Talking Terms - Positive Reinforcement

There is no shortage of recommendations for the use of positive reinforcement in today’s pet training and behavior books. Pet behavior professionals of various backgrounds have embraced positive reinforcement as a scientifically sound and effective way to change behavior without reliance on physical force or aversive stimuli. The increased focus on positive reinforcement in the dog training and behavior world has paralleled in many ways a similar movement in parenting and education – away from corporal punishment and toward a stronger emphasis on rewarding desired behavior. Because the discourse around positive reinforcement has become so prominent in the animal training world today, it’s important that we begin with a definition and clarification of positive reinforcement itself – what it is and what it is not.

Positive reinforcement is technically defined as the presentation of a stimulus or event following a response that results in an increase in some dimension of that response (its rate, its intensity, its duration, etc.) Because behavior is sensitive to its consequences, many responses will increase in probability over time when they are reliably followed by (and responsible for producing) certain stimuli or events. Colloquially, those stimuli or events are things we might characterize as desirable, rewarding, pleasing, or appetizing. Yet a stimulus is deemed a positive reinforcer ONLY when it has demonstrated effectiveness in increasing a response upon which it is contingent (or dependent).

For example, many pet owners complain that they praise and pat their dog every time she sits and are confused that she is still not sitting reliably when asked. “But I reward it all the time!” they say, “I tell her what a good girl she is and pat her on the head when she sits.” If sitting is reliably followed by praise and patting, and yet the dog is becoming less likely to sit when asked instead of more likely to do so, we might suspect the praise and pat on the head are not positive reinforcers after all. In fact, plenty of dogs might find a pat on the top of the head to be mildly aversive, as indicated by a momentarily hunkered down head or body posture when the hand reaches over for that friendly pat.

Reinforcement is defined by its effect on behavior and not by any qualities of the stimulus itself. What is reinforcing for one animal will not necessarily be reinforcing for another. Furthermore, what is reinforcing for one animal in one context may not be reinforcing for that same animal in another context. For example, we know that petting can function as a more potent reinforcer than praise alone in some cases (Feuerbacher & Wynne, 2015) yet one should not presume that petting a dog to reinforce ball retrieval will be reinforcing. In fact, the activity and arousal of the ball-retrieving moment might better be suited to another ball throw as a more potent reinforcer for the retrieval. Similarly, a dog whose responding is easily reinforced by his owner’s petting might find petting from a stranger to be frightening or off-putting (and therefore will more likely serve as a punisher than as a reinforcer).

There are many situations in which a consequence may function as a reinforcer in some contexts but not in others. In much of our pet training and behavior therapy, we rely on the use of food as reinforcement. For the vast majority of companion animals, a food treat can serve as a reinforcer, especially for simple responses that are prompted in non-stressful or non-challenging situations. Many owners, for example, with no special training, have been able to teach their dog to sit for a dry dog biscuit in the kitchen when they hold the treat in their hand and stand by the treat jar. However, when they try to use this dog biscuit at the dog park for teaching their dog to come when called away from his play with other dogs, they find the dog biscuit is insufficient to establish or maintain this response. In such cases, the owner may need to rely on chicken breast, dried liver treats, a tennis ball, or a Frisbee to get their dog to come when called. This does not mean that the dry dog biscuit is not a reinforcer. It does mean, however, that when faced with competing reinforcers (play with other dogs), the one we hold in hand must be sufficiently able to compete, and a dry dog biscuit may not do so.

Similarly, the same dog who learned to Sit for a dog treat in the kitchen but only learned to come when called for chicken breast when off leash may turn his nose up at the chicken that’s offered as an attempt to prompt a Sit while he is barking and lunging at an approaching UPS truck. Food refusal is one indicator of stress or fear in dogs, and his disinterest in food in this context does not mean such food items cannot serve as reinforcement in general, but instead that they cannot serve as reinforcers under this level of exposure to a provocative stimulus such as a UPS truck. In this case, the dog will need to be moved to a greater distance from the truck until signs of relaxation are obtained, and then presented with reinforcement for maintaining calm, relaxed behavior as the provocative stimulus is reintroduced at lower intensity (e.g., further away, parked, with engine off, etc.) Even then, certain reinforcers may function more effectively than others in this context.

When it comes to non-food reinforcers as well, people often assume a ubiquitous reinforcement effect should hold across situations. For example, I have worked with many clients who insist their dog attend the dog park because “she loves playing with dogs at our house and I want her to have the opportunity to exercise off leash in an enclosed area.” Yet when they describe that the dog cowers and runs between their legs at the dog park, avoiding or growling at other dogs, or tries to sniff along the edges of the park continuously, I have to advise them their dog might be better off without the dog park. Dogs who enjoy playing with some dogs simply don’t automatically enjoy playing with all dogs. That is, some dogs find the presence and interaction of almost all other dogs reinforcing. Other dogs find the presence and interaction of just a few specific dogs reinforcing. This latter group can live rich and fulfilling lives interacting with their regular dog buddies and skipping the dog park altogether.

Some reinforcers can begin to lose their efficacy as such through satiation. All organisms on earth will work to obtain certain reinforcers that are biologically relevant or necessary (food, water, shelter, access to mates). When they are deprived of these primary reinforcers, they will work harder to obtain them. When we use food in training and behavior therapy, for example, we can capitalize on its efficacy by conducting sessions before rather than after a meal. Thus, relative deprivation versus satiation on a particular reinforcer will, respectively, increase and decrease its momentary effectiveness.

In conclusion, when one considers the use of a positive reinforcer in training or behavior therapy, one must consider the following:

1) What is the quality of the reinforcer? Some stimuli simply will not function as reinforcement for certain individual animals. Each reinforcer must be defined by its effect on behavior  - specifically, by its ability to increase the response upon which it is contingent.

2) What are the competing reinforcers in the animal’s current environment? A stimulus that functions as a reinforcer in an environment with little to compete for the pet’s attention or behavior may not function as such when alternative reinforcers are available.

3) What is the state of deprivation or satiation on this presumed reinforcer? Some stimuli will function as reinforcers when the animal is relatively deprived of them but not when the animal is sated on that stimulus.

There are other factors that can impact the potency of a reinforcer (for example, its magnitude and the delay to its delivery) which are beyond the scope of this post. More will be written about positive reinforcement and its variables in future posts.