The Cheesecake Challenge, or an Experiment in Resisting Negative Reinforcement

One thing that training dogs has taught me is to better understand certain behavioral patterns in myself. Here’s an example.

I dislike unsolicited advice. I try to refrain from giving it to others, and it annoys me when it is bestowed on me by friends who (I feel) should know better.

And yet, a lot of the time, my response to unsolicited advice is to reinforce the unsolicited advisor by thanking them for their help. As a result, the friend is likely to give me more unsolicited advice in the future. The spiral of miscommunication continues.

Why on earth would I thank them for something I didn’t want? I recognized what I was doing, and I was puzzled by it – until I observed a completely unrelated behavioral pattern in dogs.

Learning to tug

Some dogs show little natural interest in playing with a tug toy. A common method of teaching them to tug is to excitedly tease and slap them with the toy and push the toy into the dog. Lots of dogs learn to tug this way.

Throughout the training process, a handler who uses this method is getting positively reinforced for teasing and slapping the dog with the toy: the dog grabbing and pulling on the toy is equivalent to me saying thank you to an unsolicited advisor. It’s a powerful positive reinforcer for the handler or the unwelcome advisor, respectively.

But what is happening from the perspective of the dog? Pause for a second before you continue reading past this paragraph. Is the dog’s tugging behavior strengthened and maintained by …

(A) positive reinforcement,
(B) positive punishment,
(C) negative reinforcement,
or (D) negative punishment?

Scroll down to see if you were right!

tugging, dog training, dog trainer, play

The answer is (C): negative reinforcement. The dog in our example doesn’t naturally care about tug toys. The trainer slapping and teasing him with the toy, trying to put it in his mouth, slapping some more while talking excitedly is irritating, confusing, intimidating or annoying to the dog. In an effort to escape the trainer’s behavior, he will sooner or later try grabbing the toy.

That very instant, the slapping, teasing and pushing stops. Wow! A moment of relief! The dog just learned that he can stop his owner’s craziness by grabbing and holding on to the toy. The better he understands his ability to influence the handler’s behavior, the sooner will he grab the toy. Eventually, he’ll start tugging as soon as the toy is presented. Voilà – the dog has learned to tug via a negative reinforcement procedure.

I assume many dogs eventually start to enjoy tugging as such, even if they initially learned by means of negative reinforcement. Some dogs, however, may never truly enjoy to play tug, and keep doing it mostly in order to not get slapped and teased with the toy.

Me? I’m like the latter kind of dog when it comes to thanking people for their unsolicited advice. I keep doing it to escape the pressure of being expected to respond.

Unsolicited advice

Being showered with unsolicited advice from friends – whether it is dog training advice, dating advice, or life advice – feels like being slapped with a tug toy. It’s irritating. It makes me feel yucky and disrespected. If that were the only force at play here, I’d probably actively reject the advice (applying positive punishment), or ignore it (applying extinction).

The pressure to respond and the crux of kindness

However, two other strong forces come into play as well:

  • My need to respond to my friends’ messages (the equivalent of “will to please” in a dog),
  • and the belief that kindness is more important than honesty, i.e. we should not do things that may hurt our friends’ feelings (the equivalent to living by a certain rule structure we sometimes observe in herding breeds).

When a friend writes me a message, the ball is in my court, and I feel like I NEED to give the ball back to them. As long as the ball is in my court, it’s as if the slapping and teasing with the tug toy continued. Having a friend’s ball in my court is pressure – that’s why I’m having a really hard time just ignoring unsolicited advice.

So I will eventually respond – and I’ll be nice about it. I feel like I need to respond to my friend’s good intentions rather than to the effect their unsolicited advice-giving behavior has on me.

I’ll probably at least say “Thank you” or a send a smiley face or a thumbs up emoticon. While this is the weakest reinforcer available to me, it is still strong enough to maintain my friends’ unsolicited advice-giving.

The Challenge

Now that I’m aware of what has been maintaining my grateful responses to unsolicited advice, I’m going to conduct an experiment: I’ll try to change my own reaction, and find out if I can reduce unsolicited advice giving that way. At the very least, I expect to feel better when learning to respond in a way that is in line with my feelings about unsolicited advice. (The training parallel here would be giving the dog a way to opt out of playing tug.)

