I& of Abnormal Psychology 1966, Vol. 71, No. 3. 196-208 0 RESPONSIVENESS TO FOLLOWING WAKING AND IMAGINATION INSTRUCTIONS AND FOLLOWING INDUCTION OF '

ERNEST R. HILGARD AND CHARLES T. TART2 Stanford University

2 experiments are reported to test the increase of responsiveness to tests following over responsiveness to such tests in waking and imagination conditions, an increase that has been doubted as a result of experiments by Barber and Calverley (1962, 1963). In the 1st experiment, 60 Ss were divided into groups of 20 Ss each serving under 1 of 3 conditions in a 1st session (waking, imagination, hypnosis). All received a standard hypnotic induction in a 2nd session. While the treatment effects did not yield significant differences on the 1st day, there were significant gains in responsiveness to suggestions by the waking and imagination groups in the 2nd session. In the 2nd experiment, with some methodological improvements, 90 Ss served in 6 groups of 15 Ss each, in imagination without expectation of hypnosis, imagina- tion with expectation of hypnosis, and hypnotic induction, in various combina- tions. Significant gains were found with hypnotic induction throughout. State reports (subjective responses of drifting into hypnosis) showed that those Ss withii both imagination and hypnotic induction conditions who reported them- selves as becoming hypnotized were the ones who yielded the highest sug- gestibility scores. The difficulty of obtaining significant treatment effects is noted unless Ss serve as their own controls.

Increase in suggestibility has so long been to suggestions in the waking state and follow- associated with hypnosis that such increase ing hypnotic induction, two facts stand out: has commonly been considered to be a defin- 1. There is a substantial positive correla- ing characteristic of hypnosis. Thus the titles tion between responsiveness within the wak- of standard books reflect this intimacy be- ing state and following hypnotic induction. tween hypnosis and suggestibility, for exam- This implies that responsiveness to sugges- ple, Hypnosis and suggestibility (Hull, 1933), tions carries within it a heavy loading of indi- Hypnotism: An objective study in suggesti- vidual differences common to both the waking bility (Weitzenhoffer, 1953). It is well-known, and hypnotic states. ' however, that responses to suggestions of the 2. The increase in responsiveness to sug- kind given within hypnosis may be obtained gestions over the waking state engendered by outside the hypnotic state; such suggestions a prior hypnotic induction is usually found are often called waking suggestions, a term to be significant, but the gain is typically of convenience that can be used without small "probably far less than the classical implying that hypnosis is a sleep state. When hypnotists would have supposed had the comparisons are made between responsiveness question ever occurred to them [Hull, 1933, =The investigation was carried out in the Labora- p. 2981 ." Hull's conclusions of a significant tory of Hypnosis Research, Department of Psy- but small gain following hypnotic induction chology, Stanford University, with the aid of a have been verified by Weitzenhoffer and Sjo- grant from the National Institute of Mental HeaIth, berg (1961) and by Barber and Glass (1962). Public Health Service (Grant MH-03859). Grateful acknowledgement is made of the assistance of the The second of these summary conclusions staff and research assistants in the Laboratory for has been challenged by Barber and Calverley their aid in the conduct of the experiments reported. (1962, 1963) on the grounds that the gains At the time C. T. Tart was holding a Public Health Service Postdoctoral Research Fellowship (1-F2-MH- in suggestibility found experimentally are to 14,622-01). be attributed to circumstances of experimen- 2 Now at the University of Virginia. tation, such as aroused task motivation, YD CHARLEST. TART

