Temporal Sequencing During Posthypnotic Amnesia - Semantic Scholar
Copyright 1981 by the American Psychological Association, Inc. 002I-843X/81/9005-0476S00.75
Journal of Abnormal Psychology 1981, Vol. 90, No. 5, 476-485
Temporal Sequencing During Posthypnotic Amnesia: A Methodological Critique H. Lorraine Radtke and Nicholas P. Spanos Carleton University, Ottawa, Ontario, Canada In a series of studies using the susceptibility-scale paradigm, high susceptible subjects were less likely than low susceptibles to recall the events of the hypnotic session in temporal sequence (i.e., temporal disorganization effect) following an amnesia suggestion. The primary measure of recall order was the rank-order correlation (rho scores) between the presentation order and the recall order of hypnotic experiences computed for each subject. Following a suggestion for posthypnotic amnesia, high susceptibles usually obtained lower rho scores than low susceptibles. We critically examined this research noting methodological shortcomings associated with the susceptibility-scale paradigm, inconsistent findings, and failures to replicate. We also presented two studies that found no relationship between susceptibility level and rho scores. These null results held true for subjects who recalled new information after cancellation of the amnesia suggestion (reversers) as well as for those who did not recall new information (nonreversers). Nevertheless, we did replicate previous work on differential recall of the first item. Suggestions for the development of a more adequate paradigm to assess recall order during amnesia using the susceptibility-scale paradigm are discussed.
In 1973, Evans and Kihlstrom published the first in an influential series of articles on the organization of recall during posthypnotic amnesia (Evans & Kihlstrom, 1973; Kihlstrom, 1975; Kihlstrom & Evans, 1979). This line of research has frequently been credited with demonstrating a reliable association between posthypnotic amnesia and a breakdown in the temporal organization of recall (see reviews by Bowers, 1976; Cooper, 1979; Evans, 1980; Hilgard, 1977; Kihlstrom, 1975, 1977, 1978; Kihlstrom & Evans, 1979; Maslach, 1979). Moreover, the temporal disorganization effect has served as the major empirical underpinning for a disrupted retrieval hypothesis of posthypnotic amnesia (Evans & Kihlstrom, 1973) and more recently for a dissociation hypothesis of amnesia (Kihlstrom, 1977). Nonetheless, close examination of the exThis research was supported by a Social Sciences and Humanities Research Council doctoral fellowship awarded to the first author and Natural Sciences and Engineering Research Council Operating Grant A6385 awarded to the second author. We thank H. J. Stam for critically reading an earlier version of this manuscript. Radtke is now at the Dept. of Psychology, University of Manitoba, Canada. Requests for reprints should be sent to Nicholas Spanos, Department of Psychology, Carleton University, Ottawa, Ontario, Canada K1S 5B6.
perimental paradigm employed in these studies (hereafter called the susceptibilityscale paradigm) reveals a number of serious methodological problems. Moreover, one recent study was unable to replicate the major findings (St. Jean & Coe, 1981). As a result, it is questionable whether a relationship between lack of temporal organization and hypnotic amnesia has in fact been demonstrated. The present report critically reviews this work and also presents two attempts to replicate the temporal disorganization effect using the susceptibility-scale paradigm. The susceptibility-scale paradigm assesses subjects' performance on the amnesia suggestion contained in standardized scales of hypnotic susceptibility. The scales employed have included the Harvard Group Scale of Hypnotic Susceptibility: Form A (HGSHSIA; Shor & Orne, 1962) and the Stanford Scales of Hypnotic Susceptibility, Form B (SHSS:B; Weitzenhoffer & Hilgard, 1959) and Form C (SHSS:C; Weitzenhoffer & Hilgard, 1962). On all of these scales the 12th and final test suggestion calls for posthypnotic amnesia. Subjects are told that after they are awakened they will be unable to remember anything that occurred during hypnosis until they hear a prearranged cue (i.e., "Now you
476
TEMPORAL SEQUENCING DURING POSTHYPNOTIC AMNESIA
can remember everything") that cancels their amnesia. After being awakened, but before presentation of the cancellation cue, subjects are asked to write down (or report orally, depending on the scale) everything that happened to them from the time of the hypnotic induction procedure (R-Trial 1). Three minutes later the suggestion is cancelled with the prearranged cue, and subjects are given two minutes to report anything else they now remember (R-Trial 2). Evans and Kihlstrom (1973) compared the order in which test suggestions were presented on the susceptibility scales with the order in which these suggestions were recalled by subjects on R-Trial 1. Three indices of association between presentation order and recall order were obtained for each subject. The most important index involved computation of a Spearman rank-order correlation coefficient (rho score) between presentation order of the test suggestions and recall order on R-Trial 1. A second index involved classifying each rho score as statistically significant or nonsignificant. Recall was considered to be ordered (i.e., organized) if the rho score was positive and significant (p < .05, one-tailed) and random (i.e., disorganized) in every other case. The third index simply assessed whether the first item recalled by a subject was also the first suggestion presented on the scale in question. Typically, subjects were classified as high, medium, or low in hypnotic susceptibility on the basis of their scores on the susceptibility scales, and relationships between susceptibility level and the three indices of association were assessed. Results for each index will be described separately. Rho Score Index In their initial study Evans and Kihlstrom (1973) tested subjects on the HGSHS:A, SHSS:B, and SHSS:C. Separate one-way analyses of variance (ANOVAS) using high, medium, and low susceptibility levels as the independent variable and each subject's rho score as the dependent variable were performed on the data from each scale. The ANOVAS were significant for the SHSSIB and SHSS:C but nonsignificant for the HGSHS:A. The means involved in the two significant
477
ANOVAS, however, were not consistently ordered. In the case of the SHSS:C, medium susceptibles had a lower mean rho score than low susceptibles and, on the SHSS:B, mediums had a lower mean rho score than either highs or lows. Differences involving the means for the medium susceptibles were not tested for statistical significance. However, the rho scores for high and low susceptibles were compared with multiple one-tailed t tests. High susceptibles were found to have significantly lower rho scores than low susceptibles on the HGSHS:A and the SHSS:C. On the SHSS:B this difference was not significant. At best, these findings weakly suggest a difference in rho scores between high and low susceptibles on the HGSHSIA and SHSS:C. By ignoring the medium susceptibles in their data analysis, Evans and Kihlstrom (1973) weakened the explanatory power of their hypothesis. The temporal disorganization effect was replicated by Kihlstrom (1975) using the HGSHS:A. Recently, however, St. Jean and Coe (1981) attempted unsuccessfully to replicate it. They found no significant differences in the rho scores of high, medium, and low susceptibles on either the SHSS:B or the SHSS:C.
Ordered/Random Recall Index In three samples, subjects were divided into those whose rho scores were either random (i.e., nonsignificant) or ordered (significant at p < .05, one-tailed). Subjects in the first sample had been administered the HGSHS:A, SHSS:B, and SHSS:C (Evans & Kihlstrom, 1973), those in the second sample had been given the HGSHS:A and the SHSS:C, and those in the third sample were tested on the HGSHS:A alone (Kihlstrom & Evans, 1979). Subjects in Samples 1 and 2 were classified as high, medium, or low in susceptibility and chi-square analyses were used to assess the degree of association between susceptibility level and random/ordered recall for each susceptibility scale. The three chisquare analyses were significant for Sample 1. For Sample 2, the analysis was significant for the SHSSiC but not for the HGSHS:A. According to Kihlstrom and Evans (1979), these generally significant effects indicate
478
H. LORRAINE RADTKE AND NICHOLAS P. SPANOS
that a greater proportion of high susceptibles than low susceptibles showed random recall when given an amnesia suggestion. It should be noted, however, that in four out of five analyses the proportion of medium susceptibles and high susceptibles showing random and ordered recall did not differ significantly. Also, in Sample 1 on the SHSS:B and SHSS:C medium susceptibles showed proportionately more random recall than lows.1 In Sample 3, highs were more likely to show random recall than lows; mediums were not tested. St. Jean and Coe (1981), however, found no significant differences in the proportion of high, medium, and low susceptibles showing random as opposed to ordered recall on either the SHSStB or the SHSS:C. Differential Recall of First Suggestion In their original study Evans and Kihlstrom (1973) reported that high susceptibles were significantly less likely than low susceptibles to recall first the initial test suggestion on the SHSS:C. No data were presented for performance on the HGSHS:A or SHSS:B or for medium susceptibles. St. Jean and Coe (1981) replicated and extended this differential first suggestion recall effect. They found that high and medium susceptibles were less likely than low susceptibles to recall the first suggestion first on both the SHSS:B and SHSS:C.