I’m going to start with focusing on advice given in written form, online. My alternative response to unsolicited advice will be to close the chat window instead of replying. I’m going to use a food reward for myself: I will treat myself to the world’s best cheesecake, which is sold at a local café, contingent on closing the chat window. I’ve only had this cheesecake twice because it’s fancy and expensive cheesecake, but I think about it every time I walk past the café. Starting today, for the next four weeks, I’ll have cheese cake every time I close a chat window on an unsolicited advisor. I’ll see how it goes!

How about you? Has dog training changed your understanding of your own behavioral patterns or the behavioral patterns of the humans around you? Would you like to design your own version of the cheesecake challenge? Share your thoughts in the comments!

Chrissi travels internationally learning about dogs, and makes money to support her roaming by teaching online at FDSA, in person in Guatemala, and seminars around the world. Contact Chrissi for more information.

Finding 5(0) with Kathy

I recently finished the German translation of Kathy Sdao’s book Plenty In Life is Free. It’s a great little book that makes a strong case for treating animals with kindness. One of my favorite parts is Kathy’s SMART x 50 protocol, which is presented as an alternative to NILIF (training based on the assumption that Nothing In Life Is Free for dogs). SMART stands for “See, mark and reward training.” Kathy advises her clients to count out 50 treats every morning, and use them to see, mark, and reward desired behavior throughout the day. The goal? Having an empty treat bowl in the end of the day. 50 sounds like a lot, but once you get in the habit of catching your dog being good, your treats will go like hotcakes.

I recommended that protocol to a number of clients when I first read Kathy’s book, but then somehow forgot about it. Now that I was translating Plenty in Life is Free, I decided to implement SMART for the next week with my own dogs, particularly Game – an adolescent whose house manners were a bit all over the place! For 7 days, I’d take one cup of kibble out of her daily food portion and put it on the kitchen counter every morning. Throughout the day, I fed Game any time I caught her being good, and if anything was left at night, she got it with her dinner.

Within a week of paying attention to catching Game being good, I had her mat manners back: anytime she chose to hang out on her mat and I happened to see it, a treat would materialize between her front paws. Anytime she stayed in the yard without immediately scratching the door when I went inside – a new habit she had developed recently – I’d treat her. She didn’t need to do anything specific – any behavior other than scratching the door or whining got reinforced. The first two days, I mostly fed her standing in front of the door expectantly. On day 3, she chose to lie down, and that’s what I could capture. My criterion was simple and ensured I had countless opportunities to reinforce Game: if I don’t dislike it, I’ll treat it!

In only one week, I had Game’s good house manners back. The best part: I didn’t lose any time working on these things – the training just happened while I was going about my day. All I had to do was remember to prepare a cup of kibble in the morning, and get into the habit of seeing, marking, and rewarding behaviors I liked.

I challenge you to give Kathy’s SMART x 50 protocol a try yourself! Start today, and stick to it for a week. It’s ideal for behaviors you are too lazy to work on in designated training sessions – things that seem boring rather than fun to you as a trainer. Leave me a comment to let me know how it’s going, and what you choose to reward!

PS: I’m getting ready to teach one of my favorite classes at Fenzi Dog Sports Academy. Finding Five – Training for a Busy World is starting on December 1st. It is an entire class about fitting a little bit of training time into a busy schedule, making use of games similar to Kathy’s SMART x 50. It’s also a class about reviving your relationship with your dog: if the reason you haven’t been training or spending time with her is not a lack of time, but a lack of motivation, you’ve come to the right place as well. Registration is open, and I’d love to see you in class!

Case Study: Toni learns to relax around visitors

Meet the dog

Toni is a very big, black-and-tan, floppy-eared, 3ish years old male mixed breed who was adopted by Sabrina when he was about 5 months old. With the exceptions of the two issues described below, Toni is a laid-back and mostly low energy dog.

Sabrina has four housemates, and they all have active social lives – a lot of the time, there isn’t just five people at their house, but rather ten. It’s never boring, and it’s never quiet. Toni does well in this environment – he’s a good fit for a social owner. Soon after Sabine got him, he started greeting each and every one of their visitors like old friends, even when they were new ones.

