I've never understood placebos. Naively you'd assume they work the other way around, as the body attempts to achive homeostatis. So if you think you have taken a pain killer, the body would ramp up pain sensitivity to counter it. But as far as I know a lot of studies have shown placebos work.

One thing I haven't seen discussed is that maybe there is a subconcious social pressure element to placebos. Like: I just got a big injection from the doctor, surely it is going to help, I can't complain yet again about it. Sometimes I wonder, what if people take a placebo, report that they feel better, but continue to suffer internally? And are not even able to identify this, because their internal pain gauge is affected by the placebo. They think it helps. Maybe you could detect this by non-voluntary indicators of pain - avoidance reflexes, concentration loss, etc..

Back to the homeostatis argument, there are placebo-like effects that accidentially work. In autogenic training, you can warm up your body parts by imagining fire. What I think happens is, the body anticipates heat, widens the blood vessels to regulate against the heat, but in the cold environment this actually warms up your arms or legs.