Nonrecall, Reversibility, and Organization Disorganized recall is supposedly a characteristic of hypnotic amnesia and not simply a correlate of hypnotic susceptibility (Evans & Kihlstrom, 1973). In support of this notion Evans and Kihlstrom (1973) reported that subjects who recalled three or fewer suggestions at R-Trial 1 (i.e., complete nonrecallers) were higher in susceptibility than random recallers, who, in turn, were higher in susceptibility than ordered recallers. Although this finding appears to be reliable (replicated by Kihlstrom & Evans, 1979), it is related only tenuously to the hypothesis that amnesia (as opposed to susceptibility) is associated with disorganization. An adequate test of this notion requires that amnesia be measured independently of hypnotic susceptibility. The hypothesis pre-
dicts that amnesics will show disorganized recall regardless of their level of susceptibility. Unfortunately, standardized susceptibility scales make no provision for unambiguously assessing degree of amnesia. Most investigators agree that hypnotic amnesia is a temporary recall deficit and, therefore, should be differentiated from other recall deficits in terms of its reversibility (Kihlstrom & Evans, 1976; Nace, Orne, & Hammer, 1974; Spanos & Bodorik, 1977). Thus, even a minimum requirement for a demonstration of amnesia should include the recovery of "forgotten" material after cancellation of the suggestion (i.e., information not recalled at R-Trial 1 should be recalled at R-Trial 2). This minimum requirement is not met on standardized susceptibility scales. Instead, amnesia on these scales is defined solely in terms of the number of suggestions recalled at R-Trial 1. Kihlstrom and Evans (1979) attempted to overcome this problem by devising an index of reversible recall. They compared the rho scores of subjects who recalled at least two items at R-Trial 2 that had not been recalled previously at R-Trial 1 (reversers) with those of subjects who recalled no new items at RTrial 2 (nonreversers). On the SHSS:C reversers had significantly lower rho scores than nonreversers. However, on the HGSHS:A the rho score difference between these treatments was nonsignificant (p > .05, onetailed). Thus, there is at best weak support for the hypothesis that disorganization and amnesia (as defined by reversibility) are related. Given the inconsistencies in the work reviewed above, we felt it imperative to conduct an independent replication study. We obtained two large samples. One had been tested only on the HGSHS:A, and the other had been assessed on the HGSHStA and the SHSS:C. (HGSHS:A scores, however, were available for only a subset of the second sample.) To replicate the procedures employed by Evans and Kihlstrom as closely as possible, we carried out the same statistical analyses as they and adopted the same alpha 1
These analyses are not presented in Kihlstrom and Evans (1979). We computed the chi-square on the basis of frequencies presented in that paper.
479
TEMPORAL SEQUENCING DURING POSTHYPNOTIC AMNESIA
levels. However, for analyses not conducted by Evans and Kihlstrom (1973) we conducted only two-tailed tests and set alpha consistently at .05. Although previous work has tended to ignore the performance of medium susceptibles, any theory of hypnotic behavior must surely account for the behavior of medium susceptibles (Evans & Orne, 1965). Therefore, we included data from medium susceptibles in all analyses that constituted an extension of previous work. Method Data from two samples of subjects that had been administered standard tests of hypnotic susceptibility as a screening procedure for further experiments conducted in our laboratory were available for analysis. Sample A had been tested only on the HGSHSIA in small groups of 7 to 12. The 240 subjects in this sample (93 males, 147 females) ranged in age from 17 to 49 years. Of these subjects, 54 scored 9 or above and were classified as high susceptibles, 106 scored between 5 and 8 and were classified as medium susceptibles, and 80 scored between 0 and 4 and were classified as low susceptibles.2 Sample B included 223 subjects (93 males, 130 females), aged 17 to 62 years, who had been administered the SHSS:C individually. These subjects had been administered the HGSHSIA previously in small groups of 5 to 10. However, HGSHSIA data were available only for 118(55 males, 63 females) of the total sample. All subjects were classified on susceptibility according to their scores on the SHSS:C.3 The classification criteria were as follows: (a) Highs scored 8-12, (b) mediums scored 5-7, and (c) lows scored 0-4. Of the 223 subjects tested on the SHSS:C, 105 were classified as high, 73 as medium, and 45 as low. Of the 118 for whom HCSHS:A data were available, 48 were high, 40 were medium, and 30 were low, according to their scores on the SHSS:C.
Results Rho Index Spearman rank-order correlation coefficients (rhos) were calculated for each subject who recalled three or more items. Following Evans and Kihlstrom (1973), oneway ANOVAS were performed on the rho scores. The results are summarized in Table 1. They indicate no relationship between hypnotic susceptibility and organization of recall as indexed by rho scores. Individual one-tailed / tests between high and low susceptibles revealed no difference on either scale in either sample. Subjects at each susceptibility level (high, medium, low) were then divided into those
Table 1 Summary of Rho Score Analyses With Susceptibility as the Independent Variable „ . , Sample and scale
Mean rho score Low
Medium
High
F
df
Sample A HGSHSiA n
.71 49
.76 75
.75 25
.26
2, 146
Sample B HGSHS:A n SHSSiC n
.75 21 .34 42
.73 25 .35 68
.70 23 .20 38
.09
2,66
.87
2, 145
who recalled two or more new items from R-Trial 1 to R-Trial 2 (reversers) and those who recalled fewer than two new items on R-Trial 2 (i.e., nonreversers). Separate 3 X 2 (High/Medium/Low X Reversers/Nonreversers) ANOVAS were conducted on the R-Trial 1 rho scores from the HGSHS:A in Sample A and on the R-Trial 1 rho scores from the HGSHS:A and the SHSS:C in Sample B. None of the main effects or interactions in any of the analyses approached significance. In short, we failed to replicate Evans and Kihlstrom's major findings. There were no significant differences in rho scores on any scale between subjects at different susceptibility levels or between reversers and nonreversers. Evans and Kihlstrom (1973) considered negative rho scores to be random regardless of their magnitude. The justification for this practice is unclear, and it seems more reasonable to evaluate the magnitude of the rho score and ignore the sign. After all, a rho score of —1.00 indicates perfectly ordered recall; the order is simply reversed from most recent to least recent. Therefore, we repeated our Susceptibility Level X Reversers/ Nonreversers ANOVAS on the absolute value of the rho scores. Once again, none of the main effects or interactions were significant for either scale given to subjects in Sample B. For Sample A on the HGSHS:A, however, 2 These classification criteria are those employed by Kihlstrom (1975). They enhance the likelihood that those subjects classified as high on the HGSHSIA will also score high on a more stringent test such as the SHSSiC. 3 These classification criteria are those employed by Evans and Kihlstrom (1973).
480
H. LORRAINE RADTKE AND NICHOLAS P. SPANOS
the susceptibility main effect, F(2, 143) = Table 3 3.4, and Susceptibility X Reverser/Nonre- Frequency of Ordered and Random Recall by verser interaction achieved significance, F(2, Scale and Reversibility 143) = 5.12. Pairwise comparisons of the Rho Reversers means revealed that for the medium susceptibles, reversers (Af=.87, 57) = .18) obSample A (HGSHS:A) 14(14) 46 (46) tained higher rho scores than nonreversers Ordered 4.15 35 (14) 54 (45) (M=.75, S/>=.28), f(73) = 2.76. Lows Random showed the opposite effect; reversers (M = Sample B (HGSHS:A) .45, SD = .34) obtained lower rho scores Ordered 7(7) 30 (30) 3.92 than nonreversers (M = .80, SD = .29), Random 19(14) 13(9) ?(47) = 2.19. Finally, medium reversers Sample B (SHSS:C) obtained higher rho scores than low rever- Ordered 12(12) 29(29) 7.39 sers, t(25) = 2.73. None of the other means Random 58 (41) 49 (47) differed significantly. As in Kihlstrom (1975) we calculated the Note. The frequencies in parentheses exclude those submedians for the absolute values of the rho jects who recalled three items. scores and compared high, medium, and low susceptibles using Mann-Whitney U tests. those whose rho scores were nonsignificant The only significant difference involved high or negative (random recallers). In both samand low susceptibles on the SHSS:C. Contrary ples high, medium, and low susceptibles did to Kihlstrom's (1975) findings, however, the not differ in their tendency to obtain random median rho for the lows (.53) was signifi- versus ordered rho scores. These results are cantly lower than that for the highs (.79), summarized in Table 2. U = 1792, z = 2.44. After collapsing the scores across susceptibility level, we found that reversers were more likely than nonreversers to obtain ranOrdered/Random Recall Index dom rho scores (see Table 3). This was true for the HGSHS:A given to subjects in Sample Following Evans and Kihlstrom, subjects A and for both the HGSHS:A and SHSS:C given were divided into those whose rho scores to subjects in Sample B. The classification were positive and statistically significant of a rho score as random or ordered depends (p < .05, one-tailed; ordered recallers) and heavily upon the number of items recalled because the criterion for classifying a rho score as ordered (i.e., statistically signifiTable 2 cant) becomes lower as the number of items Frequency of Ordered and Random Recall recalled increases. Specifically, when only During Amnesia by Scale and Susceptibility three items are recalled, rho is considered to be random regardless of how the items Susceptibility are ordered. Even when the three items are Rho Medium Low High perfectly ordered and rho equals 1.00, it is classified as random. Because high suscepSample A (HGSHS:A) Ordered 30 (30) 19(19) 11 (11) tibles typically recall less than low suscep45 (29) 30(21) Random 14(9) tibles, highs are more likely than lows to recall only three items. Thus, proportionally Sample B (HGSHS:A) more highs than lows will be classified as 15(15) 12(12) Ordered 10(10) 10(7) random recallers regardless of the order of Random 13(8) 9(8) their recall.4 We therefore repeated our Sample B (SHSS:C) analyses omitting those subjects who re9 (9) 20 (20) Ordered 12(12) called three items. Under these circumRandom 29 (18) 48 (41) 30 (29) Note, The frequencies in parentheses exclude those subjects who recalled only three items.