The behaviors we wanted to change

1) Toni used to get very excited when Sabrina had visitors, would try and jump on their lap as they sat on the couch, solicit attention, scratch their legs, and whine. It wasn’t easy to have a conversation with Toni in the room. Sabrina wanted to change his behavior around visitors. Taking him places wasn’t easy, either, because he insisted on being the center of attention when Sabrina was out with friends.

2) Out in the street, Toni would bark and lunge at other dogs. Sabrina wanted him to learn to pass them calmly.

Training steps: Learning to relax around human friends

Toni’s life included a lot of different people coming and going. He had the right kind of personality for it – he liked people. However, he liked them so much that he wanted to interact with them, and he had learned that the best way to do so was to pester them until they gave him attention! Sabrina had a lot of dog-loving friends, so this had been working well for him.

We decided to teach Toni to station on a blanket. This particular blanket would only come out when Sabrina wanted him to stay on it, and the blanket itself would become the cue to lie down.

Our first challenge was that Toni wasn’t interested in food rewards. He was free fed. Sabrina had a 15-kilo bag of high-value kibble in a corner of her bedroom. The bag was always open, and Toni just walked over and ate when he was hungry. He had been free fed ever since Sabrina got him. He was a little chubby, but he didn’t over-eat. He was very relaxed around food in general. Food wasn’t a limited resource. This was convenient in everyday life, but presented a training challenge!

In order to increase Toni’s interest in earning his food, Sabrina stopped free-feeding him. The first thing Toni needed to learn was that food could be the consequence of something he did – a concept he wasn’t familiar with. However, he knew how to sit. This was our starting point. Sabrina asked him to sit, clicked, and fed a cookie. She threw the next cookie to make him get up again, asked for another sit, clicked, and threw a cookie. After looking slightly perplexed in the first few sessions, Toni decided that this strange new game was fun. Now that food wasn’t available for free anymore, Toni’s interest in it had increased considerably. He liked interacting with people anyways, and these cookies weren’t all that bad, either! You could see him perk up as he realized that he had the power to make clicks happen and food appear.

With the help of a cookie pressed against Sabrina’s hand with her thumb, he soon learned to do a hand-touch as well, which earned him a click and released the cookie. Sabrina could fade the lure within a few reps. Toni learned to figure out how to get his cookie: sitting, hand touches, or “shake”: he needed to paw at a closed fist in order to get his treat!

Next, I showed Sabrina how to add an element of shaping to her training sessions. Toni was going to learn to go to the blanket we were later going to use to change his behavior around visitors. I asked Sabrina to get a new blanket Toni had never seen before. She made a big fuss about it, then put it on the floor. Toni came over to investigate – click! Sabrina threw the cookie away from the blanket. Toni chased down his treat, and since he hadn’t been done investigating the blanket just yet, he returned to give it another sniff – click! In the course of several short sessions, Toni learned to step on the blanket with all four paws. Now, Sabrina clicked him for standing on the blanket, and then lured him into a down with the reward cookie. She waited a second or two, clicked again, and threw a cookie off the blanket. After a few reps, Toni offered his first voluntary down on the blanket and got a jackpot. After every brief session, Sabrina removed the blanket. It was only out when she was working with it.

Once Toni had learned to lie down on the blanket as soon as it was presented, we put the blanket where Sabrina eventually wanted it to be when she had visitors: in one of the corners of her big couch. It was important to her that Toni could be a part of her social life. She didn’t want him to have to wait in a different room, in a crate, or in a corner. During training sessions on the couch, she could sit next to him, feed him cookies and read or work on her laptop at the same time. She could also scratch his ears while he relaxed next to her, which he loved.

Once Toni recognized the appearance of the blanket on the couch as his cue to lie down, we started adding duration. From this point onwards, we made sure that Toni wouldn’t have Sabrina’s undivided attention. She’d read a sentence – feed a cookie. Read two sentences – feed a cookie. Read three sentences – feed a cookie, and so on. As long as he continued hanging out on his blanket next to her, cookies would materialize. At the same time, we made sure her full attention and eye contact weren’t part of the picture we were creating. After all, we wanted Toni to eventually relax rather than “work,” and Sabrina wanted to be able to focus on her visitors, and not just on her dog.

