Tag: Pavlovian conditioning

I Got Trained Like Pavlov’s Dogs—Then Things Fell Apart

I Got Trained Like Pavlov’s Dogs—Then Things Fell Apart

A black and rust colored dog lies on a pink mat. Dog is lying on her side and side-eying the camera

Rinnnggg! I learned to expect something nice when I heard that sound. Then things went south.

The Sound

When I first started out as a blogger in 2012, I used a hosting site called WordPress.com. Their smartphone app has a pleasant little notification sound effect. I soon learned that the app played the sound when I got comments, likes, or follows.

Here’s the sound effect.

The sound is an arpeggiated C major triad, in the 6/3 position, pitched high (the lowest note is E6 at 1,318 Hz), with a timbre resembling a celeste. For most people accustomed to Western music, it would be a fairly pleasant sound, a lot more pleasant than, say, a buzzer.

Positive Feedback for Blogging

Getting positive feedback is fun for any blogger. But when you are just beginning and have no idea whether anyone will want to read what you write, it’s thrilling to find out that someone likes it well enough to follow. Or when they simply press the Like button. Or the absolute best, when they leave a positive comment or a question.

I didn’t realize until I started blogging how important comments are. When you write, you put your stuff out there and hope people read it. Encouraging comments act as positive reinforcement. You want to publish more, and to do that you have to write more! It was a great feeling whenever I found out that something I wrote helped somebody and their dog.

I feel lucky (most of the time) to be a writer today when immediate feedback is possible. I think about the writers of yesteryear, for whom positive responses often came only after they were dead, if then. But I can write a post and get responses on the same day.

The Classical Association and How It Was Built

You can see where this is going, right? Here’s what happened when I first started blogging and got the WordPress app.

  • **Chime**: I look at my phone and see that someone liked my post
  • **Chime**: I look at my phone and see that someone followed the blog
  • **Chime**: I look at my phone and see that someone made a positive comment
  • **Chime**: I look at my phone and see that someone liked my post
A hand holds a smartphone and a bunch of like and other symbols float in the air above it

Et cetera. We’ve got both operant and classical conditioning going on. That’s always true, but it’s especially easy to see in a situation like this. I look at my phone when I hear the chime and get reinforced for doing so. But I also get a great feeling about that chime.

The chime was meaningless the first time I heard it (a neutral stimulus) since I didn’t know what it predicted. After a few repetitions, it predicted social approval. After a couple dozen repetitions, I started getting a surge of happiness when I heard the chime!

This is one of the clearest examples to me that the stuff that goes on with our brains and emotions is chemical. I could feel happiness wash through me when the chime played. And you can bet that whenever possible, I grabbed my phone to see what had happened. The pleasure that had at first come from a like or a follow or a friendly comment had moved forward in time. It started surging in when I heard the chime—even before I saw what had arrived on the blog.

The WordPress.com notification sound is custom, not shared by other phone apps to my knowledge. It’s beneficial for their sound effect to stand out. For me, as the end user, it facilitated the classical conditioning. It meant that the pairing of “the chime” with “cheerful news about my blog” was completely consistent, so consistent and distinct that I could feel my body chemistry change when I heard it.

Expulsion from Eden: The Association Changed

Portion of Michelangelo painting Expulsion from Eden: A serpent with a woman's head is wrapped around a tree.
The serpent from Michelangelo’s Expulsion from Eden

So it was bound to happen, but I didn’t see it coming.

What happened when I got my first nasty comment on the blog?

I heard the chime and got the thrill of joyful anticipation. I looked at my phone to see what had happened. I got an eyeful of vitriol! My mind and body were primed for a treat, and I got hostility.

The happy brain cocktail had started, but cut off as I felt an unpleasant flush. My skin got prickly. A wave of nausea washed over me. I was upset and hurt.

I sound like a real baby, and maybe I am. But the above is the best description I can give of my feelings. And from my amateur observations, it may be similar to what my dogs go through when disappointed and hurt as well.

I had been floating along in a honeymoon period, and it was not in my mind that someone would respond unpleasantly. Too bad WordPress couldn’t assign a different sound to nasty comments, eh?

The important thing was that it only had to happen once to completely change my reaction to the sound.

The next time I heard the chime, I had an unpleasant dual reaction. I momentarily had the old response, then the new unpleasant one washed in. The prediction of good stuff no longer held, and the purity of the chime was history.

And worst of all, there was still a prediction! Something was waiting for me! But was it a nice thing or an icky thing?

