Once the child is comfortable and fluent with the increased demands, then you can move to a variable schedule to maintain the behavior and you can also use a variable schedule to shape behavior towards the ultimate goal. There are several effective ways to implement a variable schedule and these are all based on the response the child gives.

The schedule is governed by the child’s response and not by some arbitrary predetermined schedule or by randomly reinforcing. Creating an effective variable schedule of reinforcement should be based on criteria so that behaviors at the better end of the range are selected for reinforcement. As I mentioned previously, if you reinforce at random or on a preselected schedule, you risk reinforcing lower quality behaviors.

Because behavior is naturally variable, there will always be some better responses that are closer to the goal and some lower quality responses that are further from the goal. You can take advantage of this to generate a variable schedule of reinforcement, based on criteria rather than chance or arbitrary schedules.

Here some examples:

• To increase distance. For example moving toward or away from something, start with a distance the child can do easily. When he starts coming nearer to the goal, reinforce only within the next level of your criteria. This is what I did with my son Doug when we were working on kicking a ball. I reinforced when he was within a circle 10 feet from the ball, then 8 feet and so on.
• You can do the same thing with duration. Let’s say the child needs to learn to sit on a chair. Tag sitting for 1 sec then 2 then 3. Raise criteria when the child shows that she can do it.

Any other parameters that govern the quality of a behavior can be shaped the same way. Raise criteria for one component at a time and tag only those attempts or successes that are in the upper range of what the child is offering.

A good technique to reduce stress while you raise criteria, and thus reduce the number of reinforcements is to ping pong back and forth, while still maintaining forward progress.

For example, say you are teaching a child to sit at a work table. You might tag after 0.5 sec, then 1 sec, then 2 sec, then 1 sec, then 2 sec, then 3 sec, then 2 sec, then 3 sec , then 4 sec. In this way you are moving forward, but it is not getting harder and harder for the child on every try.

Once you get to the point where the child can sit for longer periods, the fun work going on at the table will reinforce staying at the table. You could also occasionally reinforce a longer period of sitting that is within the higher end of the range – say 5 minutes. This would only happen once in a while to help maintain the behavior at a high level. Your variable schedule in this case would be governed by the child’s behavior and would work to maintain the behavior at the higher end of the range.

Here is a diagram to explain what I mean by behavior being naturally variable. At first you may see great examples of the goal behavior only some of the time. If you can reinforce mostly behavior at the top end of what is naturally occurring, then the high end performances will become more frequent and you will have even higher quality behaviors to reinforce at the high end. The behavior distribution curve will creep to the right over time. Eventually even the behaviors at the low end of your child’s performance will be what you previously considered to be great and the examples at the high end will be even better than anything you saw when you started.

Here is another example of the ping pong method I mentioned previously. In this example, we are working on taking steps forward. Tag for increasing numbers of steps and then back off a bit, then increase and back off. This allows the child to make overall progress, without it getting harder and harder on every turn. You can adapt to balance your child’s need for progress against his tolerance for increasing difficulty so as to make optimal forward progress with minimal frustration.