage from being done and help our dogs feel a little better. And heaven forbid we punish the poor dog indeed.
Additionally, I know I’ve harped on this before, but I’ve seen first hand that if you are lucky enough to have a dog who can eat when stressed or over threshold, you can even make progress with sub-optimal counterconditioning. Eileen actually has an article on that, too, about Summer’s thunder phobia.
I agree with your last. Suboptimal, but I do think you can get some progress with it. And, of course, sometimes it really is all you’ve got.
Interestingly, I think what happens for a lot of dogs is that the process of you getting the cookie out of your pocket is so conditioned that it starts to calm them down before the cookie goes in their mouths. And then once they are chewing the cookie, the dog’s body starts releasing happier chemicals and you get some real work done. But, who knows? At some point it’s an empty intellectual exercise.
I also think (based purely on my anecdotal experience) that you can get a pairing or stimulus+food long before the stimulus actually seems happy to the dog. Silas always gets kibble while the landscapers are here running their interminable leaf blowers. It’s the only thing cheap enough to hand out in adequate quantity. I’ve had days where I can’t get him to eat at all from his stomach problems, until the leaf blowers come and he eats his kibble like it’s delicious. “Oh, that noise means I eat!” But if I hand out the kibble too slowly, he will go right over to bark at the landscapers.
Eileen’s Clara is such an interesting dog, because it doesn’t seem like she has a naturally fearful temperament. She’s like the anti-Silas, who was actually fairly well socialized to most things (definitely a few holes and some rookie mistakes) but whose instinctive first response is fear.
I think your theory makes a lot of sense! My trainer concludes her Level 1 class with Suzanne Clothier’s Really Real Relaxation Protocol, and I think that building a conditioned response to calm treat deliveries is one of the things it’s taking advantage of. That’s just a guess, since I haven’t, you know, asked her.
And yes, my anecdotal experience agrees that stimulus+food happens faster with a smart, scared dog than the emotional response, especially in those suboptimal situations. But I’ve heard people (including a relative who is a vet!) disparage CC as useless, or not worth trying, because we may not always be able to manage threshold, or may miss a pairing now and then, and I always want to preach against that even if it is to the choir, so to speak. Any possible improvement on panicked barking zoomies and a day of hypervigilance is worth trying for, I think.
I think of all the times our poor golden retriever Moses suffered through thunderstorms when we thought we weren’t supposed to comfort him. I’m so glad we learned better before we lost him. We had at least always taken him down to the basement when there was a storm to get him someplace quieter. Sometimes you have to trust your instincts when what you’re doing seems totally wrong, like not comforting a dog who is afraid.