Institute for the Study of
Healthcare Organizations & Transactions



Mechanisms of Hypnotic Analgesia

Although many early clinical studies were relatively primitive by the standards of modern research, they clearly demonstrated that hypnosis is an effective challenging agent for both clinical and experimental pain. The next question is how hypnosis works, and for that we must return to the laboratory. One theorist, taking a skeptical view of things, suggested that the reason hypnosis works is that most surgical procedures don’t really hurt anyway, and to the extent they do, hypnotized subjects try to please their physicians by reporting otherwise. I think that Esdaile had the most effective response to this claim when he wrote of the large numbers of patients flocking to his clinic for the removal of tumors:

There must be some reason for this, and I only see two ways of accounting for it: my patients, on returning home, either say to their friends similarly afflicted, "Wah! brother, what a soft man the doctor Sahib is! he cut me to pieces for twenty minutes, and I made him believe that I did not feel it. Isn’t it a capital joke? Do go and play him the same trick…". Or they say to their brother sufferers, -- "look at me; I have got rid of my burthen…, am restored to the use of my body, and can again work for my bread: this, I assure you, the doctor Sahib did when I was asleep, and I knew nothing about it…".

One thing we know is that hypnosis is not mediated by stimulating the flow of endogenous opiates. In a collaboration with Jack Hilgard, Avram Goldstein, who originally discovered the existence of specific opiate receptors in the brain, showed that naloxone, an opiate antagonist, has no effect on hypnotic analgesia (Goldstein & Hilgard, 1975). This finding has been subsequently been confirmed by other investigators (Barber & Mayer, 1977; Spiegel & Albert, 1983).  For hy_pa26.gif (9538 bytes)example, Moret and his colleagues gave unselected subjects suggestions for hypnotic analgesia during cold-pressor pain. The suggestions were successful in relieving pain by about 50% on average, but this success was not accompanied by increases in serum beta-endorphin levels. Nor did the infusion of Naloxone have any effect on the effectiveness of hypnotic analgesia. I don’t wish to sound like a dualist, but it appears that hypnotic analgesia is mediated by psychological processes, not by any indirect physiological effects. Click on the image to view an enlarged version.

Although some theorists have linked hypnosis to expectancy processes, it also appears that hypnotic analgesia is more than a placebo.  In one study,hy_pa27.gif (7963 bytes) McGlashan, Evans, and Orne recruited subjects for a study comparing hypnosis to medication in the relief of ischemic muscle pain (McGlashan, Evans, & Orne, 1969). Unbeknownst to the medical student who was running the study, during the drug trials the subjects received placebo packed in Darvon capsules. Insusceptible subjects got equivalent pain relief from hypnotic suggestion and from placebo. However, the hypnotizable subjects obtained substantially more relief from hypnosis than they did from placebo. The study suggests that hypnosis, like all effective analgesics, has a placebo component mediated by expectancies of success. Placebos are important, and they’re ubiquitous, but, in hypnotizable subjects at least, hypnosis is more than placebo. Click on the image to view an enlarged version.

As an alternative to the endorphin and placebo theories, Jack Hilgard has proposed that hypnotic analgesia involves a division of consciousness, in which an amnesia-like barrier prevents conscious awareness of pain (Hilgard, 1973; Hilgard, 1977). This proposal helps make sense of one of the paradoxes of hypnosis, which is that it alters people’s self-reports of pain but not their physiological responses to the pain stimulus. One interpretation of this difference is that the subjects are really feeling the pain after all, but we also know on independent grounds that physiological measures are relatively unsatisfactory indices of the subjective experience of pain. From the perspective of Hilgard’s neodissociation theory of divided consciousness, the reduced self-ratings are accurate reflections of the subjects’ conscious experience of pain, while the physiological measures show that the pain stimulus has been registered and processed outside of conscious awareness by the sensory-perceptual system. Put another way, hypnotic analgesia impairs the explicit perception of pain while leaving the implicit perception of pain intact (Kihlstrom, 1987; Kihlstrom, 1996; Kihlstrom, Barnhardt, & Tataryn, 1992). This dissociation is of considerable interest to those of us who are interested in consciousness and unconscious mental life. But because it is the conscious awareness of pain that bothers both experimental subjects and medical patients, the fact that hypnosis does not affect some physiological responses should not stand in the way of its clinical use.