Toni was good about relaxing for the occasional cookie, and daily sessions got Sabrina to a point where she could soon read several pages of a book between the individual cookies, and occasionally replace a cookie with ear scratches. We systematically introduced Sabrina getting up, walking around the room, and sitting down again while Toni remained in his spot. He also learned to stay when Sabrina got up, left her room, and then came right back in. We practiced this until Sabrina could get up, go to the kitchen, get a glass of water, and return without Toni getting up or getting fidgety.

The next step was practicing with various visitors. The first one was me: Toni learned that the blanket game could still be played when someone else was in the room. We first increased the rate of reinforcement again, and since Toni’s desire in this situation was to interact with the visitor, we decided that I – the visitor – would give him the occasional cookie and attention when he was on the mat. My attention made the reward even more reinforcing.

It turned out that Toni was actually able to be quite patient and well mannered now that he knew hanging out on his blanket would get him cookies and attention. His excitement hadn’t been due to high arousal and overflowing energy – he had simply learned that he had to pester people in order to get attention. Once provided with an alternative behavior, he turned out to be an easygoing big boy.

After some experimenting, we decided that the mat would come out right before a visitor came into Sabrina’s room. If he stayed on his mat, the visitor would come over and great him with a cookie right away. If he got up, the visitor would turn around and close the door behind them. Sabrina would pick up the blanket, wait a second, and then put it down again. This usually reminded Toni to lie down. Now the visitor could come in and approach again.

Sabrina then began to ask other helpers to visit her in order to train her dog. First, we worked with two of her dog-savvy housemates. Then, she would ask friends to help her. If Tony stayed on his blanket, Sabrina would instruct her friends to deliver a cookie to him and calmly talk to him.

While building this new behavior, Sabrina had people over specifically for this exercise, not in order to socialize or talk about other things. She was consistent in her training, and it showed in Toni’s progress. He learned to stay on his mat while visitors came in the room, and his overall level of excitement around human friends decreased.

Sabrina then switched from cookies to long-lasting chews and stuffed Kongs that Toni could use to entertain himself on the blanket when she had people over. At that point, she was able to actually focus on her visitors, too, and not just on training Toni. By the time Toni had gotten used to eating part of his dinner from a Kong when Sabrina had people over, she was able to start giving him more freedom again. At first, she had made sure people would leave before Toni finished his Kong. Then we tested what would happen if Toni got to finish his Kong before the visitor left: it turned out he soon dozed off while Sabirna and me were still sitting together. We then tried what would happen if Sabrina released Toni and took away the blanket after he had finished his Kong – and he would just trot over to his bed and continue dozing off there. The Kong seemed to have a calming effect, and Toni’s need to be the center of attention had disappeared now that his relaxation on the blanket got reinforced on a regular basis. As Sabrina gradually increased his post-chew freedom, he would sometimes go right to his bed, and other times, he’d jump off the couch, wag and wait for ear scratches from Sabrina or the visitor before heading over to his bed. The attention-seeking behavior and vocalization had disappeared completely.

Training steps: Learning to relax in public

Sabrina wanted to be able to take Toni more places. In order for him to relax around friends away from home, Toni needed to generalize his blanket skills. Most cafés and restaurants in Austria allow dogs, and it is pretty normal that people bring their dogs when they go out for lunch or dinner. In order to practice for this, Sabrina and I went to McDonalds. Fast food restaurants are perfect for this: you can just get up and leave anytime, and it’s perfectly fine to only spend a few minutes inside. We picked a table in a quiet corner. Sabrina would head over to the table, put down the mat, and calmly reinforce Toni for lying down on it, and for staying down. I would get our drinks from the counter and join them. Once Toni had settled, he got a frozen Kong or long-lasting chew. We would finish our drinks, keep an eye on Toni, and discuss the next training steps. Then, Sabrina would trade the Kong or the remains of the chew for a cookie, release Toni, pick up the blanket, and we would leave.

After going through these steps together, Sabrina was ready to practice at places like McDonald’s, Burger King, or Starbucks on her own, or in the company of dog savvy friends. If she went on her own, she made sure to set him up for success by having him wait in the car while she ordered her coke or coffee and put it on the table, and set up the blanket. Then, she got Toni from the car, lead him directly to her table, and rewarded him for recognizing his blanket and lying down on it.

