Monday, April 17, 2017

Set up an Experiment

In my last post I submitted that the first step to simplifying the interaction was to elicit a specific prediction.

The measurable prediction from that example ended up being:  “If the shoulder is damaged, there will be pain and weakness that won’t improve above 90 degrees of shoulder flexion.” Performing the first step well and getting good measureable predictions makes the next step so much more simple.

The next step is to set up an experiment that attempts to refute this prediction.


This step even has a name: behavioral experiments. I have long been enamored with behavioral experiments because it seemed consistent with what we do in therapy. But, as I mentioned in the first post, adding the concept of Expectancy Violation significantly improved the clarity with which we can engage targets of interventions. It is, in my view, a significant step ahead of targeting beliefs.

I have also long argued that our role as therapists is not that of one who makes the change for the patient. We don’t take pain away or heal pain. Instead, we set up the scenario in which the patient comes to their own conclusions and makes their own changes. We are context architects. We don’t perform the experiment. We set up the lab for the patient to run their own experiments. We are Alfred, not Batman.

So, at this stage the name of the game is to set up experiments for the patient to run. Here’s the thing. We WANT and are TRYING FOR a refutation of the prediction. So, we set up the experiment with this in mind and with skill.

Can we confront them with strength, as Greg Lehman puts it?
Can they show themselves several position in which the shoulder is strong and therefore opposes the narrative of the weak and damaged/fragile shoulder? Can they show the capacity to improve their strength in the position of concern, more specifically refuting the prediction?

Can we disconfirm with a different way of doing the same thing, a novel movement?
Can they get the shoulder above 90 degrees using a different approach? Maybe they can get there if the hand is planted and stationary and the body moves away from the hand (closed chain shoulder flexion). Perhaps they can simply get their arm above 90 passively, refuting the specifics of the prediction of the shoulder being above 90 in any form (and gets us to a now more specific prediction about the way in which the shoulder must get to 90)

Can we use various physiologic mechanisms like conditioned pain modulation and exercise induced hypoalgesia?
If they perform a series of isometrics in a comfortable position (or even hunting for pain, as Erik Meira puts it), does the manner in which they can subsequently raise their arm change?

Can we use desensitization?
If they repeatedly move up to and away from the position at which the protective behavior first becomes noticeable, does the behavior change? Do they gain motion or otherwise improve?

These are but a few of the many possible experimental methods in our laboratory arsenal by which the individual may test their predictions.

Some things that are important to include in this step.
  1. Again the prediction needs to be specific about the outcome of the experiment. Just like in published research we don’t want any post hoc data mining to be able to squirrel around a negative finding that makes it seem positive. We need a clear demarcation of what meets and what refutes the prediction.
  2. It may be helpful to track confidence in the prediction. “How sure are you that this bad outcome will occur if we do this?” They can rank this on a scale of 0-10. Then repeat the question after the experiment. This may help you gauge the success of the refutation. If the experiment was clearly negative (it didn’t bring about the bad outcome) but doesn’t change the confidence rating of the original prediction then something is askew and you may need to alter the manner in which the experiment is taking place or, more likely, the specifics of the prediction. Also, this re-assessment gives the individual a chance to verbalize explicitly and therefore acknowledge a change in their narrative.
  3. Try to come up with or build toward experiments that can be repeated and run by the individual on their own. More on this in the next post.

Remember that the individual will update the prediction after the experiment is complete and so eliciting the prediction step is repeated. The shift in the prediction may be subtle or it may be large. Either way, confidence in the result will need to be built. That’s the next post.

*I recorded a webinar with Modern Pain Care that touches on these topics and exposure based care if you are looking for some additional information.