rather than to the hypnotic state as such. 1. The variability in responses to hypnotic When they aroused task motivation in both suggestions from one subject to another is the waking state and the hypnotic state, dif- very great, the distributions are often flat, ferences between the two states became mini- sometimes 'bimodal, with extreme scores at mized or disappeared altogether. Their con- both ends of the scale (Hilgard, Weitzen- clusions rest upon an assumed equation of the hoffer, Landes, & Moore, 1961). Under these task motivations aroused under the two con- circumstances two risks are taken in compari- ditions. Unfortunately, their instructions dif- sons between small groups: First, that fered in important respects between their genuine differences may be obscured because waking groups and their hypnotic groups. In they are small and the variability great; addition, by contrast with the earlier Barber second, that obtained differences, meeting and Glass (1962) experiment, they no longer conventional statistical standards of signifi- used subjects as their own controls, so that cance, may arise when assignment is random other differences, besides the added task through the chance appearance in small groups motivation, occur between the experiments. of disproportionate numbers of extremely There are important methodological con- high-scoring or extremely low-scoring subjects. siderations that arise when two experimental 2. The nature of responsiveness to hyp- conditions are to be compared, when (a) re- nosis is such that within a random sample sponses between the two sets of conditions there will aDtxar a substantial number of are highly correlated, and (6) when the essentially nidsusceptible subjects, perhaps as expected differences are small. If the pre- many as two thirds of the group. Such sub- ferred hypothesis calls for no difference, this jects will dilute any comparison between result is more likely to emerge if group com- randomly selected groups assigned to waking parisons are made, and the correlation term and hypnotic conditions. Because they are neglected, as in the Barber and Calverley relatively insusceptible, gains between waking studies. If one seeks to detect any difference, and hypnosis are not to be expected; because however small, then the subject is likely to they for the most part extremely low be used as his own control, to take advantage scores they will contribute to low means and @ of the correlational term. The considerations high variability relative to the mean, thus are not entirely statistical ones, however, but reducing the probability of significant differ- depend on the appropriateness of what is ences between the compared groups. done to the actual relationships being studied. 3. At the other end of the scale, a few Because these problems are peculiarly dif- subjects with very high scores enter hypnosis ficult within hypnosis, some additional com- spontaneously withour a prior hypnotic in- ments are in order before proceeding to a duction, and for them also the changes from description of the current experiments. waking to hypnosis will be slight or absent. There are good reasons for not using sub- Because their scores are high, they also in- jects as their own controls within hypnotic crease the variability that reduces the signifi- experiments, particularly because of tenden- cance of differences between means. cies, whether deliberate or unconscious, for As a consequence, in group comparisons, subjects to modify their behavior under con- the weight must be carried bv those few trol conditions when they know that a hyp- subjectsvwho make substantial gains between notic session is to follow (Zamansky, Scharf, waking and hypnosis. An overall compari- & Brightbill, 1964). There is no assurance, son between two groups that does not take however, that expectations are circumvented into account the correlation between waking by having only one condition per subject responsiveness and hypnotic responsiveness either, if a subject knows that other subjects will understate the gains that take place for are being hypnotized. Failure to use subjects some subjects. as their own controls greatly attenuates the In view of these considerations. the experi- possibility of finding differences under two ments to be reported have been carried out conditions of experimentation. This follows in such a manner as to reveal some of the because : finer grain of the changes that take place RESPONSIVENESS between waking and hypnotic conditions, and Today you will be given tests of imagination the consequenc& of alternative treatments of under normal conditions. You will not be hypno- tized. The better you can imagine, the more you'll the data. To this end, the instructions fol- respond; try as hard as you can to concentrate lowed are essentially those of Barber and and to imagine that the things I tell you are true. Glass (1962)) rather than the task-motiva- It may be noted that a deliberate effort was made tion instructions of Barber and Calverley to keep the waking subjects from drifting into hyp- (1962, 1963). It is conjectured that the nosis; while the imagination subjects were told that Barber and Calverley results do not require they would not be hypnotized, nothing was said the task-motivation interpretation at all; about spontaneous hypnosis. In order to control the drift into hypnosis in the that is, had the Barber and Glass subjects waking condition, and to assess the extent to which not served as their own controls, significant the subject became spontaneously hypnotized under changes would not have been found. The the imagination condition, or following suggestions design of the experiments to be reported within the hypnotic induction, a "state" report was frequently called for. permits both types of comparison, that is, The "state" report represented a modified version one in which random groups are compared, of such reports earlier used by LeCron (1953), and one in which the subject serves as his Hatfield (1961), Tart (1963). Before experimenta- own control. tion began the subject was taught how to use the scale, four numerals from 0 ("zero") to 3. He was told to respond "zero" to the question "State?" if he felt in his usual, normal, wide-awake state, to respond "1" if he felt himself very relaxed or drifting off, as in going to sleep, to respond "2" if he felt himself in a mildly hypnotic state, and "3" if more Instructions and procedures. The instructions in deeply hypnotized. Because depth of hypnosis often this experiment were modeled very closely after those fluctuates within a session, these reports were called of Barber and Glass (1962). Of the 60 subjects, for repeatedly within the session, typically before from an introductory psychology class, 20 were ran- and after each test suggestion. If a subject in the domly assigned to each of three treatment conditions waking condition gave a report of "1" he was on the first day of the experiment, the groups being aroused so that he returned to a report of "0." designated Waking, Imagination, and Hypnotic In- For the hypnotic condition, the standard induc- duction. On the second day, all subjects received the tion method of Form C of the Stanford Hypnotic hypnotic induction.3 Thus the first day permitted Susceptibility Scale (SHSS) was followed (Weitzen- comparisons among randomly assigned subjects (fol- hoffer & Hiigard, 1962). This induction involves eye lowing the Barber and Calverley data treatment), fixation and eye closure, along with suggestions of while the second day permitted treating the subjects relaxation and sleep, and requires some 10 minutes. as their own controls (following the Barber and It may be noted that the waking and imagination Glass treatment). instructions were much briefer. Exactly the same set For the waking condition, the following statement of test suggestions followed waking, imagination, was made to the subject prior to his receiving the and hypnotic instructions. These represented 10 items test suggestions: of SHSS, Form C, omitting the taste hallucination and the hallucinated voice, in order to shorten the Today you will be given tests of responsiveness time required. An alternate but equivalent form, to waking suggestions. You will not be hypnotized. designed for the purpose, was used on the second We want to be sure that you do not become day .4 hypnotized; if you slip into a hypnotic state in- advertently, as indicated by your state report, we Res~ltsof Experiment I. The major results will bring you back to your normal state. It is are shown in Table 1. While the means of the still possible to respond well to hypnotic-like first day are in ascending order by conditions, suggestions in the wide-awake normal state. a simple analysis of variance, based on the For the imagination condition, the following one day only, shows that, as anticipated, statement, adapted from Barber and Glass (1962), they do not differ significantly. If these were was read to the subject: the only data collected, the authors would

3 For another purpose to supply a pool of subjects agree with the conclusions of the Barber and contrasted in their waking and hypnotic responsive- Calverley studies that hypnotic induction ness, another 20 subjects were run at the same time adds nothing significant to suggestibility, under the waking-hypnotic condition. The 20 sub- jects treated as paralleling the imagination and 4A limited supply of the alternate forms is avail- hypnotic condition represent a random half of the able on request to the Laboratory of Hypnosis Re- total 40 subjects run. search, 582 Alvarado Row, Stanford, California. TABLE 1

Scores on suggestion tests (maximum = 10) Difference: change from first to second session First session Second session Group 1 Waking condition Hypnotic induction (N = 20) M 2.88 M 4.90 M +2.02 SD 2.78 SD 2.41 (1 test, p <.0005) Group 2 Imagination condition Hypnotic induction (N = 20) M 3.70 M 5.70 M +2.00 SD 3.15 SD 2.79 (t test, p <.005) Group 3 Hypnotic induction Hypnotic induction (N = 20) i%f 4.90 M 4.55 M -0.35 SD 2.50 SD 2.53 (1 test, ns)