4 We obtained these criteria for random and ordered recall directly from Kihlstrom's (1975) thesis.
TEMPORAL SEQUENCING DURING POSTHYPNOTIC AMNESIA
stances, neither susceptibility nor reversibility were related to the classification of rho scores as random or ordered. The frequencies involved in these analyses are presented in parentheses in Tables 2 and 3. These results fail once again to support the temporal disorganization hypothesis. Differential Recall of the First Item
481
HGSHS:A. Subjects at all susceptibility levels tended to recall the initial item first (see Table 4). For Sample B on the HGSHS:A, susceptibility was marginally related to the initial item recalled for nonreversers only, X2(2) = 5.43, p < .10. Again, it was the mediums and lows that differed, x 2 0) = 3.44, p < .10 (see Table 4). On the SHSSIC, high and medium reversers were less likely than low reversers to recall the initial item first, X2(2) = 6.37 (see Table 4). There was no relationship between susceptibility and initial item recall for nonreversers. When results were collapsed across susceptibility on the SHSS:C, reversers recalled the initial item first less often than nonreversers did, X 2 (l) = 9.24 (see Table 4).
All subjects who recalled at least one item during the amnesia suggestion period were classified according to whether or not the first item they recalled was the first suggestion on the susceptibility scale. On this index Evans and Kihlstrom (1973) compared high and low susceptibles only on the SHSS:C. However, we chose to compare highs and Discussion lows on the HGSHS:A as well. For Sample A on the HGSHS:A, %2(0 = 4.46, p < .05, and Our results constitute two independent for Sample B on the SHSSIC, x 2 (0 = H.55, failures to confirm the temporal disorganip < .05, the first item recalled by low sus- zation effect using the susceptibility-scale ceptibles was more likely to be the initial test paradigm. Even when we employed statissuggestion than some other suggestion. High tical procedures identical to those of Evans susceptibles were less likely to recall the ini- and Kihlstrom (1973), we found no relatial item first than lows were.5 For Sample tionship between susceptibility and rho scores B on the HGSHS:A, highs and lows did not or between susceptibility and the classifidiffer on their tendency to recall the initial cation of rho scores as ordered or random. item first. These findings are generally con- When our analyses included medium sussistent with those reported by Evans and ceptibles and involved comparisons between Kihlstrom (1973). reversers and nonreversers, we again failed Having replicated Evans and Kihlstrom to confirm the temporal disorganization ef(1973), we repeated these analyses but now fect. The magnitude of HGSHS:A and SHSS:C included medium susceptibles. For Sample rho scores was unrelated to hypnotic suscepA, subjects' classifications as high, medium, tibility in either reversers or nonreversers. or low in susceptibility were unrelated to the Moreover, among subjects recalling more tendency to recall the initial HGSHS:A item than three items, reversers were no more first. For Sample B on the HGSHS:A, suscep- likely than nonreversers to be classified as tibility was related to the first item recalled random recallers. when all subjects were considered, x2(2) = We partially replicated Evans and Kihl7.54. Mediums were less likely than lows to strom's (1973) findings concerning differrecall the initial item first as opposed to some ential recall of the initial item. We found other item, x2( 1) = 5.31. Mediums and highs that low susceptibles were more likely to redid not differ, nor did lows and highs. On call the initial item first than high suscepthe SHSSIC, however, highs and mediums tibles in Sample A on the HGSHS:A and in were less likely than lows to recall the initial Sample B on the SHSS:C. Nevertheless, in item first, x2(2) = 12.93. The frequencies Sample B on the HGSHS:A, susceptibility was used in these analyses are presented in not related to the first item recalled. The Table 4. We then repeated our analyses for rev5 On the HGSHS:A the third suggestion presented is ersers and nonreversers separately. In both for scoring purposes treated as the first suggestion becases susceptibility was not related to the cause it is the first to be presented following the hypnotic first item recalled for Sample A on the induction procedure.
482
H. LORRAINE RADTKE AND NICHOLAS P. SPANOS
Table 4 Number of Subjects Who Recalled the Initial Scale Item First During Amnesia by Scale, Susceptibility, and Reversibility Reversers Initial item recalled
Nonreversers Susceptibility
Susceptibility High
Medium
Low
High
Medium
Low
11 2
48 9
47 3
Item 1 Any other item
17
9
Sample A (HGSHS:A) 8 23 3 5
Item 1 Any other item
11 4
Sample B (HGSHS:A) 8 3 6 0
14 4
9 8
18 3
Item 1 Any other item
16 47
Sample B (SHSS:C) 5 2
6 9
17 15
21 14
13 26
Note. Subjects who recalled no items during amnesia testing were excluded from these analyses. In Sample A, 15 highs, 21 mediums, and 19 lows were excluded. In Sample B on the HGSHS:A, 15 highs, 9 mediums, and 6 lows were excluded. In Sample B on the SHSS:C, 27 highs, 2 mediums, and 3 lows were excluded.
theoretical significance of these replications is obscure for at least four reasons. First, corresponding results were not obtained with the two major indices of temporal organization—rho scores and random/ordered classification. The latter two indices reflect the ordering of subjects' entire recall protocols and therefore are more representative and valid indices of temporal organization than an index based on the differential recall of a single item. Second, the significant effect found for the HGSHS:A in Sample A was not replicated using the same scale in Sample B. Third, significant differences on the differential recall index were not obtained between high susceptibles and medium susceptibles. Fourth, differential recall of the initial item first was not consistently related to reversibility of recall. Here again the effect was scale-specific. Reversers were less likely than nonreversers to recall the initial item first on the SHSS:C, but these groups did not differ in this regard on the HGSHSIA. Since reversibility is an important criterion of amnesia, our failure to find a consistent relationship between differential initial item recall and reversibility runs counter to the disorganization hypothesis. There are no obvious characteristics of our samples that can account for our overwhelmingly negative results. These samples were
equivalent in size to those used by Evans and Kihlstrom (1973). Furthermore, the distribution of susceptibility in our sample ensured a sufficient number of high susceptibles to adequately test the hypotheses in question. Our failures to replicate, coupled with similar failures by St. Jean and Coe (1981), force us to conclude that the relationship between temporal disorganization and hypnotic susceptibility, when assessed using the susceptibility-scale paradigm, is not a reliable phenomenon. We do not wish to argue that hypnotic amnesia and temporal disorganization are unrelated—simply that such a relationship has not been convincingly demonstrated. Another line of research, focusing on organization according to taxonomic categories during hypnotic amnesia, has demonstrated a reliable relationship between reversible amnesia and a breakdown in the organization of recall when overlearned word lists served as the to-be-forgotten material. Partial amnesics showed less clustering (i.e., organization according to taxonomic categories) during an amnesia suggestion than before it or after canceling it. Full recallers (i.e., nonamnesics), on the other hand, showed a high level of clustering across the three recall trials (Radtke-Bodorik, Planas, & Spanos, 1980; Radtke-Bodorik, Spanos,
TEMPORAL SEQUENCING DURING POSTHYPNOTIC AMNESIA
& Haddad, 1979; Spanos & Bodorik, 1977; Spanos & D'Eon, 1980; Spanos, RadtkeBodorik, & Stam, 1980; Spanos, Stam, D'Eon, Pawlak, & Radtke-Bodorik, 1980). The discrepancy between the consistent findings using the word-list paradigm and the inconsistent findings using the susceptibilityscale paradigm may be due to methodological deficiencies in the susceptibility-scale paradigm or to differences in the variables that affect temporal and categorical organization. Further research is needed to clarify this issue. At this point, however, we doubt that a relationship between temporal disorganization and amnesia can be demonstrated using standardized scales of hypnotic susceptibility to assess amnesia. These scales were simply not constructed for the purpose of assessing relationships between recall order and amnesia. Using them for this purpose is so fraught with methodological problems that failures to replicate are not surprising, and even consistent findings would be difficult to interpret. The five most obvious problems, with the way in which these scales have been used, are: 1. There are no procedures for assessing original learning. A low level of recall and/ or low rho scores at R-Trial 1 can reflect inefficient learning, amnesia, or both. In three studies Schwartz (1978, 1980, Note 1) deleted the amnesia suggestion from the SHSS:C and, in place of it, simply asked subjects to recall what had occurred during the session. In each study, high susceptibles obtained lower rho scores than lows did, thereby indicating that nonsequential recall is associated with susceptibility in the absence of amnesia. Kihlstrom and Evans (1979) failed to replicate Schwartz's findings, and the reasons for these contradictory findings remain to be clarified. Nevertheless, the failure of susceptibility scales to obtain baseline measures of recall and organization makes any interpretation of recall and rho scores following amnesia tenuous. 2. The recall task at R-Trial 1 is ambiguous. It is not clear to subjects what they are supposed to be recalling. Subjects are asked to recall "everything that happened" to them during the hypnotic experience. Nevertheless, on the HGSHS:A, SHSS:A, and
483
SHSS:B, only 9 of the 12 items are scored for amnesia, and on the SHSS:C 11 of the 12 items are so scored. Furthermore, on all four scales many events such as the suggestions for relaxation and subjects' private experiences are not considered to be scorable items. Thus, subjects often recall events that did happen but that are not considered to be scorable for amnesia (Evans, Kihlstrom, & Orne, 1973). 3. On the HGSHS:A, subjects are limited with respect to the time allowed for recall. Frequently, this time is not sufficient, and subjects must attenuate their recall before they are finished. 4. On R-Trial 2 (after cancellation of the suggestion) subjects are not asked to recall everything that they can remember. Instead, they are asked to recall only new information—events that they did not recall on RTrial 1. Because subjects are not asked to "recall everything" on R-Trial 2, it is impossible to assess the role of random fluctuations in recall between the two recall trials. More important, it is impossible to obtain a meaningful rho score between presentation order and recall order on R-Trial 2. In short, the construction of these scales makes it impossible to assess subjects' level of recall organization either before administration of the suggestion or after its cancellation. As a result, differences between subjects in obtained rhos on R-Trial 1 cannot, with any confidence, be attributed to the effects of the amnesia suggestion.6 5. Magnitude of rho is dependent on amount recalled as well as on agreement between rank orderings. For instance, the same small transposition in recall order results in a smaller rho value for low as opposed to high levels of recall. High susceptibles tend to recall less than low susceptibles and therefore may obtain lower rho scores simply as a function of lower levels of recall. In order to eliminate this problem Kihlstrom and Evans (1979) compared the rho scores of high and low susceptible subjects who had been matched at five different levels of re6 Kihlstrom (1975) modified the amnesia suggestion on the HGSHS:A so that subjects were asked to recall everything after the suggestion was lifted. Contrary to expectation, increases in amount recalled were not accompanied by increases in rho scores.
