You found my old blog. Thanks for visiting! For my new writing, visit mikesententia.com.
There’s been an idea in the back of my thoughts lately: The placebo effect is a lot like Phlogiston*. This post is me exploring that half-formed idea.
*Phlogiston is a discredited 17th-century model of fire, saying it was caused by phlogiston. But there were no predictions: You discovered something contained phlogiston when you saw it burning, and saying “X contains phlogiston” only told you that it would burn. It has become a shorthand for a non-scientific model that only explains after the fact. More on wikipedia.
First, I’m not decrying the placebo effect in general. It is well-established that people sometimes get better, and that their expectation of getting better can trigger a perceived improvement, and sometimes an actual, objective improvement. Placebo is real.
But… One of my exercises for myself, to learn to explain my magick well, is to try and explain my results away as placebo. And here’s what I find:
- Sometimes, placebo makes the person get better for a short time.
- Sometimes, it makes them get better for a long time.
- Sometimes, it causes no result.
- And there’s no pattern to it. (Aside from the actual magick techniques I used, which, if we’re assuming it’s all placebo, shouldn’t matter.)
Now, I’m not saying this model of placebo is wrong, but it sure feels like explaining away, rather than explaining. Really, it feels a lot like Phlogiston: Once you tell me the results, I can say, “Yes, placebo did that,” but it has zero predictive power.
So, what would a predictive model of placebo look like? It would have odds. It would say, “For condition X, with a highly-believable placebo, Y% of people should feel better due to placebo, and Z% of those should have lasting results.” It would tell you how much the believability of the placebo matters: A doctor from Harvard vs the local community college, or energy healing on a believer vs a skeptic. In short, it would tell you how to design an effective placebo, or how to behave to minimize the probability of placebo. It would say a lot more than, “Sometimes, people just randomly feel better.”
Is anyone aware of research like that? It seems really useful for doing small-scale experiments when you can’t do a randomized controlled trial. Thanks!If you liked this post, consider visiting my current blog at mikesententia.com.