Note.-Results of overall analysis of variance: Group effects, ns; first session differsfrom second, p < .001; Interaction. Groups X Sessions. p < .01. Results of simple analyses: Group effects, First session, ns. Group effects, Second session, ns. even though no special task motivation is and there is a significant interaction effect added, within the waking and imagination (p < .01) between the treatment conditions conditions. If the assignment of subjects was and the days, which the t tests pinpoint. truly random, and there were no order-of- The experiment just described is generally session effects, no differences would be ex- satisfactory from two points of view: (a) by pected on the second day either. A simple using their design for group comparisons, it I analysis of variance on this day also con- replicates the nonsignificant findings of firmed the conjecture of nonsignificant dif- Barber and Calverley (1962, 1963)) without I ferences. There were, however, significant requiring an explanation in terms of task gains between the two days for both the motivation, and (b) by adding a session to (5'1 waking and imagination groups, and no gain permit the study of changes by individual I (a trivial loss instead) for the group experi- subjects, significant differences are found, encing hypnotic induction on both days, a confirming the findings of Barber and Glass result that Barber and Glass (1962) had (1962)) and indicating that the Barber and also found. Calverley design is not sensitive to small but The reason that it is possible to detect the significant changes that may indeed occur effects of induction in this design, and not within hypnosis. in the random assignment one alone, is that There is a flaw in the current experiment, this design takes into account the correlations however, which is also in common with the between waking and hypnosis scores. In this experiments replicated, in that the instruc- experiment these were: for waking-hypnosis, tions were given verbally by a hypnotist who r = .65; for imagination-hypnosis, r = .66; conceivably might have unwittingly biased for hypnosis-hypnosis, r = 37. the results. The experimenters were of course Although the essential information is given aware of this danger, and sought to surmount in the simple analyses of the data as reported, it by reading all instructions verbatim and it is possible to make a somewhat more by changing experimenters from one day to elegant analysis through the use of an over- the next, the experimenter lacking knowledge all analysis of variance, in order to find an of the subject's performance on the prior interaction term. By the use of Case 15 day. A control experiment showed, however, (McNemar, 1962)) the results noted at the that the experimenter's voice was demonstra- bottom of the table are arrived at, namely, bly different after he had spent 10 minutes in that there is no significant overall treatment inducing hypnosis than after he had merely effect, but there is a significant gain overall read off very short waking or imagination with hypnotic induction on Day 2 (p < .001) instructions (Troffer & Tart, 1964). While RESPONSIVENESS

@this may not have made any difference, a responses should be made so that this experimenter second experiment was designed with tighter could record the response. The role of the assist- ing experimenter was minimal, conducting the sub- controls, in which this flaw was eliminated. ject to the room, occasionally holding a bottle under This second experiment will be described his nose, or preparing something for him to look at, before attention is directed to the state turning on the tape recorder, keeping score. The reports from both experiments. experimenter never questioned the subject, except for an interrogatory at the very end of the second day. The corresponding wording of instructions for those in the condition calling for Imagination- Expectation of Hypnosis (IE) was: Today you will be given tests of imagination while you are relaxed. It has generally been found Instructions and procedures. While the main new that exercising your imagination strongly in these control in this experiment was the use of taped experimental tasks produces a hypnotic state, even instructions and suggestion tests, with the same though we don't go through the formalities of master tape throughout, to circumvent any possible inducing hypnosis. The better you imagine, the biasing hypnotist effect, several other changes were more you'll respond. Try as hard as you can to made in order to obtain additional information concentrate and to imagine that the things I tell about the details of what goes on within a waking- you are true. imagination session as compared with a usual hyp- notic one. Three different instruction conditions were These subjects were also informed whether or not used in various combinations. These included, first, they would receive a formal hypnotic induction on an imagination condition without expectation of the second day. hypnosis (hereafter designated INE, for Imagination- Those who were to receive the formal hypnotic No-Expectation) , second, an imagination condition induction (condition HYP)were told: in which the expectation was aroused that this would lead to a hypnotic state (designated IE, for Today you will be hypnotized and given a number Imagination with Expectation), and, third, the usual of hypnotic tests. It has generally been found that standard hypnotic induction of SHSS, Form C exercising your imagination during the hypnotic (designated HYP). The uninstructed waking condi- induction and throughout the whole experiment tion of Experiment I was dropped, and no effort today will induce and maintain a hypnotic state. was made, beyond the initial instructions, to prevent The better you imagine, the more you'll respond. a spontaneous drift into hypnosis. State reports were Try as hard as you can to concentrate and to a called for as before, the voice on the tape requesting imagine that the things I tell you are true. them at the beginning and end of each suggestion. These instructions were then followed by the Another change was introduced, in that the time standard induction of SHSS, Form C. allowed for the subject "to exercise his imagination" In order to control for any of the "holding back" was equal to the time spent in hypnotic induction, that is attributed to hypnotic subjects when they so that, if relaxed immobility for 10 minutes or so know that a hypnotic session is to follow a control is important, it was equally there for the "waking" one, the control subjects, .who had nonhypnotic conditions and for the "hypnotic" ones. tests on both days, were told clearly that they The exact wording of the instructions for those were in control groups and would not be hypnotized who were in the Imagination-No-Expectation of on either day. No deception was practiced; no Hypnosis condition (INE) was as follows: subject who was told he was not to be hypnotized Today you will be given tests of imagination while was hypnotized on either session. Those who were you are relaxed. It has been generally found, not hypnotized on the first day, but were hypnotized however, that exercising your imagination strongly on the second, were told that they would be hypno- in these experimental tasks will allow you to tized on the second day. By making full disclosure I experience the things I say quite vividly, even the subject's "guessing" what would happen was though you have not been hypnotized. The better avoided. It was thus possible to compare for each \ condition what happened when there was expecta- I you imagine, the more you'll respond. Try as hard 1 as you can to concentrate, and to imagine that tion of either uniformity or change of conditions on the things I tell you are true. the next day. The total design can be summarized I most simply by a table describing the groups and In addition, the subjects were told whether or not treatment conditions. they would receive a formal induction on the second The design can most simply be understood as three day. Separate tapes were prepared for each condi- subexperiments in which pairs of groups have tion, but where the instructions were uniform, a identical conditions on the first session. For one duplicate of the original master tape was used, so member of the pair the condition is repeated on that there would not be variations in the voice, even the second day as a control for practice effects and though the same voice (that of E.R.H.) was used expectations; the other group has another condition throughout. An assisting experimenter (not E.R.H.) on the second day. Other pairings of conditions was always present, and the tape indicated when could of course be made, but these seemed to be 201 ERNESTR. HILGARDAND CHARLEST. TART