My negative commenter didn’t leave right away, so the negative feelings started being my principal response and the joyful reaction faded. Instead of happily reaching for my phone with a slight sense of euphoria, I looked at it with dread.

The Association Changes Yet Again

Fast forward a few months. I had had no aggressive commenters for a while, so when I heard the chime, I usually looked forward to checking out what was going on. I would never regain the pure joy reaction, but the chime had moved back into the positive side again.

In June 2013, I got an email from the WordPress.com staff that one of my posts was going to be featured on Freshly Pressed, the daily WordPress showcase. It was thrilling to have a post chosen out of the millions published each day. They didn’t tell me the date of the feature in advance, but I knew exactly when it happened because the chime on my phone blew up. It went off constantly for more than an hour. Wow! My post had been showcased for a potential audience of millions. All sorts of people outside the dog training community, including other writers, read my post and many followed my blog!

The chime went off at a very high rate for more than a week, and there weren’t any comments that were exceptionally hard to deal with, so all was well.

Cans of Spam on a grocery shelf

But about a month later, I noticed something. The flurry hadn’t quite died down, but my new followers didn’t look like real people from their usernames. This took a while to sink in. But when most of the usernames were things like reebok4ever, vi_gracheap, and gucciandcoach, I started to get it that not everyone who followed the blog or liked a post was passionate about dog training. They were interacting on the blog for a different reason. These bots and spammers would like a post because their icon and a link to their website could appear in a list at the bottom of the page.

Soon most chimes were predicting these spammer likes and follows. They greatly outnumbered serious followers, and I wasn’t getting any comments. So the chime became meaningless. Why would I want to know when another non-entity followed the blog?

I turned off the chime.

Dogs

This post isn’t just about me.

As a human, I have a big cerebral cortex and some cognitive skills that are unknown to dogs. I can reason and predict and justify. But I experienced the change of the chime physically, and the switch from yay to yuck was very unpleasant. Dogs have similar neurological chemicals and reactions to those of humans. And I can only imagine what it would be like to go from trusting that something great was about to happen to finding out that I might get whacked, without the cognitive skills to understand what was happening.

This is the classical conditioning version of the operant poisoned cue. I’ve written about the effort I made to replace such a cue that was negatively affecting my dog. Now, when I establish a classical pairing, or assign a cue to a behavior, I make sure in both cases that they predict only good things. Not only for effective training, but to be fair and kind to my dogs.

Here’s an example of a situation that could have gone south, but I managed to not let that happen.

I reinforce my dogs generously for getting on their mats. Most times, the mat itself is the cue. I reinforce “offered” mat behavior. So little Zani, who ceaselessly sought goodies from me, decided when we first got up in the morning and headed to the back door to run ahead of me and lie down on every mat. She was such a clever little cuss. Trouble was, she got underfoot, and some mats were in my way. I caught myself many times wanting to fuss at her for plopping down in front of me on a mat. There I was, stumbling sleepily along. I thought, damn, she should know better!

A black and rust dog is lying on a navy blue mat holding a sports shoe and looking directly at the camera
Zani on a mat with a shoe: a double bid for reinforcement

But she was doing exactly what I had daily reinforced her for doing. Mats predicted nice things happening. I hadn’t put mat behavior on stimulus control. And I was the one who put the mats in the walkway.

I know I mashed up operant and respondent learning in that example. But it was mashed up in the chime example, too. I have reinforced my dogs for being on mats so much that mats are classically conditioned as good, happy places.

So did I really want to create a similar nasty experience for my dear little dog? Did I want to switch without warning from “mats predict great things” to “getting on a mat can make Eileen pissy”?

No. Never. I didn’t want to dilute the power of her cues. I wanted that happy brain cocktail for her as part of our interactions always. And I still want it for all my dogs.

Copyright 2013, 2022 Eileen Anderson

This post was first published in 2013 under the title “Goodie or Doodie: When a Classical Pairing Gets Stomped On.” I’ve rewritten it substantially.

Spam photo from Wikimedia Commons courtesy of Angry Red Hammer Guy under this license. I cropped the photo, which originally showed that the Spam was misplaced in the Kosher section of a grocery store.

Serpent photo from Wikimedia Commons is in the public domain.

The smartphone illustration is from CanStock Photo.

The two photos of Zani are copyright Eileen Anderson.

Which Pavlov Is on Your Shoulder?

Which Pavlov Is on Your Shoulder?

Which Pavlov Is on Your Shoulder? Two photos of Pavlov, side by side, one with a halo and one with devil's horns.

