"My dog won't ___ unless I have a treat!"
The use of food is a great way to motivate and reinforce behaviors, however there are a few reasons why learned behaviors may devolve when food is not immediately available or visible.
It’s Not a Treat
You will notice throughout this post that I refrain from calling the use of food a “treat” as the definition of “treat” fundamentally does not apply to the topic of this post.
Treat (n.) -  the act of providing another with free food, drink, or entertainment;  an especially unexpected source of joy, delight, or amusement.
When utilizing food in training and behavior modification, it is both expected and not free. It is an exchange for the exertion of effort, and a positive reinforcer acting to increase the probability that a desirable behavior will be performed again in the future. Food is a primary reinforcer, in that it requires no learning to be appetitive which makes it a powerful tool for increasing behavior.
All that said, the food you use should be something your dog truly enjoys or it will not carry the reinforcement value it needs to be effective. Rewards should be enjoyed!
Keep in mind that many of the behaviors that we ask of our dogs are unnatural, most dogs are not innately motivated to learn and perform these behaviors. Dogs can be motivated to perform behaviors for a variety of intrinsic and extrinsic factors such as; social interaction, exploration, tactile stimulation, and more! If we do not address the fuel behind a behavior, we cannot adequately motivate desirable alternatives.
For example, a dog who jumps while greeting could be seeking attention, after all that is how dogs greet one another. If the dog jumps on someone and receives attention for it (yes, negative attention is still attention), that behavior is reinforced as a successful greeting.
Let’s break that down a bit further, 9 out of 10 times the dog receives a food reinforcer for sitting when approaching a person, and also receives attention for being the good boy or girl they are. 1 out of 10 times, the dog jumps, the handler shrieks “off”, and the “greeter” may still smile and laugh. From the dog’s perspective this may still feel like a successful greeting. Thus every once in a while the behavior still pops up from its apparent dormancy in the off chance it can elicit the desired response. Behavior ultimately boils down to statistics and reinforcement history.
Now, if we approach the scenario differently, if we ask the dog to sit and provide attention and food rewards only while the dog’s rear end is on the ground, we are using what the dog ultimately may want to reinforce the desired behavior. If the dog’s rear pops up, the interaction abruptly ends, whether we take a step away from a “greeter” with the dog, or turn our back on the dog and walk away. The message is clear, “if you want attention, we want to give it to you, but only when you are sitting.” We have also just given the new desirable behavior more value to your dog, the behavior is the gateway to what the dog desires. While the unwanted behavior was devalued when it did not lead to the desired outcome, this is what extinction of behavior is all about.
Keep in mind this is simply one example, and all dogs are individuals. What one dog may value in any given scenario may be different from another and it will take some observation and experimentation to determine what each dog values. Thinking critically about how to distribute resources will help you to prevent reinforcement of unwanted behaviors, and increase motivation for new skills as they provide more benefit to your dog.
Positive reinforcement is a consequence of behavior, simply meaning that it occurs after a behavior has been performed and acts to increase the probability that the selected behavior will occur again in the future.
One method of teaching new behaviors and positions is luring, in which the food comes out prior to the initiation of the behavior to prompt and guide the dog through the correct response. While luring behavior with food can be a fantastic tool for teaching new skills or positions, it can also easily become a part of the cue if it is not faded from the process strategically and quickly. Using food to lure behavior past the stage of conscious competence (learn more about the stages of learning here) may perpetuate the need for food as information. It is information in the same way that a change in hand motion can signal a different behavior, the change from a hand holding food to a hand that is empty can signify completely different information to the dog.
Generalization takes time and effort. It is our responsibility to help our dogs by breaking down the skills they learn, and help them to become not only fluent, but confident in a variety of contexts. Returning to the example of sitting when greeting, the process of fading a lure from antecedent to consequence might look like this:
Using a piece of desirable food, lure the dog’s head up and back until the rear touches the ground, mark and deliver the food reward when the dog achieves the sit position.
Repeat 3-5 times.
Holding a hand as though you had food repeat the motion luring the dog’s head up and backward until rear touches the ground, mark when the dog achieves the sit position, remove food reward from hidden pocket and deliver.
Repeat, each time speeding the motion slightly until the hand motion feels fluid and natural and response rate is as desired.