TABLE 2 sizes of these groups have been increased to C DESIGNOF EXPERIMENT11 30 subjects each. Barber and Calverley

I (1963) also found some differences of this Group 1 First session I Second session kind; when groups are large enough, and 1. INE-INE Imagination, no ex- Imagination, no ex- differences great enough, a less sensitive (N = 15) pectation of pectation of hypnosis hypnosis method can of course detect them. 2. INEHYP Imagination. no ex- Hypnotic induction Returning now to an analysis in which the (N = 15) pectation of hypnosis subjects serve as their own controls one ex- 3. IEIE Imagination, expecta- Imagination, ex- pects to find more significant differences, al- (N = 15) tion of hypnosis pectation of hypnosis though many of the comparisons will be made 4. IE-HYP Imagination. ex- I Hypnotic induction with groups of 15 subjects instead of 30 (N = 15) pectation of hypnosis as in Table 3. 5. HYP-HYP Hypnotic induction Hypnotic induction First to be considered is a control com- (N = 15) 6. HYP-INE Hypnotic induction Imagination, no ex- parison, in which no differences are to be (N = 15) pectation of expected. These are the results for those hypnosis groups that had the same conditions on each of two days. If there are differences they will the most instructive. There were 15 subjects in each represent practice effects, or perhaps slight condition, 90 in all, from classes in introductory differences between the two forms of the test. psychology. When the experiment was entirely completed, The results are given in Table 4. The analysis many of the subjects who had served in the control of variance shows all effects to be nonsignifi- groups expressed disappointment at not having had cant. These results permit acceptance with the experience of hypnosis, and requested return greater confidence of the significant changes to the laboratory. This provided the circumstances to invite back for a third day all of the subjects to be attributed in subsequent analyses to who had been in the control groups, offering a new treatment effects. They also establish the opportunity for hypnosis without violating the comparability of the suggestibility forms used agreements made with them during the experiment on the 2 days. The unusually' low mean proper. At the same time, the opportunity was taken hypnotic scores of Group 3' illustrate what to invite back also those who had had hypnosis only, in order to test them under imagination condi- was said earlier about chance effects with tions. Responses to these postexperimental invita- small groups; this group of 15 subjects hap- tions were accepted by 18 of the 30 subjects who pened to lack any highly susceptible subjects. had been in imagination groups on both days, and The low scores are not attributed either to by 6 of the 15 subjects who had been in hypnosis groups both days. Their results will be considered the use of taped inductions or to expecta- after the results of the experiment proper have been tions owing to having hypnotic induction on reported. Results of Experiment 11. The first task TABLE 3 ~ is to inspect the results as they would be were u 1 --.,A is, AND the data from the first day only to be con- CNT 11 sidered, with 30 subjects in each of the three conditions. This is not a favorable type of Scores on suggestion test Condition (Modified Form C. comparison, for the reasons indicated earlier, maximum = 10) and will detect differences only if they are Imagination, no expectation AT = 30 more substantial than those found in Experi- of hypnosis (INE) M = 2.58 ment I. The results are presented in Table 3. SD = 2.27 Contrary to expectations, there is a sig- Imagination, ex ectation of N = 30 nificant treatment effect, an analysis of vari- hypnosis (IE~ M = 3.37 = ance showing that group effects are significant SD 2.12 I between p = .O1 and .025. The effects are not Ilypnotic induction (HYP) AT = 30 striking, the Duncan multiple-range test M = 4.43 showing no difference between neighboring SD = 2.75 groups, but a significant difference between Note.-Results of analysis of variance: For Groups, F = 4.52, the extremes. It should be noted also that the .025 < p < .01. 1 RESPONSIVENESSTO SUGGESTIONS

TABLE 4

Scores on suggestion tests (Maximum = 10) Difference: change from first to second session First session Second session Group 1' Imagination, no expectation of Imagination, no expectation of (N = 15) hypnosis hypnosis M 2.37 M 2.33 M -.04 SD 1.99 SD 1.71 (1 test, ns) Group 2' Imagination, expectation of Imagination, expectation of (N = 15) hypnosis hypnosis M 2.73 M 3.23 M +.SO SD 1.99 SD 1.81 (1 test, ns) Group 3' Hypnotic induction Hypnotic induction (N = 15) M 3.83 M 3.77 M -.06 SD 2.91 SD 2.80 (1 test, ns)