On the other hand, the late Nicholas Spanos (Spanos, 1986) argued that hypnotic analgesia is achieved by the deployment of certain coping strategies identified by Donald Meichenbaum and others as stress inoculation (Meichenbaum, 1975; Meichenbaum & Turk, 1982). Now, there's no doubt that distraction, relaxation, imagining situations inconsistent with pain, and resistance to "catastrophizing" can result in substantial pain relief -- the question is whether these strategies account for hypnotic analgesia. Apparently, they do not (Hargadon, Bowers, & Woody, 1995; Miller & Bowers, 1986, 1993).

hy_pa29.gif (10443 bytes)In one test of the stress-inoculation theory, Miller and Bowers ran groups of hypnotizable and insusceptible subjects through an experiment involving cold-pressor pain. One third of the subjects in each group was hypnotized, and given suggestions for analgesia. Another third was hypnotized, but not given any analgesia suggestions. The remaining subjects were not hypnotized at all, but were instructed in the use of stress inoculation strategies of the sort that Spanos had proposed as mediators of hypnotic analgesia. The result was a highly significant, and revealing, interaction between treatment condition and hypnotizability: stress inoculation worked as expected, producing substantial pain relief, but hypnotizable and insusceptible subjects achieved the same effect. By contrast, hypnotic suggestions for analgesia were much more effective for hypnotizable than for insusceptible subjects. Note, too, that for hypnotizable subjects, hypnotic analgesia produced more pain relief than stress inoculation. But the important observation is that hypnotic analgesia is mediated by hypnotizability, while stress inoculation is not. Click on the image to view an enlarged version.

hy_pa30.gif (10066 bytes)In a later study, Miller and Bowers assigned subjects of low and high hypnotizability to stress inoculation and hypnotic analgesia conditions, but with a somewhat fiendish twist: during the cold-pressor tests the subjects were also administered a difficult vocabulary test. The idea was that consciously deployed cognitive strategies, such as those taught in stress inoculation, ought to consume attentional capacity and impair performance on the vocabulary test. This was, in fact, the outcome for the subjects in the stress inoculation condition. In the hypnotic condition, however, vocabulary scores were essentially unaffected, and for hypnotizable subjects actually went up a little. Click on the image to view an enlarged version.

hy_pa31.gif (8934 bytes)In the third and final paper in the series, Hargadon, Bowers, and Woody gave hypnotizable subjects analgesia suggestions of two types -- one suggested a lot of counterpain imagery, the other did not. Of course, some subjects in the imagery condition didn't use imagery, and some in the no-imagery condition used imagery anyway, and some just focused on the pain and catastrophized. The important result was that the use of imagery, whether instructed or spontaneous, had no impact on the success of hypnotic analgesia in these hypnotizable subjects -- suggesting, once more, that counterpain imagery is not central to hypnotic analgesia. Click on the image to view an enlarged version.

This set of results strongly suggests that whatever its underlying mechanisms, hypnotic analgesia is not mediated by stress inoculation and other consciously deployed cognitive strategies. For insusceptible subjects, incapable of responding to hypnotic suggestions, stress inoculation strategies can produce considerable benefit; but for hypnotizable subjects, hypnotic analgesia is probably preferable to stress inoculation as a psychological technique to control pain. In my view, this leaves Hilgard’s neodissociation theory as the most viable explanation of hypnotic analgesia. We may hope that our growing understanding of unconscious mental life will shed more light on the mechanisms by which this dissociation takes place (Kihlstrom, 1992a; Kihlstrom, 1997).

Click on the next section, Efficacy of Hypnotic Analgesia

Or, click on other sections:

Prelude to the Modern Era

Laboratory Research on Hypnotic Analgesia

Clinical Studies of Hypnotic Analgesia



John F. Kihlstrom, PhD

Plenary address presented at the annual meeting of the American Pain Society, Atlanta, Georgia, November 3, 2000.  The point of view represented in this paper is based on research supported by Grant #MH-35856 from the National Institute of Mental Health.  I thank Lucy Canter Kihlstrom for her comments.  Painting by Richard Bergh (1887).


Copyright © 2000 Institute for the Study of Healthcare Organizations & Transactions

Last modified:  04.08.2010 02:58 PM