484
H. LORRAINE RADTKE AND NICHOLAS P. SPANOS
call. Using multiple / tests with a one-tailed found extent of organization with number level of significance, they reported that high of items recalled (Murphy, 1979). susceptibles obtained lower rho scores than lows when these groups recalled three items (p < .001), four items (p < .005), five items Reference Note (p < .05), and six items (p < .05). No differences were detected when seven to nine 1. Schwartz, W. Trance and personal history. Paper presented at 87th annual convention of the American items were recalled. The use of such liberal Psychological Association, New York, September statistical procedures is questionable. When 1979. the results are assessed with two-tailed t tests (still a liberal procedure considering that References multiple comparisons are being made), the differences between highs and lows for five K. S. Hypnosis for the seriously curious. Belitems and six items are no longer significant. Bowers, tnont, Calif.: Wadsworth, 1976. Although the reliability of these data clearly Cooper, L. M. Hypnotic amnesia. In E. Fromm & need to be more firmly established, their R. E. Shor (Eds.), Hypnosis: Developments in research and new perspectives (2nd ed.). Chicago: Almeaning would remain ambiguous regarddine-Atherton, 1979. less of statistical findings. These data con- Evans, F. J. Phenomena of hypnosis: 2. Posthypnotic cern relationships between rho scores and amnesia. In G. D. Burrows & L. Dennerstein (Eds.), susceptibility, not between rho scores and Handbook of hypnosis and psychosomatic medicine. Amsterdam: North-Holland Biomedical Press, 1980. amnesia. In summary, even if it had been reliably Evans, F. J., & Kihlstrom, J. F. Post hypnotic amnesia as disrupted retrieval. Journal of Abnormal Psydemonstrated that high susceptibles obchology, 1913,82, 317-323. tained lower rho scores than low susceptibles Evans, F. J., Kihlstrom, J. F., & Orne, E. C. Quantifying at R-Trial 1, methodological deficiencies subjective reports during posthypnotic amnesia. Proceedings of the 81st annual convention of the Amerwould have precluded firm conclusions conican Psychological Association, 1973,5, 1077-1078. cerning amnesia. In order to obtain inter(Abstract) pretable results concerning sequential or- Evans, F. J., & Orne, M. T. Motivation, performance, ganization and amnesia, we suggest two and hypnosis. International Journal of Clinical and Experimental Hypnosis, 1965, 13, 103-116. modifications in the construction of susceptibility scales. First, recall should be assessed Hilgard, E. R. Divided consciousness: Multiple controls in human thought and action. New York: Wiley, before the suggestion to allow determination 1977. of presuggestion differences in recall orga- Kihlstrom, J. F. The effects of organization and motinization. Second, after the suggestion is canvation on recall during post-hypnotic amnesia (Doctoral dissertation, University of Pennsylvania, 1975). celled, subjects should be required to recall Dissertation Abstracts International, 1975,3(5, 2473Beverything they can remember. This would 2474B. (University Microfilms No. 75-24082) enable accurate assessment of reversibility. Kihlstrom, J. F. Models of posthypnotic amnesia. AnAs an additional advantage, these changes nals of the New York Academy of Sciences, 1977, 296, 284-301. would allow for the measurement of subjective organization.7 In this way, future studies Kihlstrom, J. F. Context and cognition in posthypnotic amnesia. International Journal of Clinical and Excould assess the relationship between preperimental Hypnosis, 1978, 26, 246-267. sentation order and recall order before the Kihlstrom, J. F., & Evans, F. J. Recovery of memory amnesia suggestion, during it, and after canafter posthypnotic amnesia. Journal of Abnormal Psychology, 1976, 85, 564-569. celling it with a measure that does not con7
In their first study, Evans and Kihlstrom (1973) collapsed their results across subjects and obtained a single subjective organization score for high susceptibles and a single score for low susceptibies. The score for the highs was lower and was taken to indicate that highs as a group are less consistent in the way they order their recall than lows are. Unfortunately, there is no way to assess the statistical significance of such differences.
Kihlstrom, J. F., & Evans, F. J. Memory retrieval processes during post-hypnotic amnesia. In J. F. Kihlstrom & F. J. Evans (Eds.), Functional disorders of memory. Hillsdale, N.J.: Erlbaum, 1979. Maslach, C. Negative emotional biasing of unexplained arousal. Journal of Personality and Social Psychology, 1979, 37. 953-969. Murphy, M. D. Measurement of category clustering in free recall. In C. R. Puff (Ed.), Memory organization and structure. New York: Academic Press, 1979.
TEMPORAL SEQUENCING DURING POSTHYPNOTIC AMNESIA Nace, E. P., Orne, M. T., & Hammer, A. G. Posthypnotic amnesia as an active psychic process: The reversibility of amnesia. Archives of General Psychiatry, 1974, 31, 257-260. Radtke-Bodorik, H. L., Planas, M., & Spanos, N. P. Suggested amnesia, verbal inhibition, and disorganized recall for a long word list. Canadian Journal of Behavioral Science, 1980, 12, 87-97. Radtke-Bodorik, H. L., Spanos, N. P., & Haddad, M. G. The effects of spoken versus written recall on suggested amnesia in hypnotic and task-motivated subjects. American Journal of Clinical Hypnosis, 1979, 22, 8-16. Schwartz, W. S. Time and context during hypnotic involvement. International Journal of Clinical and Experimental Hypnosis, 1978, 26, 307-316. Schwartz, W. S. Hypnosis and episodic memory. International Journal of Clinical and Experimental Hypnosis, 1980, 28, 375-385. Shor, R. E., & Orne, E. C. The Harvard Group Scale of Hypnotic Susceptibility, Form A. Palo Alto, Calif.: Consulting Psychologists Press, 1962. Spanos, N. P., & Bodorik, H. L. Suggested amnesia and disorganized recall in hypnotic and task-motivated subjects. Journal of Abnormal Psychology, 1977,86, 295-305.