Note.-Results of analysis of variance: Groups effect, ns; session effects, ns; interaction, Groups X Sessions, ns. Two-day correlations: Group 1' (INE versus INE), r = .65 (N = 15); Group 2' (IE versus IE), r = .40 (N = 15); Group 3' (HYP versus HYP).r = .98 (N = 15). both days. Expectations are no different from explanation of the low mean hypnotic score those subjects in Group 3 of Experiment I of Group 3'. (Table I), who averaged a point higher. The The main results are found in the taped induction can scarcely be blamed be- relationships of Table 5. The analysis of cause the other 15 subjects of Experiment I1 variance shows the interaction effects of who were hypnotized by tape on the first groups and sessions to be highly significant. day averaged 5.03 compared with the first (p < .005), which means that the gains from day average of 4.90 for the 20 subjects of imagination conditions to hypnotic induction Experiment I, in which induction was live. (and the loss when hypnotic induction comes A random effect in the composition of a first) have been demonstrated. The t tests small sample of subjects remains the best for each group are satisfactory, particularly

TABLE 5

Scores on suggestion test (Maximum = 10) Difference: change from first to second session Fit session Second session Group 1 Imagination, no expectation of Hypnotic induction (N = 15) hypnosis M 2.80 M 3.70 M +.90 SD 2.57 SD 3.08 (t test, p <.025) Group 2 Imagination, expectation of Hypnotic induction (N = 15) hypnosis M 4.00 M-5.13 M +1.13 SD 2.11 SD 2.42 (t test, p <.025) Group 3 Hypnotic induction Imagination, no expectation of (N = 15) hypnosis M 5.03 M 3.53 M -1.50 SD 2.52 SD 2.38 (t test, p <.025)

Note.-Results of analysis of variance: groups ?as. sessions ns' interaction, Groups X Sessions 9 < 005. 2-day comda- tions: Group 1 (INE versus HYP), r = .86 (N = 15)'; G&UP 2 (I$ ve;(sus HYP), r = .72 (N = 15); ~r&3 ~HYPversus INE), r = .52 (N - 15). ERNESTR. HILGARDAND CHARLEST. TART

TABLE 6

Scores on suggestion test (Maximum = 10) Difference: change from second to third Second session (repeating session conditions of first) Third session (unplanned) INE-INE-HYP Imagination, no expectation of Hypnotic induction (N = 8) hypnosis M 2.75 M 4.00 M +1.25 SD 0.60 SD 1.73 IE-IEHYP Imagination, expectation of Hypnotic induction (N = 10) hypnosis M 3.55 M 3.10 M -0.45 SD 1.83 SD 2.20 HYP-HW-INE Hypnotic induction Imagination, no expectation of (N = 6) hypnosis M 5.00 M 2.42 M -2.58 SD 3.26 SD 1.71

