Your Dog The Gambler: The Magic Of Variable Rate Reinforcement

A very common concern or problem with reward-based training is that the dog only performs when he knows you have the reward. The secret to success is to understand the difference between a lure and a reward, and work on giving your dog a gambling addiction.

Variable rate reinforcement is powerful stuff, just look at the money casinos, lotteries, and other types of gambling bring in. They dole out small prizes to string people along, and the lure of a jackpot is always there to encourage people to keep on playing. People know logically that they ultimately lose more money than they get out of it, but they keep playing. Why? Because when properly motivated, our brains will encourage us to work harder and longer for a random reward. This is just as true of dogs as it is of people, and understanding this bit of dog psychology is a huge advantage when training.

In behavior studies, variable rate reinforcement, meaning that rewards are given at  random intervals or are of variable value, produce the most consistent performance. Think about it, if you sometimes reward your toddler with a candy bar in the grocery store when she cries, what will she do when you don’t? Probably cry harder, louder and longer. The same is true of adults. What do you do if the soda machine fails to dispense your soda? You push the button harder, and more times. If something works sometimes, both people and dogs are wired to do it more to make it work again.

To apply this to dog training, you must first understand the difference between luring your dog to behave, and rewarding him after he’s behaved. When you’re first training a behavior,  it’s perfectly acceptable to use food as a lure to encourage your dog to perform the desired behavior, but once he consistently shows that he understands the cue, it’s time to stop luring and start rewarding randomly. The best way to do this is to set yourself up so that you can produce a treat without a lot of fanfare. You can wear a treat pouch all the time, so that your dog begins to ignore it. You can have bowls of treats available around the house, positioned so that you can get a treat without a bunch of noise or trouble that will alert your dog that you have the reward, or you can hide treats for your dog to find when he does the right thing. If your dog absolutely knows that you have food, and that he will get it the second he does X, it’s a lure. We’re aiming for a slightly surprising reward.

This is where a clicker can come in handy, because it’s an easily and precisely timed cue to your dog that, yes, he did do the correct thing. It’s not a reward in itself, but because it’s very frequently paired with a reward, some of the same happy things will go off in your dog’s brain. It’s a way to direct your dog’s attention to you, and not when your hand is near the treat bag. Dogs are visual learners, so it’s easy for them to pick up on subtle physical cues, and some people find that having the clicker as an intermediary between their hand and the treat bag prevents them from having an obvious tell.

To get your dog addicted to gambling, you’ll vary both the timing and the quality of his rewards, just like a casino. They should be consistently high value enough to motivate him in a given situation, and may require a temporary use of lures in high stress environments. For instance, the vet’s office deserves a much stronger motivator than the living room, because it’s a much more stressful and distracting environment. It’s not a reward if your dog is too stressed or distracted to enjoy it. At the vet, that probably means a high rate of reinforcement with a moist, smelly, flavorful treat, rather than a bit of kibble or a dry biscuit. You can vary the rewards for the length of time your dog stays in a position, the number of times he’s performed a task, and the difficulty of the environment. Many treat pouches have two or more compartments for different types of treats, so you can vary what you’re giving in a given training session.

It’s also good to “Jackpot!” your dog occasionally. An example given by trainer, Kathy Sdao, is to hide a hamburger, or some other wonderful and rare treat that’s appropriate for your dog, along the route you’ll take for an off-leash walk. Ask your dog to come, and make it an easy recall, then “magically” produce the hamburger as a reward. The hamburger was hidden, and he couldn’t smell it and anticipate that you had it, so to your dog, it seems as though you could give him a HUGE reward for a relatively easy command at any time. It makes it much more rewarding overall to respond promptly to commands, and you can use fewer and lower value treats to maintain the response. It also makes it harder for your dog to decide that the reward he’ll get for responding is less rewarding than what he’s already doing. He doesn’t know what he might win, so it’s probably worth it to check. No one would play the lottery if the prizes maxed out at $5, but because the lure of the big prize is out there, people continue to play with very few rewards. It’s the same concept.

Jackpotting can also be accomplished with many pieces of a small treat, and many dogs actually find this more rewarding. It’s the difference between a pile of Hershey’s Kisses and an equal weight of bar chocolate. The Kisses will be perceived to be as more, because they are more numerous, and take longer to eat.

The more times your dog responds to your commands, the more worn the mental pathways for that response become, and the more likely he is to respond correctly under stress. Variable rate reinforcement helps intensify this effect, and keeps your dog from prefacing his response with an evaluation of what he’s going to get out of it. To a dog that’s accustomed to being lured into a behavior, checking that you have the treat is part of his response. A dog that’s been rewarded randomly just performs and then hopes he’s a winner. The behavior rapidly becomes more automatic and consistent.

I hope this dispels the myth that dogs trained with reward-based techniques can’t be reliable without the treats on hand. All it takes is a little preparation and savvy, and you’ll have your dog gambling in no time.


By Laura-C

Hopes to someday train her dogs not to be douchebags.

2 replies on “Your Dog The Gambler: The Magic Of Variable Rate Reinforcement”

I just adopted a puppy, and I’ve found this is already working really well in just one day. I’ve gotten him to sit and heed me by once in a while give him a small treat but most of the time just having him do it and then rewarding him with pets, haha. It’s working very well.

Leave a Reply