Gradually, Sabrina increased the time Toni could spend at a fast food restaurant, and decreased the attention he got from her until she was able to take him to other restaurants as well and actually have conversations with her friends while he relaxed on his blanket under the table.

Why we chose this training approach

Toni wasn’t a high energy dog to begin with. That made the blanket a good choice. He wasn’t torn between finding an outlet for his energy, and staying on the mat. He just needed an acceptable way to solicit attention when people were around. Hanging out on a blanket was congruent with his base personality: a big, friendly, laid-back dog.

Toni’s excitement around people wasn’t based on anxiety or insecurity. He genuinely liked people, and wanted to meet them. Not knowing how to get their attention was frustrating to him. Interacting with them was reinforcing, not stressful. This made it possible to integrate visitors into his reinforcement protocol.

It was important to Sabrina that our training plan would allow Toni to keep being part of her social life. She wanted her friends to be his friends, too. Using a blanket on the couch achieved just that.

Check back next week for how we worked on Toni’s second issue: barking and lunging at strange dogs in the street.

Satiation vs. Deprivation: Ethics and Smart Training Choices

 

If you are like me, you probably love chocolate. If you wanted to teach me something using a high-value food reinforcer, chocolate would be the way to go. However, there are times when chocolate loses some of its reinforcing power: right after a big lunch, I’ll be full and not particularly interested in chocolate. And if I’ve already had a sundae that day, I’d rather work for something savory – maybe riffle chips (the red pepper flavored kind). In order to be an effective trainer of Chrissi, you should know these things about me.

And there is more: if you deny me access to chocolate for several days or even weeks, chocolate will be much more attractive – it’ll feel special and gain even more reinforcing power. You could also get a particularly high value kind of chocolate (Swiss chocolate truffles, for example) that I don’t have access to on a daily basis and achieve a similar reward-boosting effect.

A third way to strengthen the power of chocolate as a reinforcer is to not feed me all day, and then ask me to work for chocolate. I’ll be hungry, and when I’m hungry, I’ll crave that chocolate bar even more. (Note that if I haven’t eaten in a very long time and my blood sugar is low, my performance will suffer – no matter how much I want the chocolate, I won’t be able to concentrate well.)

Let’s say you don’t want to work with chocolate. You’d rather use a healthier reinforcer. How about seafood? It’s supposed to be good to have some on a regular basis! Well – I don’t like seafood. I’m not going to work for your seafood. Unless, that is, I’m SO hungry that I don’t care what I get as long as it’s edible. Starve me for a day, and I’ll be more willing to perform for a shrimp.

The variables you’ve manipulated in these examples are satiation and deprivation. I’m not the only one who is affected by them: the same goes for our dogs. (Note that these diagrams are in no way scientifically accurate – they are just meant to illustrate a point.)

Satiation Diagram A

Diagram A: The relationship between satiation and reward value

In Diagram A, the X axis depicts Satiation. The further to the right, the more satiated the dog. On the far right, the dog has eaten a bit too much, and now his tummy aches a bit. On the very left, the dog is starving and desperate for something – anything – to eat.

Let’s assume you’re working with a medium to low value reward. Your dog will always take it, unless he is feeling sick or has eaten way too much to care. The blue dots show that the more deprived your dog, the higher the value of the same reward. It’s a linear development.

Satiation Diagram B

Diagram B: The relationship between satiation and performance

Diagram B shows the relationship between your dog’s level of satiation and her training performance. As in the previous graph, we have satiation on the X axis.

The Y axis depicts your dog’s performance. The red dots show the relationship between the two. Unlike in the previous graph, the development of the performance is not linear. Up until a certain point, the dog’s performance increases with deprivation. However, at a certain point, it starts to decrease again. In the example of reinforcing Chrissi with chocolate, the peak of the graph would be my best performance. It would likely occur after starving me for a few hours, but not an entire day. If you starve me for too long, my blood sugar will drop too far. I won’t be able to concentrate on the task you ask of me, or perform a well-known behavior at top speed.

The Ethics of Working with Deprivation

Satiation Diagram C

Diagram C: Where do you draw the line?