The trainer Bob Bailey is often quoted as saying that when one is training an animal, “Pavlov is on your shoulder.” He is reminding us that while we are training operant behaviors (sit, down, fetch, weave), there are also respondent behaviors and respondent conditioning occurring. Respondent behaviors are behaviors that are generally involuntary and include both outwardly visible reflexes and internal surges of hormones.

But there’s another part that is not quoted as often. Bob Bailey also says that while Pavlov is on one of your shoulders, Skinner is on the other. (B.F. Skinner was known for exploring operant conditioning.) In his Fundamentals of Animal Training DVD, Bailey states that both fellows are always on your shoulders. Depending on what you are doing, one may shrink while the other grows in importance. (Great visual, eh?)

When we are training operantly, there are still classical associations going on. And when we are doing classical conditioning, there are operant behaviors that come along and get reinforced.

I find these mixtures fascinating and have some posts and video examples.

• In this post, I describe and show how I was using Skinner for operant behaviors but Pavlov was on my shoulder. You can see the results of good associations (food, play) with agility behaviors. I was teaching operant behaviors, but many aspects of the training (including me) got a positive dose of classical conditioning.

• In this post and video, I discuss how I performed Pavlovian conditioning. But Skinner came tagging along close behind. You can see the results of classically conditioning my dog Clara to respond favorably to another dog barking. Yes, she even drools. We can assume that her body is getting ready to ingest food. But we also see tail wagging, orientation to me (food lady), and in the end, running to me when she hears barking. Those are all operant behaviors. She was performing operant behaviors that reflected her emotional state and expedited the food delivery, and those behaviors got reinforced.

What About Fear Conditioning?

Wait, what?

skinner box with rat inside. The box has a food dispenser, a floor with an electrified grid, an audio speaker, and lights that can blink
Skinner box. Note electrified grid in the floor.

We often use classical conditioning procedures with the goal of alleviating fear. We do this by classically conditioning an appetitive response, and, via desensitization, slip it in place of the fear response. But the term “fear conditioning” means something else in the literature. It refers to using classical pairing to create a fear response to a formerly neutral stimulus. Scientists including Pavlov did this. Such experiments have been performed for decades. Animals learned that a buzzer, light, or other stimulus predicted a shock to the floor of their cage or to some apparatus. They started showing fearful behaviors and a general suppression of behavior when exposed to the predictive (conditioned) stimulus.

carolina wren closeup
Carolina wren: I am little and very noisy

These pairings happen in life all the time. Some years back, a pair of Carolina wrens were nesting in a rubber bucket on my back porch, but I didn’t know it. The bucket was up on a table and one day I went to put it away. I grabbed it and a bird flew out noisily right into my face. It scared the crap out of me. Even though the startle wasn’t painful, nor was my life in danger, that’s the way I reacted. And I got leery of that bucket. I avoided moving it again until six months later in the dead of winter. Even then I touched it only after getting on a ladder to peer into it from a distance. I remained wary of the bucket for a year or two after as well. The bucket was not directly responsible for my scare. It was just a bucket. But I couldn’t shake the anxiety that got attached to it for quite a while.

Not all associations are dramatic or come from trauma. Repeated unpleasant experiences can become associated with the place they happened or the sight of the person who caused them. How about when that person who bugs you comments on Facebook? All you have to see is her name—not the content of the current remarks—to get a sinking feeling.

I’ve written another post about a stimulus that changed from appetitive to aversive to dismally neutral.

Back to dog training. People who train using aversive stimuli also have Pavlov on their shoulders. But trainers who use shock and prong collars, molding and body pressure, or who throw things are not generally accompanied by Nice Pavlov*. Not the one who floods the animal’s body with “let’s eat!” or “let’s play!” chemicals and causes pleasant anticipation. They get Bad Pavlov, the one associated with fear conditioning. You can see some examples of that in my Shut Down Dogs post.

There’s nuance in reading the body language, of course. If an animal has enough training to know how to control the aversive stimulus with its behavior, it won’t necessarily look miserable. Also, you can see happy body language on an animal being trained with aversives if the activity itself is fun. Hence the dogs who are said to “get excited” when the prong collar comes out—because it predicts a walk. Disassociate the walk and you will find that the prong collar—surprise—is not intrinsically fun. In a mixed case like this, though, I envision the aversive control as a heavy weight that always has the potential to pull the dog’s emotions in a negative direction or to suppress its behavior. The dog’s happiness is despite the aversive stimulus, not because of it.