Note.-: test, all hypnotic induction scores minus all imagination scores. $ < .05. in view of the small numbers of subjects in be very desirable, in experiments of this sort, each group. Within each day the treatment to be able to compare results only for sub- means are not significantly different, reflect- jects capable of influence by hypnotic induc- ing the smaller sample size as compared with tion. The state reports are of help in this the significant first-day effects presented in respect, because they are relatively inde- Table 3. pendent of the responses to the various tests The additional data from the subjects who of suggestion. An attempt will be made to returned for a third day suggest another show that they do indeed provide infonna- comparison: What happens when an unantici- tion that is supplementary, and throw addi- (" pated third session provides a change from tional light upon what is happening in the uniformly experienced 2 days of the experiments of this kind. regular experiment? Although the groups are Reliability of state reports. Within any 1 small, the overall results are significantly in day there were a number of state reports the same direction as the main findings obtained, so that a,mean state report could (Table 6). The only reversal is a failure of be computed for each subject with a value those who had had 2 days of imagination between 0 and 3. The reliabilities of such with expectation of hypnosis to improve reports are shown by the 2-day correlations under hypnotic induction, but this result, of Table 7, in which all groups with the same with 10 subjects only, is too slight to destroy condition on 2 days have been studied. the mean effect for the total of 24 subjects For the hypnotized groups, these are in the participating in this third session. .go's; while for the nonhypnotized groups The results of the second experiment con- they are lower, they are still in the .80Js. firm those of the first: In all comparisons, Any spread in scores for the nonhypnotized with the minor exception just noted, hypnotic induction produces an increase in response TABLE 7 over noninduction conditions, a gain which, RELIABILITYOF STATEREPORTS in its overall effect, meets acceptable sta- Common treatment Correlation between tistical standards of significance. on 2 days mean state reports Experiment 1 HW-HYP .91 (N = 20) Experiment 2 HYP-HYP .99 (N = 15) IE-IE .83 (N = 15) In view of the large number of subjects INE-INE .8l (N = 15) insusceptible to hypnotic induction it would TABLE 8 subjects can destroy the correlation in a CORRELATIONBETWEEN STATE REPORTS AND OBJECTIVE small sample of this kind. The other positive SUGGESTIONSCORES, ALL CONDITIONS correlations may be taken as much more typical, based as they are on large samples. Correlation between mean state report The positive correlations between state re- Subject group and suggestibility scores ports and responsiveness to suggestions raise a ticklish problem. May it not be that the Experiment I Waking (N = 40) (first day) .22a state reports are themselves reflections of Imagination (N = 20) (first day) .67 the fact that the subject finds himself re- Hypnosis (N = 80) (second sponding to suggestions? The answer cannot day only) .68 be a straightforward one, for suggestions of Experiment I1 drowsiness, and so on, if responded to, in part Imagination, no expectation of hypnosis (INE) (N = 45) characterize the "state" that is being re- (first day with this condition) .65 ported. But there are some data that show Imagination, expectation of that this is not simply an alternative form of hypnosis (IE) (N = 15) (first day only) -.01 report on suggestibility, based on the sub- Hypnosis (N = 60) (first day ject's self-scoring of his responses within the with this condition) .75 suggestibility tests. .This is attenuated by the fact that the subject was aroused The first point to be made is that the if a state report above 0 was given. Even so, enough sub~ects yielded such state reports often enough to produce the small initial state report, following waking instruc- correlation. tions, imagination instructions, and hypnotic induction, tends to persist throughout the ses- groups means, of course, that some have sion, despite the subject's later responsiveness drifted into a borderline state or one that or lack of responsiveness to the suggestions. they were willing to characterize as similar to The facts for the first day of experimentation hypnosis; otherwise there would be no vari- are summarized in Table 9, which combines ance and no correlation. the results for both experiments. It is evident State score a functia of instructions and that the initial report is highly predictive of inductions, not of responsiveness to suggesti- later ones: x2 = 153.89 with 4 df, p < .001, bility to test items. There is usually a correla- and the contingency coefficient (Siegel, tion between the state report and responsive- 1956) is .69. Of those who start with but a ness to suggestions, as shown in Table 8. The report of 0 (normal aroused waking state), one failure of correlation, that for the IE 92 of 97 average below 1.0 in their later group, is puzzling, but the number of cases reports, and none averages as high as 2.0. is small (N = 15) and it happens that two Correspondingly, of those who initially report subjects giving high state reports were very 2 or 3 (hypnotized) 17 of 23 remain there, low in their suggestibility scores. Such deviant and none averages below 1.0. Thus, despite the correlation between the state report and TABLE 9 the later suggestibility measures, the state report does not appear to depend upon the self-observation of these later experiences. The second point is that the statistical regression may differ by conditions, even Mean state report within day 1 (disregarding condition) when the correlations between state report Initial state report and suggestibility scores are nearly alike. 0-0.9 1.0-1.9 2.0-2.9 Total Useful comparisons from Experiment I are ---- available, in which the state reports correlate 0 92 5 0 97 1 15 29 6 50 with suggestibility .67 for the imagination 2-3 0 6 17 23 group and .68 for the hypnosis group, and 107 40 23 170 another comparison from Experiment 11, in which the correlations are .65 for the INE Note.