Move to different locations and repeat, generalizing the skill to different parts of the home, outdoors, and public.
4. The Compromise Conflict
Occasionally performance of a known behavior fails despite a consistent history of reward, because they also have a consistent history of negative association in which the delivery of food was a concession for the loss of some high value desire.
When teaching the “drop” behavior for example, we often instruct clients to view the transaction as a trade, so that the dog does not become suspicious about giving up whatever it is that they have possession of. The simplest way to think of this transaction is a deposit into a bank account. We want the positive experiences to remain in a surplus. When experience hits a deficit problems can arise. That said, the exchange typically looks like this:
The dog begins with possession of a dirty sock that is always just out of reach. (10)
Cue the “Drop” behavior and offer a piece of food that they love, but also have access too every few days. (6)
The dog breaks from chewing the sock to come eat the food reward, and while they are eating the sock is taken away. (-10, +6) at the end of the interaction the dog feels like they are left with (-4) despite having received something else that was desirable, only less desirable.
Once the sock is gone the dog may move on to another activity, though the next time they find something amazing they may not be willing to take a break for food because at the end of the pervious interaction, they felt a loss. Over time this can develop into suspicion and guarding of valuable resources.
There is a concept known as the Premack Principle that put very simply, means that we can use behaviors that the dog wants to engage in, as reinforcement for behaviors that they are less likely to perform. For the sake of illustration, let’s reframe the “drop” scenario:
The dog has a special occasion bone that is taken out when you want some quiet time.
Cue the “Drop” behavior and offer a piece of food that they love, but also have access too every few days.
The dog breaks from chewing the bone to come eat the food. Immediately after they swallow the food you send them back to their bone for a few moments.
The core of the problem is that we tend to only practice skills in the situations in which we need them. Dogs being the masters of pattern recognition they usually are, quickly determine that this pattern often doesn’t end in their favor, are no longer motivated to perform the behavior despite the food reward. Remember that behavior ultimately boils down to statistics and reinforcement history. Practicing skills when you do not NEED your dog to perform perfectly will increase the chances that they will perform when it is vital. By using the Premack Principle and sending the dog back to their desired activity as a reward for cooperatively leaving the activity when given a cue helps to dissolve the compromise conflict and build a surplus of positive outcomes.
Another great example is the behavior of recall (come when called). Many pet parents often tell me that their dog is fabulous at coming when called at home, or indoors, but as soon as they are free in the backyard or the dog park it is nearly impossible to get their dog to come back. We have to first assess what is fueling the dog’s motivation and the result of performing the behavior. In these scenarios, we often call dogs as a way of ending their freedom. We call them in from the backyard when it is time for us to go to work, or at the dog park when it is time to leave. Sure they may get a great piece of food, but after a while they begin to weigh the value of the food against the consequences of cooperation. Instead of utilizing the recall behavior only when you need the dog to come to you, use it a few times when you first let your pup free! Call them back, give them a reward for stopping by, and send them away again. The food becomes the bonus for cooperating that it is intended to be, and the suspicion regarding which cue is the cue that leads to an unpleasant outcome will dissipate as the ratio of positive experiences to negative increases.
5. Food as a Distraction
Another common problem occurs when trying to use food to rein in the dog’s attention. Going back to the dog that jumps when greeting, if your approach procedure looks something like this:
Dog goes wild.
Food comes out.
Dog focuses, sits.
Eats food, while person approaches for greeting.
Jumps on person.
The problem may be in a lack of communication. In this scenario, the food is a distraction from the arousing stimulus of a person approaching. The dog is momentarily redirected, but when the food is eaten and gone, the excitement for the person remains. We may have effectively rewarded the brief sit, but may not have not communicated the importance of the calmness, impulse control, and a stationary position in the greeting scenario. Instead we must return to the concept of motivation. The treat should reward the behavior you want to continue, however in this scenario a brief sit is not the only criteria for success.
In this scenario, the dog is going wild because it wants to say hello (this hypothetical dog is friendly, please do not assume any dog going wild wants to greet). Start to reframe your actions so that this outcome is contingent on the desired response. This takes patience, but it is effective. The process of greeting would more likely resemble this scenario:
Ask anyone who would like to greet to stop out of the dog’s reach.
