AKC is a participant in affiliate advertising programs designed to provide a means for sites to earn advertising fees by advertising and linking to akc.org. If you purchase a product through this article, we may receive a portion of the sale.
Training your dog with rewards is a great way to teach and maintain behavior. But how often should you give those rewards? You can give your dog a treat every single time they do a behavior right, or you can toss a treat every once in a while. Which is better? Understanding what’s known as “schedules of reinforcement,” and the science behind them, can help you to reward your dog most effectively when training.
What’s a ‘Schedule of Reinforcement?’
Positive reinforcement dog training uses rewards to reinforce or encourage the behavior you want your dog to perform. Positive reinforcement applies to people, too. Most people are more likely show up to work because they are compensated in some way. If payment stopped, most people wouldn’t continue to do their work. But is a paycheck the best motivator? Or do bonuses drive efforts? Some people might prefer to be paid by the task, not by the week.
In a similar way, dogs are more likely to repeat behavior that is rewarded. A reward could be a training treat or the toss of a squeaky ball—whatever your dog finds exciting. There are many ways to provide rewards, and these are known as schedules of reinforcement. That’s a technical way of describing different rules for how payment or rewards are dispensed. Let’s look at the different schedules and how they apply to your dog.
Continuous Schedule of Reinforcement
A continuous schedule of reinforcement (CR) rewards every instance of a behavior. For example, you might give your child dessert every time they eat their broccoli. With your dog, you could give them a biscuit every time they sit. This is the perfect way to teach new behaviors, because it makes the association between the behavior and the reward really clear. If you don’t reward them every time they sit, your dog might struggle to understand what you expect and when they are performing correctly.
Think about your water taps: they provide CR. When you turn them on, refreshing water comes out. However, if you turn on your tap and nothing happens, you won’t keep trying and trying to turn it on. Instead, you’ll give up and call the plumber. This isn’t the most practical schedule, though. You might not always have a treat on hand, and eating too many treats isn’t healthy for your pet.
Once your dog understands what you want, it’s time to stop CR. Otherwise, they might only work for you if they see a treat in your hand. CR is also helpful when a dog completes complex behaviors, such as performing a trick (like dropping a ball through a hoop).
In this case, CR helps to maintain the entire “behavior chain.” A behavior chain is a combination of two or more behaviors performed in a sequence, where each behavior serves as a cue for the next and the reinforcer comes after the last step in the chain. For example, retrieving a dumbbell involves picking up the dumbbell, bringing it back, then placing it in the handler’s hand. Or performing a freestyle dance move might involve walking backwards, then spinning in a circle before jumping into the handler’s arms. Each behavior is originally taught and reinforced individually before they are strung together into a chain.
Extinction, or No Schedule of Reinforcement
We don’t recommend to stop rewarding your dog when they’ve learned a behavior. Remember a water tap that doesn’t dispense anything when you turn it on. Your tap-turning behavior stopped when the reward stopped flowing. That’s known as extinction. And when your dog stops getting rewards, eventually their responses will dry up too. That can be helpful when dealing with problem behaviors like jumping up or begging. But when reinforcement and encouraging certain behaviors is your goal, you want to ensure your dog stays engaged and willing to respond. That’s where intermittent rewards come into play.
Intermittent Schedules of Reinforcement
If you don’t reward your dog for every response you receive, you’re using an intermittent or partial schedule of reinforcement. Dogs learn more slowly with this type of reward system, but once they have learned a given behavior, it will be less susceptible to extinction. There are four categories of intermittent reinforcement.
Fixed Ratio
A fixed ratio schedule (FR) provides a reward after a fixed number of responses. Think about a production line. Workers might get paid for every 20 gadgets they build. Completing the assembly of the first 19 gadgets earns them nothing, but number 20 brings payday. With your dog, you might play tug-of-war after every fifth sit and do nothing for sits one through four.
FR is easy to implement, but your dog will quickly learn the routine. They will respond best right before the reward is due, while their performance will drop off right after. So FR might not be the best way to work on improving your dog’s behavior. Consider that production line. The workers will work fast to build those gadgets, but in their haste, they might not be doing their best quality work.
