Thursday, November 10, 2011

Placebo Engineering

Since we know the placebo response is a localized Bayes' like update that factors in importance, we can come up with ways to game the system.

The obvious first few are:

  • Focus attention on the suggestion. Fire up emotions. Make it feel somehow important. Make it salient.
  •  Lower the Signal to Noise Ratio. Add noise and suggest small effect sizes.
  •  Distract attention away from alternate interpretations. Frame it as "You can notice how well it works" instead of "Either it works or it doesn't". Reduce salience of unwanted responses.

But there are more clever and more powerful tricks.

It is very easy to start an information cascade.  Framing it as "You will feel better because of the placebo effect" focuses attention on the belief itself, rather than the damage that the belief is supposed to represent! In theory, if it is interpreted exactly as stated, you have created a free-floating beliefThis alone works well. Doctors are missing a low hanging fruit here; they should be handing sugar pills out like candy.

In practice, it's not going to be completely free floating, but that's fine. Directing attention (salience) towards the preferred interpretation can easily push the loop gain over 1 once the loop is referring to the state of the map and not the territory. Over unity loop gains means instability. Once it's started, it can pretty quickly max out if nothing else gets in the way. 

Once I figured this out, I realized that my high susceptibility to motion sickness definitely fit the bill. Using the same tricks, I was able to get it under control and become less susceptible to motion sickness than most people (in about two minutes in the back seat of a car while reading and driving on curvy roads, no less :))

It is possible to fold this process back on itself (either implicitly or explicitly) and give them a placebo that "increases response to placebo response". This is a big part of a lot of hypnosis. The expectancy induction demonstrates this well, as it is explicitly expectation based. When done the meta part is done explicitly, it is known as the "super suggestion". I would have named it the "meta suggestion", but whatever.

EDIT: Here's how I got my motion sickness under control

I first realized and acknowledged that it was an information cascade and that there was no reason it had to stay this bad. I realized that I should be able to make a small improvement by placebo, then 'save game' there (since it's a floating node) and make another improvement.

I imagined that I had a dial indicator that represented my level of motion sickness. I let the needle move automatically instead of forcing it - I want it directly tied to my level of sickness, not to consciously force it to have the same level. Once I got that, I started tugging on the needle and noticing what effects it does have on my comfort level (making sure not to break the association by forcing it to zero when i still feel somewhat sick). I pretty quickly became comfortable enough to read in the back of a car driving up a curvy mountain road. I have since had problems with motion sickness, but it hasn't been nearly as bad and I've used other tricks too.

Both my seasick cousins that I showed this technique to reported that it helped. Pushing the needle up instead of down makes for a fun (if slightly mean) convincer!


  1. "Using the same tricks, I was able to get it under control and become less susceptible to motion sickness than most people." I'm very susceptible to motion sickness, and I'd love to get rid of it. But I can't tell how to do it, from your description. "Using the same tricks"? I don't see any tricks, unless you mean the ones in the bullet points. Those are *way* too abstract for me though. Any chance you could spell it out in more (lots of) detail?

    1. Does my addition to the post help?

    2. It does, thank you! I'm not good at visualisation, but I'll definitely give it a try next time I'm in a car or bus. BTW, can you recommend a good *practical* site about self-hypnosis?

    3. Hmm... I can't really =\

      Self hypnosis isn't any easier than hetero-hypnosis, and a quick google search turns up sites with one formulaic approach (which *can* work, but...)

      A practical 'how to' for self hypnosis is on my to-do list, but not at the top - partially because I struggle with it myself and don't feel like the expert on it.