Too much of a good thing is going to be inevitably bad. So how do we scale things back?

Differential Reinforcement of Low Rates of Responding, or DRL is another procedure to help curb bad habits through differential reinforcement. What sets this one apart from its cousins, DRA, DRI, and DRO, is that this procedure doesn’t want to eliminate the behavior. Instead, it just wants to decrease how often the behavior occurs. 

DRL is usually used to target behaviors that aren’t problematic by themselves – the only issue is that they occur too often. To achieve these goals, DRL is commonly paired with a fixed interval schedule of reinforcement. If a subject performs a behavior fewer times than a specific number, then they will receive reinforcement at the end of a time period. 

Alternatively, a study has reported that DRL in practice is used to eventually bring a problematic behavior down to a frequency of 0. The reason for this is DRL offers a structured way to steadily encourage a subject to reduce the problematic behavior themselves. DRA and DRI would reinforce the subject for doing a behavior that isn’t the problematic behavior (whether it achieves the same goal in a socially acceptable way or makes the problematic behavior impossible are the primary concerns of DRA and DRI respectively). DRO comes the closest to achieving a similar effect, but it would work better as a way to maintain a frequency of 0 rather than teaching the subject to slowly bring their response rate to 0.  

It sounds a bit complicated, but let’s look at some examples to clear things up.

Lab Example

Here’s the link again to the study mentioned above, but in that study, DRL was used on students in a school setting to reduce the number of times they performed attention-seeking behavior.

An important note, if you are reading a scientific study on behavior, always pay attention to how they operationally define the targetted behavior- as in, how they define a behavior so that it’s measurable and observable. An operational definition must be clear enough that any observer can look at what your subject is doing and say “ah yes, they are indeed doing the behavior we are recording.” I’ll probably dive into this more when we get to ABC analysis of behavior. 

Going back to the study, the subjects had a high rate of attention-seeking behavior. In this scenario, it was operationally defined as “repeated requests for assistance when the children should have been working independently on tasks matched to their ability levels, but also included requests for approval.” Asking for help isn’t inherently a problematic behavior, but when a child is doing it too often, then it can become a problem. Thus, DRL is the best solution. The teacher in this study also made a note that the children who would perform these behaviors were also capable of completing the task by using classroom resources like dictionaries or encyclopedias, instead of constantly asking for help. 

Each interval for the children was a single 20-minute session. Just like any good study, baselines were observed for each child, and the acceptable rate of response, how often the child was allowed to perform the target behavior, was created to match each child.

Each child was given an index card with boxes on it equal to the number of times they were allowed to ask for help or attention + 1. For example, Elin, one of the children in the study, was allowed to ask for help or attention 3 times, so her card had 4 boxes. Each time Elin asked for help, the teacher would walk over, initial the box, and then help Elin. However, if the children had even one empty box by the end of the 20-minute session, then she would earn a point for her team, which could be traded for more computer time or longer breaks. This way, the children were reinforced for keeping their response rates low at the end of the 20-minute session. 

Dog Example

In my eyes, dog examples of DRL are difficult to come by. It’s difficult to teach a dog that only a certain number of responses is allowed, but after that, they won’t receive any reinforcers. 

Another layer of difficulty is that a good number of problematic dog behaviors are also self-reinforcing. Let’s take for example a common behavior that people would make a good target for DRL – excessive barking. While DRL makes sense in this scenario, it’s difficult to perform because barking is by definition self-reinforcing. Some dogs enjoy barking and barking at stuff is fun for them.

In my experience, I’ve only successfully used DRL on a dog once. It was with a gorgeous Akita named Kato (pronounced Kay-toe). In our shelter, there were two sides to each of the kennels. The front, where guests could walk by and look at each of the dogs, and the back where trainers and volunteers could get into the kennel to interact with the dogs. If you approached Kato from the backside of the kennel, he was the happiest dog on the planet. In fact, I loved saying “Hi” to him during my shifts because he would always do an adorable tippy-tap dance when I walked by. 

However, if you approached Kato from the front, he would immediately bark at you non-stop. You could be the same person, and he would still continue barking at you. Personally, my guess for this behavior was that Akitas usually attach themselves to a “family” to protect. I think Kato saw the backside people as family, while the front side were all strangers. 

Regardless, I started a DRL procedure with him because someone was interested in adopting Kato thanks to his alertness. 

Since I didn’t want to extinguish his guard barking entirely, DRO, DRA, and DRI procedures were out of the question. DRL would help me lower the number of barks he did and help prevent a barking frenzy on the front end. 

To start this procedure, what I did was first enter the kennel through the back and teach Kato how to speak. This would teach him how to bark on command. Soon, I got him into a consistent pattern of 3 speak commands and a sit command would be a reinforcer.

After that, I tried teaching this to him from the front end. Man, that was an uphill battle. If you haven’t worked with dogs before, dogs are very poor at generalizing commands. What can work in one context, will not apply well to another for them. 

For Kato, this means that everything I taught him from the back, I had to now teach him from the front when he was much less cooperative. Over time, with the help of some mat calming routines, I got him to relearn “sit” from the front. After that, I set up a white-board with 4 large white squares. Each time Kato barked, I would fill in a white square with an X. At the end, if there were any white squares left, I would tell Kato to sit, tap the board, and then reinforce him. Otherwise, if the board was full, then Kato wouldn’t receive any reinforcement. 

I couldn’t tell if this idea worked or not, but Kato did eventually reduce his barking. I just couldn’t get his barking down to only 3 barks. But this routine did reduce the amount of barking – which was something I was more than happy to settle for.

Conclusion

Differential Reinforcement of Low Rates of Responding is a behavioral procedure where your goal is to reinforce the subject for a low rate of behavior. This is usually used to target behavior whose only problem is its high frequency. However, this procedure can also be used to gradually decrease a problem behavior and then switch into a DRO procedure. 

The study I talked about shows a great way DRL was used to lower the response rates of 3 kids who were asking for help and attention too often. 

Kato, one of my favorite Akitas I got to work with, had an issue with barking too much. While DRL doesn’t usually work well with dogs, Kato was one of the few dogs I decided to try it on with uncertain results. It did eventually lower his barking, but not to the degree I wanted it to. 

If you liked this article or thought it was entertaining, please hit the like button. If you think I forgot something in the article or have an idea of how I could have better used DRL on Kato, then please write a comment down below. 

Thank you for reading!