Skip to content

Hypnosis and Placebos - Response Expectancy as a Mediator of Suggestion Effects

Source here.

The paper starts off with the usual and expected (ha) pro-sociocognitive preamble. Some notes…

  • Spontaneous amnesia is much less prevalent now, and is very likely a byproduct of suggestion and expectation. (They cite examples of where this happens based on expectation.)
  • In contrast to there being regularly “highly hypnotizable subjects,” there is (according to the paper, written in 1999) not a clear indicator of someone who will likely respond to placebo. The paper argues hypnotic response and placebo “are mediated by response expectancy.”
  • Unlike instructions, suggestions call for the response to be emitted automatically or without volition.
  • Meanings are contextual. Suggestions can be implied in the visual qualities of a pill, or by who is giving the utterance. Telling your roommate to sleep at 4am is clearly a command, but it’s clearly a suggestion coming from a stage hypnotist.
  • Simple compliance is not a successful response - it’s supposed to be non-volitional.
  • Nonverbal suggestions are still suggestions.
  • Even if someone doesn’t respond to a suggestion, it’s still a suggestion.

Kirsch explains expectancies and suggestions elegantly.

Accepting a suggestion means coming to expect that the suggested event will occur.

Kirsch, I. (1999). Hypnosis and placebos: Response expectancy as a mediator of suggestion effects. Anales de Psicología, 15(1), 99–110.

Response expectancies are often self-fulfilling prophecies - if we expect to be awake after drinking a cup of (unknowingly decaf) coffee, we’ll likely feel more awake.

  • Direct experience is more useful than vicarious experience (social learning) in producing responses.
  • Self-efficacy is a core component of hypnotic expectancies. EG - someone that often experiences anxiety may feel incapable they can control it, and in the same stroke someone may feel incapable of experiencing amnesia automatically. This component is a mediator, not determinant of response.
  • The expectation that something happens non-volitionally may confirm itself. EG - they expect the arm to move in a non-voluntary fashion, it moves, and the thought that it would move involuntarily is confirmed. The subject intends to move the arm in an involuntary fashion, the same way you’d intend to move it voluntarily.

Data indicates placebo cannot be explained by the following…

… rapport, trust, faith, hope, anxiety reduction, or endorphin release.

Kirsch, I. (1999). Hypnosis and placebos: Response expectancy as a mediator of suggestion effects. Anales de Psicología, 15(1), 99–110.

(The paper cites studies demonstrating this.)

Oof. If you’re unfamiliar with classical conditioning, this section will feel dense.

The naive (outdated) models of classical conditioning does not line up with the data. In this, the assumption would be that real medication (US) changes into a CS, producing CRs.

Modern takes (EG, Rescorla) on classical conditioning view it as a way organisms acquire information. Because of this, CRs differ from URs. The CRs are in preparation for the CS. CS and US parings only result in conditioning when there is valuable information. Kirsch posits that expectations mediate behavior - so if expectations change, the behavior does too.

Placebo response as a CR to a placebo CS doesn’t make sense, since it’s not in preparation for the CS.

Traditional (old) models don’t line up with the data:

  • Tranquilizers as UCs weaken the placebo response instead of strengthening it.
  • The CR is inversely proportional to the US. (If you shot someone with an absurdly strong tranquilizer, and the second one was a dud, you’d probably be confused.)
  • Placebos work for years and resist extinction.
  • Placebos can be stronger than active drugs.
  • Expectancies, when they differ from the CS, create effects that are different than the URs.

The Pavlovian model doesn’t explain placebos.

The paper posits that the aforementioned components (trust, faith, hope etc) are unlikely to mediate between expectancy and response. For example…

Positive expectancy => trust in the hypnotist => expected response

Later in the paper, they clearly mention expectancy may be just a mediator, not a cause of hypnotic response.

Hell, while I’m ranting - I think a friend brought this up at some point, but it finally clicked. If we always expected a response, why do people seem so damn surprised and amazed when they can’t remove their hand from a table? Sure - you could argue that if you hopped in an Ariel Atom, fully expecting it to accelerate quickly, and you’d still shit yourself in surprise when punching the gas. This doesn’t seem like it works so cleanly in the expectancy frame, where they’re self-fulfilling prophecies. Expectations would never work better than expected. There could be nuance I’m missing in the paper… but… it still feels like a gap.

Back to the paper - Kirsch highlights that expectancies shouldn’t be generalized, likely in preparation for talking about hypnosis.

Even the psychological construct of expectancy is likely to be too broad. Instead, we need to establish the physiological correlatives of specific expectancies, such as expectations of alterations in arousal, pain sensitivity, nausea, and so on.

Kirsch, I. (1999). Hypnosis and placebos: Response expectancy as a mediator of suggestion effects. Anales de Psicología, 15(1), 99–110.

The first section here is Kirsch’s usual take with hypnosis as a non-deceptive placebo.

  • Inductions (while trance is a pain in the ass to sociocognivisits) create expectations, like placebo
  • Kirsch insists hypnosis behaves like a placebo, despite others arguing against it. I believe this is still being debated today. Kirsch argues that some people expect hypnosis to work better than something called a placebo, therefore you can call hypnosis a placebo. In recreation - I don’t think this matters much.
  • Kirsch argues that variance in response (in high/mid/low responders) could be due to measurement error, and it’s all expectation. However, I feel like measurement error could be used carelessly to explain anyway anything that doesn’t line up with one theory. (EG - I don’t believe in dissociated control theory, but if they said “well measurement is often inaccurate, and if we account for variance, it’s more likely we’re right” it wouldn’t make their argument more convincing. TBH this thought is not completely original - it’s mostly a byproduct of something a helpful friend said who also read the paper. I also don’t want to point out their name and indirectly make them responsible for my academic-tourist-level understanding of hypnosis.)

They highlight a nuance in response and “secondary/sensory” suggestions…

  • Response expectancies elicit… well, responses.
  • Sensory (secondary) suggestions elicit change in a stimulus (EG - feeling.)
  • Sensory changes may be, in the framework of placebo, less of a self-fulfilling prophecy. The suggestion that someone will become less sensitive to pain is different from telling them the pain will lessen.
  • They say “internal states are more ambiguous than external stimuli.” (Highlighting that there may be a difference in suggestion there.)
  • Hypnosis often (but not always) differs from placebo in that expectancies have different targets.
  • Expectancy is, at minimum, a critical variable in the effects on behavior and experience.
  • A skilled facilitator will work with a client and their personal expectations. They should frame failures as temporary setbacks and highlight small successes to increase expectancy.
  • The concept of trance often does more harm than good in a clinical setting.