"Suggestible You" by Erik Vance explores this phenomenon, and really turned me around on placebos. I'm, generally speaking, not particularly susceptible to placebo effects, and always kind of (arrogantly) chalked placebo effects up to people just being gullible and wanting to see something that wasn't there. It turns out there are actually genes that might be responsible for the physical impact of the placebo effect in the brain.
There's an enzyme - catechol-O-methyltransferase - which is coded for by the COMT gene. This enzyme catalyzes the metabolism of dopamine in the brain. COMT has three common variations - AA/AG/GG - which substantially alters how effective the enzyme is at metabolizing dopamine. AA results in significantly reduced enzyme activity, which can result in dopamine built-up in the brain, which results in increased sensitivity to stress, anxiety, and pain, but comes with the bonus of enhanced cognition, motor skills, and memory.
AA genotypes also tend to be "placebo responders", while GG genotypes tend to be non-responders. This tends to imply that placebo responsiveness isn't purely psychological, but physical - and indeed, it turns out that in AA "responders" you can turn off the placebo effect by administrating naloxone (which works by binding to opioid receptors)!
This has really interesting implications for pharmacological research, too - if there are people who are genotypically predisposed towards or against placebo effects, then a drug trial that stacks responders in the trial group and non-responders in the placebo group which would produce a drug efficacy signal that could be just the placebo effect.