485
Spanos, N. P., & D'Eon, J. L. Hypnotic amnesia, disorganized recall, and inattention. Journal of Abnormal Psychology, 1980, 89, 744-750. Spanos, N. P., Radtke-Bodorik, H. L., & Stam, H. J. Disorganized recall during suggested amnesia: Fact not artifact. Journal of Abnormal Psychology, 1980, 89, 1-19. Spanos, N. P., Stam, H. J., D'Eon, J. L., Pawlak, A. E., & Radtke-Bodorik, H. L. The effects of social psychological variables on hypnotic amnesia. Journal of Personality and Social Psychology, 1980, 39, 737750. St. Jean, R., & Coe, W. C. Recall and recognition memory during posthypnotic amnesia: A failure to confirm the disrupted-search hypothesis and the memory disorganization hypothesis. Journal of Abnormal Psychology, 1981, 90, 231-241. Weitzenhoffer, A. M., & Hilgard, E. R. Stanford Hypnotic Susceptibility Scale, Forms, A and B. Palo Alto, Calif.: Consulting Psychologists Press, 1959. Weitzenhoffer, A. M., & Hilgard, E. R. Stanford Hypnotic Susceptibility Scale, Form C. Palo Alto, Calif.: Consulting Psychologists Press, 1962. Received March 9, 1981
Journal of Abnormal Psychology 1981, Vol. 90, No. 5, 476-485
Temporal Sequencing During Posthypnotic Amnesia: A Methodological Critique H. Lorraine Radtke and Nicholas P. Spanos Carleton University, Ottawa, Ontario, Canada In a series of studies using the susceptibility-scale paradigm, high susceptible subjects were less likely than low susceptibles to recall the events of the hypnotic session in temporal sequence (i.e., temporal disorganization effect) following an amnesia suggestion. The primary measure of recall order was the rank-order correlation (rho scores) between the presentation order and the recall order of hypnotic experiences computed for each subject. Following a suggestion for posthypnotic amnesia, high susceptibles usually obtained lower rho scores than low susceptibles. We critically examined this research noting methodological shortcomings associated with the susceptibility-scale paradigm, inconsistent findings, and failures to replicate. We also presented two studies that found no relationship between susceptibility level and rho scores. These null results held true for subjects who recalled new information after cancellation of the amnesia suggestion (reversers) as well as for those who did not recall new information (nonreversers). Nevertheless, we did replicate previous work on differential recall of the first item. Suggestions for the development of a more adequate paradigm to assess recall order during amnesia using the susceptibility-scale paradigm are discussed.
In 1973, Evans and Kihlstrom published the first in an influential series of articles on the organization of recall during posthypnotic amnesia (Evans & Kihlstrom, 1973; Kihlstrom, 1975; Kihlstrom & Evans, 1979). This line of research has frequently been credited with demonstrating a reliable association between posthypnotic amnesia and a breakdown in the temporal organization of recall (see reviews by Bowers, 1976; Cooper, 1979; Evans, 1980; Hilgard, 1977; Kihlstrom, 1975, 1977, 1978; Kihlstrom & Evans, 1979; Maslach, 1979). Moreover, the temporal disorganization effect has served as the major empirical underpinning for a disrupted retrieval hypothesis of posthypnotic amnesia (Evans & Kihlstrom, 1973) and more recently for a dissociation hypothesis of amnesia (Kihlstrom, 1977). Nonetheless, close examination of the exThis research was supported by a Social Sciences and Humanities Research Council doctoral fellowship awarded to the first author and Natural Sciences and Engineering Research Council Operating Grant A6385 awarded to the second author. We thank H. J. Stam for critically reading an earlier version of this manuscript. Radtke is now at the Dept. of Psychology, University of Manitoba, Canada. Requests for reprints should be sent to Nicholas Spanos, Department of Psychology, Carleton University, Ottawa, Ontario, Canada K1S 5B6.
perimental paradigm employed in these studies (hereafter called the susceptibilityscale paradigm) reveals a number of serious methodological problems. Moreover, one recent study was unable to replicate the major findings (St. Jean & Coe, 1981). As a result, it is questionable whether a relationship between lack of temporal organization and hypnotic amnesia has in fact been demonstrated. The present report critically reviews this work and also presents two attempts to replicate the temporal disorganization effect using the susceptibility-scale paradigm. The susceptibility-scale paradigm assesses subjects' performance on the amnesia suggestion contained in standardized scales of hypnotic susceptibility. The scales employed have included the Harvard Group Scale of Hypnotic Susceptibility: Form A (HGSHSIA; Shor & Orne, 1962) and the Stanford Scales of Hypnotic Susceptibility, Form B (SHSS:B; Weitzenhoffer & Hilgard, 1959) and Form C (SHSS:C; Weitzenhoffer & Hilgard, 1962). On all of these scales the 12th and final test suggestion calls for posthypnotic amnesia. Subjects are told that after they are awakened they will be unable to remember anything that occurred during hypnosis until they hear a prearranged cue (i.e., "Now you
476
TEMPORAL SEQUENCING DURING POSTHYPNOTIC AMNESIA
can remember everything") that cancels their amnesia. After being awakened, but before presentation of the cancellation cue, subjects are asked to write down (or report orally, depending on the scale) everything that happened to them from the time of the hypnotic induction procedure (R-Trial 1). Three minutes later the suggestion is cancelled with the prearranged cue, and subjects are given two minutes to report anything else they now remember (R-Trial 2). Evans and Kihlstrom (1973) compared the order in which test suggestions were presented on the susceptibility scales with the order in which these suggestions were recalled by subjects on R-Trial 1. Three indices of association between presentation order and recall order were obtained for each subject. The most important index involved computation of a Spearman rank-order correlation coefficient (rho score) between presentation order of the test suggestions and recall order on R-Trial 1. A second index involved classifying each rho score as statistically significant or nonsignificant. Recall was considered to be ordered (i.e., organized) if the rho score was positive and significant (p < .05, one-tailed) and random (i.e., disorganized) in every other case. The third index simply assessed whether the first item recalled by a subject was also the first suggestion presented on the scale in question. Typically, subjects were classified as high, medium, or low in hypnotic susceptibility on the basis of their scores on the susceptibility scales, and relationships between susceptibility level and the three indices of association were assessed. Results for each index will be described separately. Rho Score Index In their initial study Evans and Kihlstrom (1973) tested subjects on the HGSHS:A, SHSS:B, and SHSS:C. Separate one-way analyses of variance (ANOVAS) using high, medium, and low susceptibility levels as the independent variable and each subject's rho score as the dependent variable were performed on the data from each scale. The ANOVAS were significant for the SHSSIB and SHSS:C but nonsignificant for the HGSHS:A. The means involved in the two significant
477
ANOVAS, however, were not consistently ordered. In the case of the SHSS:C, medium susceptibles had a lower mean rho score than low susceptibles and, on the SHSS:B, mediums had a lower mean rho score than either highs or lows. Differences involving the means for the medium susceptibles were not tested for statistical significance. However, the rho scores for high and low susceptibles were compared with multiple one-tailed t tests. High susceptibles were found to have significantly lower rho scores than low susceptibles on the HGSHS:A and the SHSS:C. On the SHSS:B this difference was not significant. At best, these findings weakly suggest a difference in rho scores between high and low susceptibles on the HGSHSIA and SHSS:C. By ignoring the medium susceptibles in their data analysis, Evans and Kihlstrom (1973) weakened the explanatory power of their hypothesis. The temporal disorganization effect was replicated by Kihlstrom (1975) using the HGSHS:A. Recently, however, St. Jean and Coe (1981) attempted unsuccessfully to replicate it. They found no significant differences in the rho scores of high, medium, and low susceptibles on either the SHSS:B or the SHSS:C.
Ordered/Random Recall Index In three samples, subjects were divided into those whose rho scores were either random (i.e., nonsignificant) or ordered (significant at p < .05, one-tailed). Subjects in the first sample had been administered the HGSHS:A, SHSS:B, and SHSS:C (Evans & Kihlstrom, 1973), those in the second sample had been given the HGSHS:A and the SHSS:C, and those in the third sample were tested on the HGSHS:A alone (Kihlstrom & Evans, 1979). Subjects in Samples 1 and 2 were classified as high, medium, or low in susceptibility and chi-square analyses were used to assess the degree of association between susceptibility level and random/ordered recall for each susceptibility scale. The three chisquare analyses were significant for Sample 1. For Sample 2, the analysis was significant for the SHSSiC but not for the HGSHS:A. According to Kihlstrom and Evans (1979), these generally significant effects indicate
478
H. LORRAINE RADTKE AND NICHOLAS P. SPANOS
that a greater proportion of high susceptibles than low susceptibles showed random recall when given an amnesia suggestion. It should be noted, however, that in four out of five analyses the proportion of medium susceptibles and high susceptibles showing random and ordered recall did not differ significantly. Also, in Sample 1 on the SHSS:B and SHSS:C medium susceptibles showed proportionately more random recall than lows.1 In Sample 3, highs were more likely to show random recall than lows; mediums were not tested. St. Jean and Coe (1981), however, found no significant differences in the proportion of high, medium, and low susceptibles showing random as opposed to ordered recall on either the SHSStB or the SHSS:C. Differential Recall of First Suggestion In their original study Evans and Kihlstrom (1973) reported that high susceptibles were significantly less likely than low susceptibles to recall first the initial test suggestion on the SHSS:C. No data were presented for performance on the HGSHS:A or SHSS:B or for medium susceptibles. St. Jean and Coe (1981) replicated and extended this differential first suggestion recall effect. They found that high and medium susceptibles were less likely than low susceptibles to recall the first suggestion first on both the SHSS:B and SHSS:C.