-x' = 153.89, p < .001, 4 df. C = .69 (contingency coefficient). group and .75 for the HYP group. When the 2 05 ERNESTR. HKGARDAND CHARLEST. TART regression lines are plotted in raw score form, TABLE 10 it becomes evident that for an equal amount of suggestibility response there is a lesser state report for the nonhypnotized than for the hypnotized subjects (Figure 1). This argues also for some separateness between the Suggestibility scores (Modified judgments of state and those of responsive- Form C) ness to suggestions. Initial state report, prior to suggesti- Imagination Hypnotic The state reports furnish a somewhat crude bility tests instructions induction (Either INE or IE) (HYP) but useful answer to the need for evidence that a person is hypnotized, apart from re- NMSD NMSD sponsiveness on tests of suggestibility, for, as 3-very hypnotized 0 - - 5 6.9 1.9 just indicated, there is no absolute corre- 2-hypnotized 5 5.0 2.2 21 5.7 2.5 spondence between hypnotic state and respon- 1-drifting 38 3.5 2.2 43 4.0 2.2 siveness to standardized suggesticnls. If one 0-normally awake 38 2.2 0.8 12 1.3 1.4 were willing to accept the state measure Totals 81 3.0 1.5 81 4.2 2.1 following either waking instructions or hyp- The groups of 81 subjects have 63 subjects who served in notic induction as a sorting measure, elimi- both conditions. If a subject served in the same condition nating those from the waking group who twice, the first experience only was used. feel hypnotized despite lack of induction, and from the hypnotic group those who do that it would make is presented in Table 10 not feel hypnotized despite the attempted which shows the suggestibility scores follow- induction, the obtained differences between ing imagination instructions and following. waking and hypnosis would be greatly en- hypnotic induction of those who gave vari hanced. This is a tricky matter, of course, state reports at the end of the instruction because of the extent to which the state induction period but before the tests of s report is itself a response to suggestions. gestion were made. It can be seen that those Yet some such sorting basis would be highly who felt themselves unaffected by the instruc- desirable. An illustration of the difference tions or hypnotic induction (state report 0) scored at the bottom of the group, whether or not there had been attempted hypnotic induction. On the other hand, those five sub- jects who gave state reports of 2 ("hypno- tized") following imagination instructions yielded suggestibility scores equaling those of the subjects who gave the same report follow- ing a standard hypnotic induction. Thus in group comparisons in which various alterna- tives to induction are used, it may be pre- sumed that some subjects actually become hypnotized. The similar means for responsive- ness to suggestions after equivalent initial state reports are striking; of course the num- bers of cases in each category reflect the greater effectiveness of the hypnotic induc- ~0123456789lOtion. MEAN SUGGESTlON TEST SCORES The relationsJiip between waking scores FIG. 1. Regression lines for state reports according and hypnotic scores for idividuds. Informa- to degree of responsiveness to suggestions. (Note tion about individuals is easily lost within that despite similar correlations, higher suggestibil- pooled data. For example, the relatively high ity scores are associated with lower state reports under imagination conditions than under hypnotic correlations between waking and hypnotic conditions.) performances indicate that they have some- RE~PONSIVENE~S v thing in common, but they do not reveal any- absent quadrant is the lower-righthand one, thing about individual departures from this which means that those high in waking sug- generalization. This information is best ob- gestion are never found low in hypnotic sug- tained from a scatterplot of the scores fol- gestibility. This is a statistical justification lowing waking (or imagination) instructions for selecting good hypnotic subjects by means and following attempted hypnotic induction. of waking hypnosis tests; some good subjects Throwing together for this purpose all sub- will be missed, but no very poor ones will jects who have contributed one score under be chosen. Those whose suggestibility is un- a waking or imagination condition and an- affected by hypnosis lie in the diagonals; as other following hypnotic induction, yields a expected, more of these are poorer subjects total of 129 subjects from the two experi- (20 scoring &4), but there are the 11 high- ments. A scatterplot of the relationship be- scoring ones also (scoring 5-10). tween waking response and hypnotic response Many of the contradictory statements is given in Figure 2. The overall correlation about the effectiveness of hypnotic induction is r = .63; the plot is not significantly curvi- can be understood in the light of Figure 2. linear, the value of eta being .67, not signifi- Depending upon the manner in which sub- cantly different from r. Even so, the scatter- jects are selected, and what kind of statistical plot is somewhat triangular, with cases falling treatment is employed, it is easy to demon- in three of the four quadrants as shown. strate either that induction has a substantial What this means is that many who fall into effect or almost none at all. The overall find- the low-responsive group in tests of waking ing is, of course, that the effect of hypnotic suggestions are indeed quite hypnotizable fol- induction is very real for a number of sub- lowing a standard induction. These are the 34 jects. Thus for the 99 subjects who score cases in the upper-left quadrant. There are 0-4 in waking or imagination, 35 gained also a substantial number who are quite under hypnosis, many of them substantially, susceptible in the waking state, though many 20 remained the same, and 12 decreased a of them go further under hypnosis (the 30 point or two. cases in the upper right quadrant). It is + DISCUSSION of interest that 11 of these scored higher under hypnosis, 11 of these scored as high The results support the usual belief that in waking as in hypnosis, and 7 scored following hypnotic induction there is a sta- slightly lower in the hypnotic test. The tistically significant increase in responsiveness to suggestions of the kind familiar within hypnotic lore, although the amount of this