Nice Pavlov Is Not Automatic

Just as the presence of aversives doesn’t always squelch all of the pleasure out of a situation, the presence of appetitive stimuli doesn’t guarantee rainbows and unicorns. We tend to assume that if you train with positive reinforcement, you automatically get a positive conditioned emotional response.  But it ain’t automatic. There are ways to mess it up.

Bad Pavlov was on my shoulder
See below for the reason Bad Pavlov is hanging around poor Clara.

If you are generous with reinforcement, minimize extinction frustration, and don’t frustrate or scare your dog—Nice Pavlov is probably on your shoulder. Your dog will build good associations to the training experience and to you.  But what if you frustrate or scare your dog? What if you repeatedly get in your dog’s space and don’t notice that she doesn’t like the pressure? Is Nice Pavlov going to show up and save the day just because you are using food? Probably not. You are not going to get a sweet, positive conditioned emotional response to your cues, to your training sessions, or to you if you are regularly letting aversive events creep into your training.

Here’s what it can look like if our own training session—with food—is less than fun for the dog.  I’ve set the link to start the video in the middle where training of my older dog, Summer, begins. It was a moderately stressful session for her, with too many competing, slightly scary stimuli from the environment. I was also asking for stationary “leave-it” behaviors that need a lot of self-control, and I was using kibble, a low-value food. Not a great combination.

Summer was a trouper, worked hard, and although she was clearly anxious through much of the session, succeeded at what I asked of her. I wouldn’t say this session damaged our relationship or tainted training in general—we had much too strong a history of happy training. But consider this: what if every training session I ever had with her was like that? Mildly scary stuff, hard tasks, low-value treats. Even though it would still be “R+ training” I don’t think I would have built up much of a positive response to training. Nice Pavlov would not have joined us—or if he had, he would have been smaller than Bad Pavlov. (Too many dudes on our shoulders!)

The photo above shows Clara on a day in 2015 after she knew I was getting ready to trim her nails. I have always used tons of good food, and Clara had been comfortable with nail trims for quite a while. But nice Pavlov was not in sight. Clara was recovering from Rocky Mountain Spotted Fever. While she had been sick, her joints had become painful, and having her feet handled and nails trimmed hurt. (I should not have trimmed them.)

Are Pavlov and Skinner All Jumbled Up?

Yes and no. Operant and respondent behaviors are both going on all the time. But I’ve also seen some trainers use it as an excuse. They say that since Pavlov and Skinner are both present, we shouldn’t get “hung up” on which one is primary. Well sure, we can’t crawl inside the animal’s body to check. But our training procedure should reflect our goal(s). And we should definitely get hung up on whether the dog is enjoying herself.

Teach Your Children Puppies Well

Positive reinforcement-based training done even moderately well comes with lovely side effects for the trainee (and also the trainer).  But badly attempted R+ training that regularly lets aversives, coercion, or too much difficulty in the picture can stress out dogs.  If I grab my dog’s collar to move her and I haven’t conditioned her to collar grabbing, she may dodge when I grab the next time. That was aversive. If I push into her space to get her to move and she starts jumping back when I approach her, that was aversive. If I play a game with a fearful dog where they need to come closer and closer to me to get the treat and I go too fast—I am the aversive.

So take this friendly reminder from someone who has made plenty of mistakes. Yes, Pavlov is always on your shoulder. But even if we use food to train, it doesn’t mean we will automatically get a beautiful positive conditioned response to training. There are other stimuli that can creep into our training sessions that can knock the fun right out of it.

*Pavlov wasn’t actually what we would consider “nice” to dogs in this day and age, although he was likely better than many experimentalists. I’m using some rhetorical license here.

Copyright 2018 Eileen Anderson

Pavlov photo credit Wikimedia Commons. Additions in color are mine.
Skinner Box diagram credit Wikimedia Commons.
Carolina Wren photo credit Wikimedia Commons.
Photo of Clara credit Eileen Anderson

Higher-Order Conditioning: Did it Happen To My Dog?

Higher-Order Conditioning: Did it Happen To My Dog?

tan dog with black muzzle and smaller black dog stare at the camera expectantly
Stares of expectation from Clara and Zani. Note Clara’s tail wag.

The other day I was sitting in my bedroom with Clara and Zani and the doorbell rang. And there was dead silence. Continue reading “Higher-Order Conditioning: Did it Happen To My Dog?”

Copyright 2021 Eileen Anderson All Rights Reserved By accessing this site you agree to the Terms of Service.
Terms of Service: You may view and link to this content. You may share it by posting the URL. Scraping and/or copying and pasting content from this site on other sites or publications without written permission is forbidden.