Nonrecall, Reversibility, and Organization Disorganized recall is supposedly a characteristic of hypnotic amnesia and not simply a correlate of hypnotic susceptibility (Evans & Kihlstrom, 1973). In support of this notion Evans and Kihlstrom (1973) reported that subjects who recalled three or fewer suggestions at R-Trial 1 (i.e., complete nonrecallers) were higher in susceptibility than random recallers, who, in turn, were higher in susceptibility than ordered recallers. Although this finding appears to be reliable (replicated by Kihlstrom & Evans, 1979), it is related only tenuously to the hypothesis that amnesia (as opposed to susceptibility) is associated with disorganization. An adequate test of this notion requires that amnesia be measured independently of hypnotic susceptibility. The hypothesis pre-
dicts that amnesics will show disorganized recall regardless of their level of susceptibility. Unfortunately, standardized susceptibility scales make no provision for unambiguously assessing degree of amnesia. Most investigators agree that hypnotic amnesia is a temporary recall deficit and, therefore, should be differentiated from other recall deficits in terms of its reversibility (Kihlstrom & Evans, 1976; Nace, Orne, & Hammer, 1974; Spanos & Bodorik, 1977). Thus, even a minimum requirement for a demonstration of amnesia should include the recovery of "forgotten" material after cancellation of the suggestion (i.e., information not recalled at R-Trial 1 should be recalled at R-Trial 2). This minimum requirement is not met on standardized susceptibility scales. Instead, amnesia on these scales is defined solely in terms of the number of suggestions recalled at R-Trial 1. Kihlstrom and Evans (1979) attempted to overcome this problem by devising an index of reversible recall. They compared the rho scores of subjects who recalled at least two items at R-Trial 2 that had not been recalled previously at R-Trial 1 (reversers) with those of subjects who recalled no new items at RTrial 2 (nonreversers). On the SHSS:C reversers had significantly lower rho scores than nonreversers. However, on the HGSHS:A the rho score difference between these treatments was nonsignificant (p > .05, onetailed). Thus, there is at best weak support for the hypothesis that disorganization and amnesia (as defined by reversibility) are related. Given the inconsistencies in the work reviewed above, we felt it imperative to conduct an independent replication study. We obtained two large samples. One had been tested only on the HGSHS:A, and the other had been assessed on the HGSHStA and the SHSS:C. (HGSHS:A scores, however, were available for only a subset of the second sample.) To replicate the procedures employed by Evans and Kihlstrom as closely as possible, we carried out the same statistical analyses as they and adopted the same alpha 1
These analyses are not presented in Kihlstrom and Evans (1979). We computed the chi-square on the basis of frequencies presented in that paper.
479
TEMPORAL SEQUENCING DURING POSTHYPNOTIC AMNESIA
levels. However, for analyses not conducted by Evans and Kihlstrom (1973) we conducted only two-tailed tests and set alpha consistently at .05. Although previous work has tended to ignore the performance of medium susceptibles, any theory of hypnotic behavior must surely account for the behavior of medium susceptibles (Evans & Orne, 1965). Therefore, we included data from medium susceptibles in all analyses that constituted an extension of previous work. Method Data from two samples of subjects that had been administered standard tests of hypnotic susceptibility as a screening procedure for further experiments conducted in our laboratory were available for analysis. Sample A had been tested only on the HGSHSIA in small groups of 7 to 12. The 240 subjects in this sample (93 males, 147 females) ranged in age from 17 to 49 years. Of these subjects, 54 scored 9 or above and were classified as high susceptibles, 106 scored between 5 and 8 and were classified as medium susceptibles, and 80 scored between 0 and 4 and were classified as low susceptibles.2 Sample B included 223 subjects (93 males, 130 females), aged 17 to 62 years, who had been administered the SHSS:C individually. These subjects had been administered the HGSHSIA previously in small groups of 5 to 10. However, HGSHSIA data were available only for 118(55 males, 63 females) of the total sample. All subjects were classified on susceptibility according to their scores on the SHSS:C.3 The classification criteria were as follows: (a) Highs scored 8-12, (b) mediums scored 5-7, and (c) lows scored 0-4. Of the 223 subjects tested on the SHSS:C, 105 were classified as high, 73 as medium, and 45 as low. Of the 118 for whom HCSHS:A data were available, 48 were high, 40 were medium, and 30 were low, according to their scores on the SHSS:C.
Results Rho Index Spearman rank-order correlation coefficients (rhos) were calculated for each subject who recalled three or more items. Following Evans and Kihlstrom (1973), oneway ANOVAS were performed on the rho scores. The results are summarized in Table 1. They indicate no relationship between hypnotic susceptibility and organization of recall as indexed by rho scores. Individual one-tailed / tests between high and low susceptibles revealed no difference on either scale in either sample. Subjects at each susceptibility level (high, medium, low) were then divided into those
Table 1 Summary of Rho Score Analyses With Susceptibility as the Independent Variable „ . , Sample and scale
Mean rho score Low
Medium
High
F
df
Sample A HGSHSiA n
.71 49
.76 75
.75 25
.26
2, 146
Sample B HGSHS:A n SHSSiC n
.75 21 .34 42
.73 25 .35 68
.70 23 .20 38
.09
2,66
.87
2, 145
who recalled two or more new items from R-Trial 1 to R-Trial 2 (reversers) and those who recalled fewer than two new items on R-Trial 2 (i.e., nonreversers). Separate 3 X 2 (High/Medium/Low X Reversers/Nonreversers) ANOVAS were conducted on the R-Trial 1 rho scores from the HGSHS:A in Sample A and on the R-Trial 1 rho scores from the HGSHS:A and the SHSS:C in Sample B. None of the main effects or interactions in any of the analyses approached significance. In short, we failed to replicate Evans and Kihlstrom's major findings. There were no significant differences in rho scores on any scale between subjects at different susceptibility levels or between reversers and nonreversers. Evans and Kihlstrom (1973) considered negative rho scores to be random regardless of their magnitude. The justification for this practice is unclear, and it seems more reasonable to evaluate the magnitude of the rho score and ignore the sign. After all, a rho score of —1.00 indicates perfectly ordered recall; the order is simply reversed from most recent to least recent. Therefore, we repeated our Susceptibility Level X Reversers/ Nonreversers ANOVAS on the absolute value of the rho scores. Once again, none of the main effects or interactions were significant for either scale given to subjects in Sample B. For Sample A on the HGSHS:A, however, 2 These classification criteria are those employed by Kihlstrom (1975). They enhance the likelihood that those subjects classified as high on the HGSHSIA will also score high on a more stringent test such as the SHSSiC. 3 These classification criteria are those employed by Evans and Kihlstrom (1973).
480
H. LORRAINE RADTKE AND NICHOLAS P. SPANOS
the susceptibility main effect, F(2, 143) = Table 3 3.4, and Susceptibility X Reverser/Nonre- Frequency of Ordered and Random Recall by verser interaction achieved significance, F(2, Scale and Reversibility 143) = 5.12. Pairwise comparisons of the Rho Reversers means revealed that for the medium susceptibles, reversers (Af=.87, 57) = .18) obSample A (HGSHS:A) 14(14) 46 (46) tained higher rho scores than nonreversers Ordered 4.15 35 (14) 54 (45) (M=.75, S/>=.28), f(73) = 2.76. Lows Random showed the opposite effect; reversers (M = Sample B (HGSHS:A) .45, SD = .34) obtained lower rho scores Ordered 7(7) 30 (30) 3.92 than nonreversers (M = .80, SD = .29), Random 19(14) 13(9) ?(47) = 2.19. Finally, medium reversers Sample B (SHSS:C) obtained higher rho scores than low rever- Ordered 12(12) 29(29) 7.39 sers, t(25) = 2.73. None of the other means Random 58 (41) 49 (47) differed significantly. As in Kihlstrom (1975) we calculated the Note. The frequencies in parentheses exclude those submedians for the absolute values of the rho jects who recalled three items. scores and compared high, medium, and low susceptibles using Mann-Whitney U tests. those whose rho scores were nonsignificant The only significant difference involved high or negative (random recallers). In both samand low susceptibles on the SHSS:C. Contrary ples high, medium, and low susceptibles did to Kihlstrom's (1975) findings, however, the not differ in their tendency to obtain random median rho for the lows (.53) was signifi- versus ordered rho scores. These results are cantly lower than that for the highs (.79), summarized in Table 2. U = 1792, z = 2.44. After collapsing the scores across susceptibility level, we found that reversers were more likely than nonreversers to obtain ranOrdered/Random Recall Index dom rho scores (see Table 3). This was true for the HGSHS:A given to subjects in Sample Following Evans and Kihlstrom, subjects A and for both the HGSHS:A and SHSS:C given were divided into those whose rho scores to subjects in Sample B. The classification were positive and statistically significant of a rho score as random or ordered depends (p < .05, one-tailed; ordered recallers) and heavily upon the number of items recalled because the criterion for classifying a rho score as ordered (i.e., statistically signifiTable 2 cant) becomes lower as the number of items Frequency of Ordered and Random Recall recalled increases. Specifically, when only During Amnesia by Scale and Susceptibility three items are recalled, rho is considered to be random regardless of how the items Susceptibility are ordered. Even when the three items are Rho Medium Low High perfectly ordered and rho equals 1.00, it is classified as random. Because high suscepSample A (HGSHS:A) Ordered 30 (30) 19(19) 11 (11) tibles typically recall less than low suscep45 (29) 30(21) Random 14(9) tibles, highs are more likely than lows to recall only three items. Thus, proportionally Sample B (HGSHS:A) more highs than lows will be classified as 15(15) 12(12) Ordered 10(10) 10(7) random recallers regardless of the order of Random 13(8) 9(8) their recall.4 We therefore repeated our Sample B (SHSS:C) analyses omitting those subjects who re9 (9) 20 (20) Ordered 12(12) called three items. Under these circumRandom 29 (18) 48 (41) 30 (29) Note, The frequencies in parentheses exclude those subjects who recalled only three items.
