This week’s blog is an excerpt from

“Dog Smart: Evidence-based Training with The Science Dog.”

---

Even those who are not Pink Floyd fans understand this principle.

“You cannot have dessert until you first eat your vegetables”

This common admonition means that you can only have the thing that you really want if you first complete a task that you have are far less motivated to do.

Parents everywhere use this technique. By rewarding vegetable-eating with dessert, mothers attempt to make Johnny’s vegetable-eating behavior more likely to happen (or so the theory goes). Behaviorists refer to this as using a high probability behavior to reinforce a lower probability behavior – aka The Premack Principle.

Dessert reinforces eating vegetables; going swimming reinforces getting the yard mowed; drinking a glass of wine reinforces finishing the vacuuming. You get the picture.

---

The Premack Principle is not only popular with parents; a lot of dog trainers also seem to like it. Often referred to as “life rewards”, this principle is in effect when trainers use the opportunity to greet a friendly person (something the dog really wants to do) as a positive reinforcer for approaching slowly or sitting prior to greeting (something that the dog is not so keen on). Another popular example that we see in the literature (but I have yet to see it work well in real life) is attempting to teach a dog that walking on a loose lead  will be (eventually) rewarded by an opportunity to sniff a bush at the end of the block. In these two examples, sitting to greet and loose-lead walking are the vegetables. Greeting and getting to sniff are the dessert.

Sounds great, right?

Sure. Except that it generally does not work very well with dogs. (Nor with kids, I would argue). Just as with extinction, attempting to reward something that a dog finds difficult or is not motivated to do with something that he wants to do very much, can cause unwanted emotional fall-out. Here’s why:

Problems with Premack

The problem is that Premack ignores the emotional components of the behaviors it is meant to influence. Let’s return to little Johnny and his vegetables for a moment. Imagine Johnny, finishing the green beans as instructed, only so that he can get his pudding. Do we expect this approach to increase Johnny’s enjoyment of vegetable eating? Probably not. This is because vegetable eating is turned into a means to an end when Premack is in play. There are three possible outcomes when Johnny (and dogs) are told

Do something that you rather dislike in order to get a reward that you really like

1. Rushing through: Johnny rapidly shoves the remaining green beans into his mouth, choking them down as quickly as possible in order to get to his dessert.

• Stonewalling: Johnny folds his arms and refuses to eat the dreaded green beans. He reasons (correctly) that at some point his mom will capitulate and allow him at least to leave the table (which he would like) if not actually receive dessert.

• Scamming the system: Johnny hides the vegetables under his plate or, if he is lucky, slips them to his vegetable-loving dog who is waiting under the table.

---

Of course a fourth possibility is that the promise of future dessert causes Johnny to suddenly take a new look at those green beans and decide that he really does love vegetables! Perhaps, but highly unlikely because the Premack Principle does absolutely nothing to increase the value of the behavior that is of interest – in this case eating and enjoying one’s vegetables.

---

And so it goes also with dogs.

In many of the situations in which a low probability behavior (sitting to greet, loose-lead walking) is presumably being reinforced by a high probability behavior (greeting or sniffing), the dog does exactly what Johnny did. She will attempt to rush through – walking rapidly and frenetically (perhaps even throwing in a vertical leap and frustrated whines) while still (trying to) maintain a loose lead. She may stonewall by avoiding the target behavior altogether (lying down), or she may try to scam the system by offering another (equally unwanted) behavior (barking in excitement). Regardless of the dog’s response, the emotional outcome is not going to be a dog who enjoys sitting to greet or walking on a loose lead precisely because Premack turns these behaviors into a means to an end and does absolutely nothing to make the behavior itself enjoyable (reinforcing) for the dog.

An Alternative to Premack

An alternative approach is to increase the value of the behavior you wish to influence. In Johnny’s case, perhaps his mom could prepare green beans in a way that is more palatable for little kids, or make a game out of eating each bean, or have Dad sit at the table with Johnny and make a show of how much he loves to eat vegetables.

For dogs, it is even simpler –beef up the behavior you want. Positively reinforce desired behaviors as they are occurring and use differential reinforcement as the dog becomes proficient. Loose-lead walking, sitting for greeting and many other “low probability” behaviors can be turned into high probability behaviors by reinforcing these behaviors directly. The use of high-value treats that are available to the dog only during walks can be used to reinforce a loose-lead for a few steps at a time. Having a greeting person approach with hands lowered and offering lots of yummy treats for a quick sit prior to greeting can help to shift “sitting for greeting” towards a high probability behavior. A quick fix? Of course not – loose-lead walking and sitting for greeting are two really difficult behaviors for excited and friendly dogs to accomplish. However, an approach that avoids the emotional fall-out of frustration and that attempts to increase pleasure in the targeted response will always be first on my list of go-to training techniques.

By not turning desired behaviors into means to ends, we can increase a dog’s enjoyment of the targeted behaviors and reduce the frustration that is associated with making dogs “eat their vegetables in order to get their dessert”.

---

From: “Dog Smart: Evidence-Based Training with The Science Dog” by Linda P. Case

---