Author: Barry Morgenstern

In 1975, Robert Hawkins published a paper describing the case of a behavior analysis student who was planning to teach a student with severe impairments cursive writing. He describes in detail why it was a very poor choice of target behavior and how behavior analysis must get better at selecting appropriate target behaviors.

In the last 45 years, I think that the field has definitely done that. Now it would be rare to find a behavior analysis student that would make the kind of suggestion that Dr. Hawkins described.

But that doesn’t mean there isn’t a lot of room for Poogi. Most people accept a target behavior as appropriate if it has a “Student Centered Rationale.” The problem is that many, many behavior changes can sound perfectly reasonable when considered in isolation. But in my view, working on the wrong behavior changes substantially lowers the quality of the program.

A common problem is that behavior analysts pick too many target behaviors for the amount of time that is available to work with the student. Parents, behavior analysts, and others often want to address every single thing that is a problem – right away.  But those good intentions go awry when the skills do not maintain over time. It can take tremendous time and effort to do the proper work to obtain generalization and maintenance. If we have too many behavior change programs simultaneously, you are likely going to be too busy to do the appropriate generalization and maintenance work.

It is important to remember that each behavior change introduced adds many things that require time and effort – student time, staff time, materials prep, data, graphing, troubleshooting, program writing, reporting, staff training, generalization programming, and maintenance programming. This is part of the reason behavior analysts are so absurdly busy.

Most behavior analysts are no longer like Robert Hawkins’ student, working on things that don’t make any sense. Much more likely behavior analysts are working on many things that offer the child benefits, but those benefits are too small to be worth the time and effort that would be required to get generalization and maintenance. Our biggest time waster is working on too many behavior changes without proper generalization and maintenance, leading to little improvement overall.

A helpful tip: Judge how well students are doing – not by the acquisition data or the behavior reduction data – but by the generalization and maintenance data. If you don’t have time to do generalization and maintenance, maybe that skill wasn’t that important in the first place?

Many parents and behavior analysts have experienced the behavioral Whac-a-mole effect. Specifically, you work really hard and you solve a problem: Wow, she sits at the table for dinner! It’s amazing that the lawn mower doesn’t bother him anymore. He doesn’t get upset sharing with his brother! Everyone is celebrating, but… then you get the Whac-a-mole effect.

Specifically, as fast as you solve problems and produce important changes, new problems are starting to crop up. As soon as you successfully treated falling to the floor, all of a sudden you are dealing with throwing, which wasn’t a problem before.

Why? What causes this problem? Over the last few years, I’ve learned the cause from Greg Hanley. Greg very convincingly argues from recent research why this occurs.

Specifically, Greg argues that for any particular child, problem behavior occurs for multiple reasons, and those reasons are similar regardless of the variety of ways the child actually behaves. For example, sometimes he might just whine. But sometimes he yells, falls to the floor, hits, bites, punches, kicks, or knocks over furniture. If those behaviors occur together, there are probably occurring for the same reason.

So the Whac-a-Mole problem turns out to be fairly simple. If you only treat yelling when the child engages in a whole range of other behaviors for the same reason, we would naturally expect those other behaviors to increase. Almost as if the child is thinking “huh, yelling doesn’t work anymore? How about this?” This is still controversial at the time of this writing. But I think Greg is correct. The only long-term solution is a skill building approach, where the behavior meets a natural contingency.

So everything has been going great in therapy! He hasn’t had any problem behaviors in three months! He is playing with friends! The teacher reports he is ahead of most of the class in reading and math!

As always, we want to know if these gains will last over the long haul. A great day to test – Thanksgiving. That’s because Thanksgiving is often chaos. There are lots of relatives and friends who the child might not see frequently. It’s loud. The schedule is off. What happens is unpredictable.

If your Thanksgiving goes well, great! But if not, there is still more Poogi to do. Our goal is not success in therapy. Our goal is success in life.

At some point in the last 35 years or so, the phrase behavior modification went out of favor and the field started to use the term behavior analysis. There were good reasons for this. First, some people had done some terrible things in the name of behavior modification and there was a desire to not associate with the term. Second, behavior modification is not what good programming is about. Good programming comes from the analysis. Unfortunately, despite what it is called, many of us are still behavior modifiers and not really behavior analysts despite what the certification and ethical rules say.

A common phrase among poor behavior modifiers is to “throw spaghetti against the wall and see what sticks.” So, when we work with kids, we try things to see what works:

• He won’t come away from the computer. How about trying a timer?
• He doesn’t want to sit for instruction. Have you tried a token board?
• She is getting bored easily. What about mixing and varying some different tasks?
• He gets upset during transitions. Maybe we try giving him a visual schedule?

There is nothing wrong with these interventions if used as part of a carefully planned program. But often they aren’t.

There is no doubt that these types of interventions work sometimes. If they don’t, poor behavior modifiers will just throw more spaghetti. It’s likely that eventually an intervention will be effective (at least temporarily). But we are fooling ourselves.

No matter how much spaghetti we throw, we will always end up with a big problem. Maybe the spaghetti will last for two weeks, two months, or even two years. But the problem always comes up: The intervention doesn’t last over the long-term. That’s because we have a huge pot of spaghetti and no analysis. We are judging the intervention on the wrong criteria. Is there problem behavior? Sure, that’s important- but just one aspect of the problem. What do we expect the parent or teacher to do? Remember to implement this carefully crafted list of often cumbersome strategies that are often impractical in natural settings? Forever?

One thing this is great for- blaming parents and teachers when the intervention doesn’t last. We can say, “well the plan worked great with us, but the teacher, parent, paraprofessional, or grandma won’t implement it.”

The only thing that matters in the long run when treating problem behaviors is: Are the alternative behaviors naturally reinforced? If not, we should not expect it to maintain.

Why We Need to Aim for Natural Contingencies

In an old article on treatment of problem behavior, the researchers compared two treatments for problem behavior maintained by attention from adults.

The first treatment was a time-out. The second treatment was communication training which taught the children to say “Am I doing good work?” The researchers found that both treatments were highly successful.

But then the twist: The researchers had the children work with naïve adults who didn’t know the history. Under those conditions, the children who had previously had been treated with time-out went right back to problem behavior. The children treated with communication training remained successful in the new situation. The students who were originally treated with time-out became successful with the naïve adults after they received the communication treatment.

The article demonstrates how we can fool ourselves into thinking our treatment is more effective than it really is. All the time-out students were looking good. Imagine teachers raving about the student progress at IEP meetings. But all it takes is a novel person showing up and the behavior falls apart. In reality, treatments in practical settings are almost always dramatically more complicated than the article describes, which makes it even more difficult to maintain progress. It is not if the behavior will deteriorate, it’s when.

Now, do we believe all these children who experienced the brief communication training treatment were likely successful in the long-run? Not likely unless they got a lot more therapy.

But I do think the article beautifully captures what our ultimate goal should be. The behaviors we teach have to be effective in the real world. And in order to be effective, they must meet a natural contingency. That means it was reinforced without any specific person having to think about it or plan it. The children said, “Am I doing good work?” Then the naïve adult was able to respond in a way that was reinforcing to the child. No one had to train the adults to follow the behavior plan since most adults will naturally respond to “am I doing good work” with attention. There are many examples in the research literature, but not enough. If you don’t get to this point, the behavior is unlikely to maintain.