4 We obtained these criteria for random and ordered recall directly from Kihlstrom's (1975) thesis.
TEMPORAL SEQUENCING DURING POSTHYPNOTIC AMNESIA
stances, neither susceptibility nor reversibility were related to the classification of rho scores as random or ordered. The frequencies involved in these analyses are presented in parentheses in Tables 2 and 3. These results fail once again to support the temporal disorganization hypothesis. Differential Recall of the First Item
481
HGSHS:A. Subjects at all susceptibility levels tended to recall the initial item first (see Table 4). For Sample B on the HGSHS:A, susceptibility was marginally related to the initial item recalled for nonreversers only, X2(2) = 5.43, p < .10. Again, it was the mediums and lows that differed, x 2 0) = 3.44, p < .10 (see Table 4). On the SHSSIC, high and medium reversers were less likely than low reversers to recall the initial item first, X2(2) = 6.37 (see Table 4). There was no relationship between susceptibility and initial item recall for nonreversers. When results were collapsed across susceptibility on the SHSS:C, reversers recalled the initial item first less often than nonreversers did, X 2 (l) = 9.24 (see Table 4).
All subjects who recalled at least one item during the amnesia suggestion period were classified according to whether or not the first item they recalled was the first suggestion on the susceptibility scale. On this index Evans and Kihlstrom (1973) compared high and low susceptibles only on the SHSS:C. However, we chose to compare highs and Discussion lows on the HGSHS:A as well. For Sample A on the HGSHS:A, %2(0 = 4.46, p < .05, and Our results constitute two independent for Sample B on the SHSSIC, x 2 (0 = H.55, failures to confirm the temporal disorganip < .05, the first item recalled by low sus- zation effect using the susceptibility-scale ceptibles was more likely to be the initial test paradigm. Even when we employed statissuggestion than some other suggestion. High tical procedures identical to those of Evans susceptibles were less likely to recall the ini- and Kihlstrom (1973), we found no relatial item first than lows were.5 For Sample tionship between susceptibility and rho scores B on the HGSHS:A, highs and lows did not or between susceptibility and the classifidiffer on their tendency to recall the initial cation of rho scores as ordered or random. item first. These findings are generally con- When our analyses included medium sussistent with those reported by Evans and ceptibles and involved comparisons between Kihlstrom (1973). reversers and nonreversers, we again failed Having replicated Evans and Kihlstrom to confirm the temporal disorganization ef(1973), we repeated these analyses but now fect. The magnitude of HGSHS:A and SHSS:C included medium susceptibles. For Sample rho scores was unrelated to hypnotic suscepA, subjects' classifications as high, medium, tibility in either reversers or nonreversers. or low in susceptibility were unrelated to the Moreover, among subjects recalling more tendency to recall the initial HGSHS:A item than three items, reversers were no more first. For Sample B on the HGSHS:A, suscep- likely than nonreversers to be classified as tibility was related to the first item recalled random recallers. when all subjects were considered, x2(2) = We partially replicated Evans and Kihl7.54. Mediums were less likely than lows to strom's (1973) findings concerning differrecall the initial item first as opposed to some ential recall of the initial item. We found other item, x2( 1) = 5.31. Mediums and highs that low susceptibles were more likely to redid not differ, nor did lows and highs. On call the initial item first than high suscepthe SHSSIC, however, highs and mediums tibles in Sample A on the HGSHS:A and in were less likely than lows to recall the initial Sample B on the SHSS:C. Nevertheless, in item first, x2(2) = 12.93. The frequencies Sample B on the HGSHS:A, susceptibility was used in these analyses are presented in not related to the first item recalled. The Table 4. We then repeated our analyses for rev5 On the HGSHS:A the third suggestion presented is ersers and nonreversers separately. In both for scoring purposes treated as the first suggestion becases susceptibility was not related to the cause it is the first to be presented following the hypnotic first item recalled for Sample A on the induction procedure.
482
H. LORRAINE RADTKE AND NICHOLAS P. SPANOS
Table 4 Number of Subjects Who Recalled the Initial Scale Item First During Amnesia by Scale, Susceptibility, and Reversibility Reversers Initial item recalled
Nonreversers Susceptibility
Susceptibility High
Medium
Low
High
Medium
Low
11 2
48 9
47 3
Item 1 Any other item
17
9
Sample A (HGSHS:A) 8 23 3 5
Item 1 Any other item
11 4
Sample B (HGSHS:A) 8 3 6 0
14 4
9 8
18 3
Item 1 Any other item
16 47
Sample B (SHSS:C) 5 2
6 9
17 15
21 14
13 26
Note. Subjects who recalled no items during amnesia testing were excluded from these analyses. In Sample A, 15 highs, 21 mediums, and 19 lows were excluded. In Sample B on the HGSHS:A, 15 highs, 9 mediums, and 6 lows were excluded. In Sample B on the SHSS:C, 27 highs, 2 mediums, and 3 lows were excluded.
theoretical significance of these replications is obscure for at least four reasons. First, corresponding results were not obtained with the two major indices of temporal organization—rho scores and random/ordered classification. The latter two indices reflect the ordering of subjects' entire recall protocols and therefore are more representative and valid indices of temporal organization than an index based on the differential recall of a single item. Second, the significant effect found for the HGSHS:A in Sample A was not replicated using the same scale in Sample B. Third, significant differences on the differential recall index were not obtained between high susceptibles and medium susceptibles. Fourth, differential recall of the initial item first was not consistently related to reversibility of recall. Here again the effect was scale-specific. Reversers were less likely than nonreversers to recall the initial item first on the SHSS:C, but these groups did not differ in this regard on the HGSHSIA. Since reversibility is an important criterion of amnesia, our failure to find a consistent relationship between differential initial item recall and reversibility runs counter to the disorganization hypothesis. There are no obvious characteristics of our samples that can account for our overwhelmingly negative results. These samples were
equivalent in size to those used by Evans and Kihlstrom (1973). Furthermore, the distribution of susceptibility in our sample ensured a sufficient number of high susceptibles to adequately test the hypotheses in question. Our failures to replicate, coupled with similar failures by St. Jean and Coe (1981), force us to conclude that the relationship between temporal disorganization and hypnotic susceptibility, when assessed using the susceptibility-scale paradigm, is not a reliable phenomenon. We do not wish to argue that hypnotic amnesia and temporal disorganization are unrelated—simply that such a relationship has not been convincingly demonstrated. Another line of research, focusing on organization according to taxonomic categories during hypnotic amnesia, has demonstrated a reliable relationship between reversible amnesia and a breakdown in the organization of recall when overlearned word lists served as the to-be-forgotten material. Partial amnesics showed less clustering (i.e., organization according to taxonomic categories) during an amnesia suggestion than before it or after canceling it. Full recallers (i.e., nonamnesics), on the other hand, showed a high level of clustering across the three recall trials (Radtke-Bodorik, Planas, & Spanos, 1980; Radtke-Bodorik, Spanos,
TEMPORAL SEQUENCING DURING POSTHYPNOTIC AMNESIA
& Haddad, 1979; Spanos & Bodorik, 1977; Spanos & D'Eon, 1980; Spanos, RadtkeBodorik, & Stam, 1980; Spanos, Stam, D'Eon, Pawlak, & Radtke-Bodorik, 1980). The discrepancy between the consistent findings using the word-list paradigm and the inconsistent findings using the susceptibilityscale paradigm may be due to methodological deficiencies in the susceptibility-scale paradigm or to differences in the variables that affect temporal and categorical organization. Further research is needed to clarify this issue. At this point, however, we doubt that a relationship between temporal disorganization and amnesia can be demonstrated using standardized scales of hypnotic susceptibility to assess amnesia. These scales were simply not constructed for the purpose of assessing relationships between recall order and amnesia. Using them for this purpose is so fraught with methodological problems that failures to replicate are not surprising, and even consistent findings would be difficult to interpret. The five most obvious problems, with the way in which these scales have been used, are: 1. There are no procedures for assessing original learning. A low level of recall and/ or low rho scores at R-Trial 1 can reflect inefficient learning, amnesia, or both. In three studies Schwartz (1978, 1980, Note 1) deleted the amnesia suggestion from the SHSS:C and, in place of it, simply asked subjects to recall what had occurred during the session. In each study, high susceptibles obtained lower rho scores than lows did, thereby indicating that nonsequential recall is associated with susceptibility in the absence of amnesia. Kihlstrom and Evans (1979) failed to replicate Schwartz's findings, and the reasons for these contradictory findings remain to be clarified. Nevertheless, the failure of susceptibility scales to obtain baseline measures of recall and organization makes any interpretation of recall and rho scores following amnesia tenuous. 2. The recall task at R-Trial 1 is ambiguous. It is not clear to subjects what they are supposed to be recalling. Subjects are asked to recall "everything that happened" to them during the hypnotic experience. Nevertheless, on the HGSHS:A, SHSS:A, and
483
SHSS:B, only 9 of the 12 items are scored for amnesia, and on the SHSS:C 11 of the 12 items are so scored. Furthermore, on all four scales many events such as the suggestions for relaxation and subjects' private experiences are not considered to be scorable items. Thus, subjects often recall events that did happen but that are not considered to be scorable for amnesia (Evans, Kihlstrom, & Orne, 1973). 3. On the HGSHS:A, subjects are limited with respect to the time allowed for recall. Frequently, this time is not sufficient, and subjects must attenuate their recall before they are finished. 4. On R-Trial 2 (after cancellation of the suggestion) subjects are not asked to recall everything that they can remember. Instead, they are asked to recall only new information—events that they did not recall on RTrial 1. Because subjects are not asked to "recall everything" on R-Trial 2, it is impossible to assess the role of random fluctuations in recall between the two recall trials. More important, it is impossible to obtain a meaningful rho score between presentation order and recall order on R-Trial 2. In short, the construction of these scales makes it impossible to assess subjects' level of recall organization either before administration of the suggestion or after its cancellation. As a result, differences between subjects in obtained rhos on R-Trial 1 cannot, with any confidence, be attributed to the effects of the amnesia suggestion.6 5. Magnitude of rho is dependent on amount recalled as well as on agreement between rank orderings. For instance, the same small transposition in recall order results in a smaller rho value for low as opposed to high levels of recall. High susceptibles tend to recall less than low susceptibles and therefore may obtain lower rho scores simply as a function of lower levels of recall. In order to eliminate this problem Kihlstrom and Evans (1979) compared the rho scores of high and low susceptible subjects who had been matched at five different levels of re6 Kihlstrom (1975) modified the amnesia suggestion on the HGSHS:A so that subjects were asked to recall everything after the suggestion was lifted. Contrary to expectation, increases in amount recalled were not accompanied by increases in rho scores.
