Carmen B. Pingree Training and the Scientific Method

Today, Cristina and I went to Carmen B. Pingree and had a training session while Bertrand was in preschool. (Bertrand did well again!)

I really enjoyed the training because it's grounded in the sound psychological and scientific principles. The Pingree approach aggressively adapts and tests methods shown to work in the research literature. The bulk of their approach is based on well-known techniques from operant conditioning. (They don't use the term operant conditioning, but that's what it was called when I took psychology.)

The high-level approach is straightforward: when a child does something good, you reinforce that behavior with a reward; when a child does something bad, you negatively reinforce that behavior to deter it. Honestly, it's not unlike training a dog. (Cristina and I trained our Chihuahuas to use litter boxes when we lived in an apartment.)

It's also important to emphasize that negative reinforcement need not be a painful punishment. With children--even autistic children--depriving a child of your attention is a much more effective negative reinforcement than something like a spanking. In fact, an instinctive reaction to misbehavior--yelling at the child--may positively reinforce them with precisely the attention they craved.

If you want to turn a behavior into a habit, then you use three phases of conditioning: (1) continuous reward, (2) fixed ratio reward and (3) variable ratio reward.

With continuous reward, if the child does something good, you reward immediately each time.

With fixed-ratio reward, you reward every nth action.

And, with variable-ratio reward, you reward every nth action on average.

Continuous reward makes it easy for the child to realize what a good behavior is. The ratio-based systems make the good behavior more difficult to "extinct." That is, if a child is trained only on continuous rewards, they'll stop doing the good behavior when the rewards stop. However, if a child is trained with ratio rewards, it takes longer for them to stop doing the good behavior in the absence of rewards. Research has shown that children (or animals) trained on variable ratio rewards take the longest to extinct their training when the reward is taken away. Over time, you can gradually increase the average ratio between rewards to strengthen the habit.

The other half of the Pingree approach is data. The Pingree staff collect data as meticulously as Cristina and I do. In some cases, they make a note of a child's behavior every 30 seconds. The goal of all this data collection is to allow an unbiased judgment of what really works for a child. They look for statistical trends in a child's progress associated with specific methods they've been using in the classroom. In this way, they can find out what works best for each child.

We know data collection is critical, because Cristina has data on every seizure Bertrand has had since entering the ketogenic diet. But she has more than that; she records every event of his life. She records the time of every event down to the minute; the amount and type of every food down to the milliliter; the calories by fat, protein and carbohydrate of each meal; the length of time and type of each activity (napping, in stander, reading etc.); the composition of each bowel movement; the amount of each medication taken; the type and duration of every seizure; his height and weight each day; and his ketone (and now glucose) levels with each urination.

Plotting data like this, clear trends emerge. First, the number of seizures have fallen from perhaps hundreds each day before the diet, to dozens after initiation, to no more than five on a bad day. We can also see strong clusters of seizure activity. Almost all of Bertrand's seizures now happen during feedings. Most of the rest are clustered around times when he's exhausted. The remaining few are scattered around stressful events like bumps, blood draws and bruises (and broken arms). Hardly any seizure occurs without a triggering event.

We're hopeful that the Pingree method and the decades of combined expertise on their staff will reveal what works and what doesn't when it comes to getting Bertrand to learn, to communicate and to interact. Stay tuned for the results!