Diagram C has a dotted grey line parallel to the Y axis. This line is defined by your ethics, and it’ll look a little different for every trainer, and for every reinforcer that trainer uses in her training: you’re okay working to the right of the dotted grey line – on the green line parallel to the X axis that defines the satiation level. You aren’t okay working to the left of the dotted grey line: you feel like it’s not fair to deprive your dog to this level for no other reason than to strengthen the reward value to the point of top performance.

Where do you draw the line?

Think about where YOU draw your line. What satiation level is both in line with your ethics, and gives you the highest reinforcement value (the double-headed green arrow on the X axis)? That’s your ideal training space.

Primary Reinforcers: Food and Water

When it comes to food, my line of ethics doesn’t go all the way to top performance/high satiation. I won’t deprive my dogs of their daily food in order to increase the value of their reinforcers. However, I am perfectly fine using their regular meals for training, or training right before or during their dinner or breakfast time. Naturally, they’ll be more hungry at this point than after a meal. My ideal satiation level is right before and during regular meal times. When I want to make my food reinforcers extra valuable, that’s when I’ll train.

If you train for quite a while, you may see your dog’s interest in his treats decrease as the session goes on: the fuller he is, the lower their value gets. While some dogs never lose interest in treats, others show this effect consistently. If that’s the case with your dog, make sure to keep your food training sessions short, or use high value treats!

Water is something my dogs always have access to when we’re at home. I won’t ever deprive them of it. Therefore, water isn’t a reinforcer I can use in my training – it’s a primary reinforcer (all animals need to drink), but its value tends to be low because it’s freely available. However, that doesn’t mean I can’t use it as a reinforcer when the opportunity just happens to present itself! On hot days, I’ll bring water for myself and the dogs on a hike, to the beach, or to the training field. After lots of running around or a high energy training or play session, my dogs will be thirsty, and I’ll offer them water. They are going to get the water anyways, but why not throw in a quick recall or ask them to walk up to the water bowl on a loose leash? You could also practice impulse control by asking them to hold their position while you put the bowl on the ground, and then release them to it. If my dog just happens to be thirsty, it’ll be heavily reinforcing. If he doesn’t comply for whatever reason, he’ll get his water anyways, of course. Using things your dog has free or regular access to in moments when they just happen to be stronger reinforcers than usual isn’t the same as depriving your dog of his meal or his water – it’s making a smart training choice.

Game’s recall is being reinforced by an opportunity to drink some fresh water (I don’t use a marker cue for this):

https://youtu.be/1KKKbGpLzyo

Secondary Reinforcers: Toys, and the Opportunity to Train

When it comes to toys, dogs tend to be highly motivated at the beginning of a session. However, the more often they have already chased the ball or tugged their toy, the less interesting the toy gets: your dog – unless he is a ball junkie – is becoming satiated by the play.

If you have a dog who doesn’t like to play for hours on end, the smart training choice is to keep your training sessions short and end them before the dog is satiated by your toy reinforcer. That way, you’re always training with a strong reinforcer.

Another secondary reinforcer is a more specific kind of attention: training time! While some dogs are training junkies and never get tired of working for their human (Phoebe would fall into this category), others are more easily satiated by training. It’s not that they don’t like spending time with you – it’s just that training is fun for them, but so are lots of other things like taking a walk, relaxing on the couch, playing with your other dog, or watching squirrels through the kitchen window. Their training drive is satiated easily, and then they are ready to move on to a different activity. If you have a dog like this, less can be more: rather than training 3 sessions a day, train only 1 or 2! And instead of training 7 days a week, take the weekends off! Reducing training time and keeping sessions extra short can boost the motivation of a lower-drive dog. Of course, reducing training time doesn’t mean that you can’t spend as much time with your dog as you want – it just means that instead of spending every free minute training, you’ll have a cuddle session on the couch or a nice walk instead.

What are the primary and secondary reinforcers in your dog’s life? How could you manipulate them in terms of satiation/deprivation? And where is that sweet spot of strongest possible reinforcer, best performance, and an ethical training session?

These are just some of the questions we’ll be looking at in my upcoming FDSA class, May the reinforce be with you! We’ll also talk about marker words and reward placement, how to select reinforcers based on the emotional state you are looking for, and how sometimes, small changes to your reinforcement protocol can have a big impact on your dog’s precision and enthusiasm. Join me in class if that’s your kind of geeky!