Wait for the dog to be calm and check in.
When you have the dog’s attention, ask for a sit.
If they sit, reward, but do not automatically release to greet.
Have the person slowly begin to approach while the dog is sitting.
If the dog pops up from the sit, have the person retreat.
Either wait for the dog to reset the sit, or cue the sit, and repeat.
When the person finally arrives at the dog, release the dog to greet while the person keeps their hands at nose level to help give the dog a low point to focus on.
Allow greeting for as long as the dog can remain calm, this may be a fraction of a second at the beginning, and call away before they attempt to jump, reward for leaving and repeat.
As your dog begins to learn that remaining calm draws the person closer and extends the length of the interaction, it brings value to the concept of impulse control. The calmer they can be, the more greetings they will have, and the faster the person will get to them, and the more attention they will receive. Using the food to hold the dog’s focus in some scenarios serves as a distraction not only from arousal, but from the information that could help them learn to cope with it. The moral of the story is to be strategic, food is a communication tool. Think critically about the message you are sending and how you can adjust the environment and your procedure to convey a more specific message towards a mutual goal.
6. Fading Reinforcement Strategically
To maintain the value in using food as a reinforcer we want to use food in exchange for effort. That said, returning to our 5 stages of learning, when we reach Stage 4: Unconscious Competence it is time to begin strategically fading reinforcement as there is little conscious effort needed to perform the learned behaviors.
If we suddenly stop reinforcing the behaviors altogether, we could develop extinction. Extinction occurs when a behavior is never reinforced and therefore disappears because no reason or motivation for the behavior currently exists. Extinction can be great when we are removing reinforcement for undesired behaviors, but it can be a problem when we want the dog to keep performing a particular behavior. To avoid extinction, we can utilize a schedule of reinforcement to assist us in fading out reinforcement effectively.
Up until this point you have likely been utilizing a Continuous Schedule of Reinforcement (CRF). This means that every correctly performed behavior reinforced. This is necessary when building fluency and proofing behaviors to a variety of contexts.
Fading reinforcement, means changing the reinforcement schedule gradually over time so that reinforcement is intermittent rather than continuous. Some intermittent schedules of reinforcement include:
Fixed Schedule of Reinforcement: The correct response is reinforced consistently after a fixed duration or number of performances.
e.g. sit - sit - sit - treat - sit - sit - sit- treat - sit - sit - sit- treat - etc.
e.g. 10 seconds - treat - 10 seconds - treat - 10 seconds - treat - etc.
One possible problem with fixed schedules is that the dog may quickly learn when the reinforcement will be delivered and will therefore put in their best effort immediately before reinforced event. A dog asked to sit 3 times between treats will often perform his best sit on the 3rd cue.
Variable Schedule of Reinforcement: The correct response is reinforced after a random number of repetitions or duration that is constantly changing.
e.g. sit - treat - sit - sit - sit - treat - sit - treat - sit - sit - treat - sit - sit - sit - sit - treat - etc.
e.g. 10 seconds - treat - 15 seconds - treat - 5 seconds - treat - 15 seconds - treat - etc.
Variable schedules are the most resistant to extinction and performance drop-offs because the reinforcement is unpredictable, the dog performs at the same level consistently “just in case” reinforcement is right around the corner. The trick to fling reinforcement effectively is to build on the process gradually, randomize starting in small intervals 2 behaviors, 4 behaviors, 3 behaviors between rewards and build your way up to more at a time gradually.
The most effective way to fade food reinforcers from your dog’s routine, is to employ a Variable Schedule of Reinforcement with Reinforcement Variety (VRRV) by using other things that your dog likes as additional non-food reinforcers that bridge the gap between food reinforcement.
e.g. Sit - Praise - Down - Play Tug - Touch (Targeting) - Treat - Sit - Down - Throw Ball - Touch (Targeting) - Sit - Treat - etc.
While this post is intended to give you some potential insight into obstacles to performance maintenance, it is not a comprehensive list nor an adequate diagnostic or treatment tool for behavior problems. If you are having trouble with your training or behavior modification plan, we suggest reviewing this information with a qualified trainer or professional to ensure that your plan is suited to you and your pet’s individual needs. For more information, please contact Simply Animal Behavior and Training Center at (847) 834-4765 or [email protected]