increase, in agreement with Hull's early find- , ings, is small. The reasons for the small aver- age gains have been clarified by showing that many low-scoring subjects change very little as a consequence of induction, while others yield high scores on waking suggestion tests, and hence also gain little under hypnosis. Unless methods of investigation permit the use of the subject as his own control, these small gains are difficult to detect, and hence may lead to the conclusion of no difference, the conclusion frequently reached in recent publications by Barber and his associates. Treatment differences reached significance for

SUGGESTIBILITY MLLOWNG WAKING OR IMAGINATION INSTRUCTIONS randomly selected groups in only one of the comparisons in the present experiment FIG. 2. Scatterplot of suggestibility scores following waking or imagination instructions. (The scores all (Table 3) in which each of the three groups fall within three of the four quadrants.) contained 30 subjects, and even then one of the imagination groups was not significantly standable according to the order-of-treatments different from the hypnotic group. When sub- hypothesis. The fact that their best is so rela- jects tested under these same imagination tively poor (mean = 3.55)) weakens this in- conditions were later tested following hypnotic terpretation. Probably the most telling injunction, they went on to a significant in- argument against the order-of-treatments hy- crease in their scores. The basic data are pothesis is that from the correlation between all therefore consistent with the findings state reports and responsiveness to sugges- of others, but the evidence refutes the tions. Those subjects who felt themselves to interpretation of lack of influence of hypnotic be hypnotized were the ones, regardless of induction. instructions, or order of treatment, who The most serious criticism of experiments yielded the higher suggestibility test scores. in which subjects serve as their own controls To argue that they felt hypnotized only be- is that they may temper their performances cause they expected to become hypnotized is in waking conditions in order to enhance merely to assert that there is in fact some them in hypnosis, as in the experiments of responsiveness to suggestion involved in hyp- Scharf and Zamansky (1963) and Zamansky, nosis; to deny responsiveness to suggestion Scharf, and Brightbill (1964). The current would make the procedure of hypnosis mean- authors' experimental arrangements permit a ingless. This constraint is not implied in number of comparisons to test the presence looking for effects due to order of treatments. of this order-of-treatments effect. If it is A serious problem is raised in making re- present, it is not clearly demonstrable, for sponsiveness to suggestion a criterion of hyp- individual differences in susceptibility across nosis, for it is evident that some subjects, not groups are far more important. For example, hypnotized, can yield high responses to sug- the means in Table 5 can be studied for these gestion. The state reports show that this is not effects. Consider Group 1, with imagination too serious a matter because of the substan- in the first session, hypnosis in the second, as tial correlations between state reports and compared with Group 3, with hypnosis first responsiveness to suggestions, even when the and imagination second. The hypothesis is state reports are made after induction, and fi that the imagination group would hold back before the suggestion tests are administered. on the first day for Group 1, probably not The fact that the regression lines are dis- on the second day for Group 3. The means tinguishable, even for correlations of about are in that order: 2.80 for Group 1 and 3.53 the same size (Figure 1))means that there is for Group 3. But the same argument would a difference, however, between the subject hold that both groups should enhance their responding to suggestions following hypnotic hypnotic performances. It turns out that induction and without such an induction. It Group 3 has a mean for hypnosis of 5.03 and would be desirable to develop some measure Group 1 a mean of 3.70, a little higher than of the extent to which a person is actually the imagination mean for Group 3. An ex- hypnotized other than measures of his respon- planation more plausible than the holding- siveness to standard suggestions; it is quite back hypothesis is that the subjects of possible that in the deeper stages of hypno- Group 1 were merely less susceptible than sis responsiveness to some suggestions will those of Group 2; otherwise there is no reason disappear, without this meaning any weak- for their low hypnosis scores on the second ening of the hypnosis. The ordinary tests of day. In Table 6 there is one comparison that responsiveness to suggestions following stand- might fit the expectation hypothesis, the re- ard hypnotic inductions are still quite satis- sults for the IE-IE-HYP group. On the factory as measures of individual differences, assumption that these subjects, expecting and their validity as measures of individuality hypnosis through imagination, and not ex- is supported by the correlations between wak- pecting to be formally hypnotized at all, gave ing and hypnotic responsiveness as well as by all they had in their 2 days, then the the correlations between responsiveness on I failure to gain when brought back for hyp- these tests and state reports. This does not nosis on an unexpected third day is under- preclude further refinements, however, that would go beyond suggestibility, and might LECRON,L. M. A method of measuring the depth get at some of the other important aspects of of hypnosis. Journal of Clinical and Exfietimental HyMosis, 1953, 1, 4-7. the hypnotic state. MCNEMAR, Q. Psychological statistics. (3rd ed.) New York: Wiley, 1962. REFERENCES SCHARF,B., & ZAMANSKY,H. S. Reduction of word- BARBER,T. X., & CALVERLEY,D. S. "Hypnotic be- recognition threshold under hypnosis. Perceptual havior" as a function of task motivation. Journal 8 Motor Skills, 1963, 17, 499-510. . of Psychology, 1962, 54, 363-389. SIEGEL,S. Non-parametric statistics for the behav- BARBER,T. X., & CALVERLEY,D. S. The relative ioral sciences. New York: McGraw-Hill, 1956. effectiveness of task-motivating instructions and TART, C. T. Hypnotic depth and basal skin resist- trance-induction procedures in the production of ance. International Journal of Clinical and Ex- "hypnotic-like" behaviors. Journal of Nervous perimental Hypnosis, 1963, 11, 81-92. and Mental Disease, 1963, 137, 107-116. TROFFER,S. H., & TART,C. T. Experimenter bias in BARBER,T. X., & GLASS,L. B. Significant factors in hypnotist performance. Science, 1964, 145, 1330- hypnotic behavior. Journal of Abnormal and So- 1331. cial Psychology, 1962, 64, 222-228. WEITZENHOFFER,A. M. Hypnotism: An objective HATFIELD,E. C. The validity of the LeCron method study in suggestibility. New York: Wiley, 1953. of evaluating hypnotic depth. International Jour- WEITZENHOFFER,A. M., & HILCARD,E. R. Stanford nal of Cliniccal and Experimental Hypnosis, 1961, Scale, Form C. Palo Alto, 9, 215-221. Calif.: Consulting Psychologists Press, 1962. HILCARD,E. R., WEITZENHOFFER,A. M., LANDES,J., WEITZENHOFFER,A. M., & SJOBERC,B. M., JR. Sug- & MOORE,R. K. The distribution of susceptibility gestibility with and without "induction of hyp- to hypnosis in a student population: A study using nosis." Journal of Nervous and Mental Disease, the Stanford Hypnotic Susceptibility Scale. Psy- 1961, 132, 204-220. chological Monographs, 1961, 75 (8, Whole No. ZAMANSKY,H. S., SCHARF,K, & BRIGHTBILL,R. The 512). effect of expectancy for hypnosis on prehypnotic HULL, C. L. Hypnosis and suggestibility: An ex- performance. Journal of Personality, 1964, 32, perimental approach. New York: Appleton-Cen- 236-248. tury, 1933. (Received May 26, 1965) - would go beyond suggestibility, and might LECRON,L. M. A method of measuring the depth get at some of the other important aspects of of hypnosis. Journal of Clinical and Experimental Hypnosds, 1953, 1, 4-7. the hypnotic state. MCNEMAR, Q. Psychological statistics. (3rd ed.) New York: Wiley, 1962. REFERENCES SCHARF,B., & ZAMANSKY,H. S. Reduction of word- BARBER,T. X., & CALVERLEY,D. S. "Hypnotic be- recognition threshold under hypnosis. Perceptual havior" as a function of task motivation. Journal & Motor Skills, 1963, 17, 499-510. of Psychology, 1962, 54, 363-389. SIECEL, S. Non-parametric statistics for the behav- BARBER,T. X., & CALVERLEY,D. S. The relative ioral sciences. New York: McGraw-Hill, 1956. effectiveness of task-motivating instructions and TART, C. T. Hypnotic depth and basal skin resist- trance-induction procedures in the production of ance. International Journal of Clinical and Ex- "hypnotic-like" behaviors. Journal of Nervous perimental Hypnosis, 1963, 11, 81-92. and Mental Disease, 1963, 137, 107-116. TROFFER,S. H., & TART,C. T. Experimenter bias in BARBER,T. X., & GLASS,L. B. Significant factors in hypnotist performance. Science, 1964, 145, 1330- hypnotic behavior. Journal of Abnormal and So- 1331. ci~.lP~ychol~gy, 1962, 64, 222-228. WEITZENHO~R,A. M. Hypnotism: An objective HATFIELD,E. C. The validity of the LeCron method study in suggestibility. New York: Wiley, 1953. of evaluating hypnotic depth. International Jour- WEITZENHO~R,A. M., & HILCARD,E. R. Stanford nal of Clinical and Experimental Hypnosis, 1961, Hypnotic Susceptibility Scale, Form C. Palo Alto, 9, 215-221. Calif.: Consulting Psychologists Press, 1962. HILCARD,E. R., WEITZENHOFPER,A. M., LANDES,J., WEITZENHOFFER,A. M., & SJOBERG,B. M., JR. Sug- & MOORE,R. K. The distribution of susceptibility gestibility with and without "induction of hyp- to hypnosis in a student population: A study using nosis." Journal of Nervous and Mental Disease, the Stanford Hypnotic Susceptibility Scale. Psy- 1961, 132, 204-220. chological Monographs, 1961, 75 (8, Whole No. ZAMANSKY,H. S., SCHARF,B,, & BRIGHTBILL,R. The 512). effect of expectancy for hypnosis on prehypnotic HULL, C. L. Hypnosis and suggestibility: An ex- performance. Journal of Personality, 1964, 32, perimental approach. New York: Appleton-Cen- 236-248. tury, 1933. (Received May 26, 1965)