Variable Ratio
A variable ratio schedule (VR) provides a reward after an unpredictable number of responses. This is what gambling is based on (and why it’s so addictive). When you use a slot machine, you never know which spin will pay off. So, you keep playing, convinced your money might be right around the corner. You might use a VR to teach your dog to reliably heel by your side on walks. You can feed a treat every 10 steps on average, but sometimes it will be after two steps and sometimes it will be after 15.
But using VR can be tricky. The human brain likes patterns, so you might find yourself using an FR (dispensing rewards after a fixed number of responses) when you thought you were using a VR (dispensing rewards after a variable number of responses). It can help to use a computer to produce a random reward schedule for you. Or include life rewards like a chance to sniff on a walk. Sometimes your dog will get a treat, sometimes a toss of a toy, and sometimes nothing at all. Because your dog doesn’t know what’s coming and when, they will consistently work hard and steadily for you.
Fixed Interval
With a fixed interval schedule (FI), you only provide a reward if the response occurs after a fixed length of time has passed. It’s similar to a child’s weekly allowance. The child only gets their money on Sunday evening if they’ve completed all their chores for the week. Regular feeding times are an FI situation. If you teach your dog to sit by their food bowl before you lower it to the ground, sitting by the food bowl will only be rewarded once the time between meals has passed.
FI is predictable, so like just like FR, your dog can learn the routine. They will work hard right before the reward is due, but performance will drop off at the beginning of the next interval, just like a child who waits to do their chores until Sunday afternoon.
Variable Interval
A variable interval schedule (VR) provides a reward only if a response occurs after an unpredictable length of time has passed, similar to pop quizzes in school. Getting an A on a quiz might mean a gold star from the teacher, but students will never know the length of time between quizzes. They’re more likely to stay on top of their homework, just in case. VR can be helpful when training your dog to fetch to your hand. You can vary the length of time you wait before asking for a drop it, so your dog never knows when the reward will come and will hold the item until you ask.
Ratio
(The dog must respond a certain number of times before it’s reinforced) |
Interval
(The dog is only reinforced after a certain amount of time has passed) |
|
Fixed
(Number of responses or time between responses is the same) |
Result: Dog performs well right before the reward, but response rate drops right after | Result: Dog only reinforced after a certain amount of time has passed |
Variable
(Number of responses or time between responses isn’t always the same) |
Result: Dog performs well right before the reward but response rate drops right after | Result: Dog’s performance is steady without any pauses, but longer intervals have lower response rates |
Putting Reinforcement Schedules Into Practice
CR is great for teaching new behaviors, but to maintain behavior, intermittent schedules can be more effective. The unpredictable schedules of VR and IR may lead to a steadier response. If you can, be random. If your dog never knows when their human slot machine will pay off, they will respond with enthusiasm each time you ask for the behavior.
Even though interval schedules apply to certain specific behaviors, most of the time you want your dog to hold a behavior, such as a stay, for a given duration instead. So, although they aren’t technically schedules of reinforcement, add duration schedules to your training toolbox too. That involves varying the length of time your dog maintains a behavior before providing a reward. You can use a fixed duration schedule or a variable one, but the variable will be more effective.
It’s also helpful to practice differential reinforcement, which means rewarding only certain responses based on a set criterion. For example, this means only rewarding only the tightest heels or the fastest downs. This is a great way to improve performance and will add variability to your rewards as well. Before you add differential reinforcement, first introduce intermittent reinforcement gradually and purposefully. That will build your dog’s confidence and keep them interested in working for you. Then, when you start getting picky about what earns a reward, your dog will already understand that treats don’t come every time.
Finally, know your dog. You don’t want them to get frustrated or give up (extinction) due to infrequent rewards, but you also want to prevent boredom as well. While some breeds or individual dogs may happily perform the same behavior 20 times before receiving a treat, other pups or breeds might wander away if you don’t frequently pay them for their efforts. High rates of reinforcement keep dogs engaged.