484
H. LORRAINE RADTKE AND NICHOLAS P. SPANOS
call. Using multiple / tests with a one-tailed found extent of organization with number level of significance, they reported that high of items recalled (Murphy, 1979). susceptibles obtained lower rho scores than lows when these groups recalled three items (p < .001), four items (p < .005), five items Reference Note (p < .05), and six items (p < .05). No differences were detected when seven to nine 1. Schwartz, W. Trance and personal history. Paper presented at 87th annual convention of the American items were recalled. The use of such liberal Psychological Association, New York, September statistical procedures is questionable. When 1979. the results are assessed with two-tailed t tests (still a liberal procedure considering that References multiple comparisons are being made), the differences between highs and lows for five K. S. Hypnosis for the seriously curious. Belitems and six items are no longer significant. Bowers, tnont, Calif.: Wadsworth, 1976. Although the reliability of these data clearly Cooper, L. M. Hypnotic amnesia. In E. Fromm & need to be more firmly established, their R. E. Shor (Eds.), Hypnosis: Developments in research and new perspectives (2nd ed.). Chicago: Almeaning would remain ambiguous regarddine-Atherton, 1979. less of statistical findings. These data con- Evans, F. J. Phenomena of hypnosis: 2. Posthypnotic cern relationships between rho scores and amnesia. In G. D. Burrows & L. Dennerstein (Eds.), susceptibility, not between rho scores and Handbook of hypnosis and psychosomatic medicine. Amsterdam: North-Holland Biomedical Press, 1980. amnesia. In summary, even if it had been reliably Evans, F. J., & Kihlstrom, J. F. Post hypnotic amnesia as disrupted retrieval. Journal of Abnormal Psydemonstrated that high susceptibles obchology, 1913,82, 317-323. tained lower rho scores than low susceptibles Evans, F. J., Kihlstrom, J. F., & Orne, E. C. Quantifying at R-Trial 1, methodological deficiencies subjective reports during posthypnotic amnesia. Proceedings of the 81st annual convention of the Amerwould have precluded firm conclusions conican Psychological Association, 1973,5, 1077-1078. cerning amnesia. In order to obtain inter(Abstract) pretable results concerning sequential or- Evans, F. J., & Orne, M. T. Motivation, performance, ganization and amnesia, we suggest two and hypnosis. International Journal of Clinical and Experimental Hypnosis, 1965, 13, 103-116. modifications in the construction of susceptibility scales. First, recall should be assessed Hilgard, E. R. Divided consciousness: Multiple controls in human thought and action. New York: Wiley, before the suggestion to allow determination 1977. of presuggestion differences in recall orga- Kihlstrom, J. F. The effects of organization and motinization. Second, after the suggestion is canvation on recall during post-hypnotic amnesia (Doctoral dissertation, University of Pennsylvania, 1975). celled, subjects should be required to recall Dissertation Abstracts International, 1975,3(5, 2473Beverything they can remember. This would 2474B. (University Microfilms No. 75-24082) enable accurate assessment of reversibility. Kihlstrom, J. F. Models of posthypnotic amnesia. AnAs an additional advantage, these changes nals of the New York Academy of Sciences, 1977, 296, 284-301. would allow for the measurement of subjective organization.7 In this way, future studies Kihlstrom, J. F. Context and cognition in posthypnotic amnesia. International Journal of Clinical and Excould assess the relationship between preperimental Hypnosis, 1978, 26, 246-267. sentation order and recall order before the Kihlstrom, J. F., & Evans, F. J. Recovery of memory amnesia suggestion, during it, and after canafter posthypnotic amnesia. Journal of Abnormal Psychology, 1976, 85, 564-569. celling it with a measure that does not con7
In their first study, Evans and Kihlstrom (1973) collapsed their results across subjects and obtained a single subjective organization score for high susceptibles and a single score for low susceptibies. The score for the highs was lower and was taken to indicate that highs as a group are less consistent in the way they order their recall than lows are. Unfortunately, there is no way to assess the statistical significance of such differences.
Kihlstrom, J. F., & Evans, F. J. Memory retrieval processes during post-hypnotic amnesia. In J. F. Kihlstrom & F. J. Evans (Eds.), Functional disorders of memory. Hillsdale, N.J.: Erlbaum, 1979. Maslach, C. Negative emotional biasing of unexplained arousal. Journal of Personality and Social Psychology, 1979, 37. 953-969. Murphy, M. D. Measurement of category clustering in free recall. In C. R. Puff (Ed.), Memory organization and structure. New York: Academic Press, 1979.
TEMPORAL SEQUENCING DURING POSTHYPNOTIC AMNESIA Nace, E. P., Orne, M. T., & Hammer, A. G. Posthypnotic amnesia as an active psychic process: The reversibility of amnesia. Archives of General Psychiatry, 1974, 31, 257-260. Radtke-Bodorik, H. L., Planas, M., & Spanos, N. P. Suggested amnesia, verbal inhibition, and disorganized recall for a long word list. Canadian Journal of Behavioral Science, 1980, 12, 87-97. Radtke-Bodorik, H. L., Spanos, N. P., & Haddad, M. G. The effects of spoken versus written recall on suggested amnesia in hypnotic and task-motivated subjects. American Journal of Clinical Hypnosis, 1979, 22, 8-16. Schwartz, W. S. Time and context during hypnotic involvement. International Journal of Clinical and Experimental Hypnosis, 1978, 26, 307-316. Schwartz, W. S. Hypnosis and episodic memory. International Journal of Clinical and Experimental Hypnosis, 1980, 28, 375-385. Shor, R. E., & Orne, E. C. The Harvard Group Scale of Hypnotic Susceptibility, Form A. Palo Alto, Calif.: Consulting Psychologists Press, 1962. Spanos, N. P., & Bodorik, H. L. Suggested amnesia and disorganized recall in hypnotic and task-motivated subjects. Journal of Abnormal Psychology, 1977,86, 295-305.
485
Spanos, N. P., & D'Eon, J. L. Hypnotic amnesia, disorganized recall, and inattention. Journal of Abnormal Psychology, 1980, 89, 744-750. Spanos, N. P., Radtke-Bodorik, H. L., & Stam, H. J. Disorganized recall during suggested amnesia: Fact not artifact. Journal of Abnormal Psychology, 1980, 89, 1-19. Spanos, N. P., Stam, H. J., D'Eon, J. L., Pawlak, A. E., & Radtke-Bodorik, H. L. The effects of social psychological variables on hypnotic amnesia. Journal of Personality and Social Psychology, 1980, 39, 737750. St. Jean, R., & Coe, W. C. Recall and recognition memory during posthypnotic amnesia: A failure to confirm the disrupted-search hypothesis and the memory disorganization hypothesis. Journal of Abnormal Psychology, 1981, 90, 231-241. Weitzenhoffer, A. M., & Hilgard, E. R. Stanford Hypnotic Susceptibility Scale, Forms, A and B. Palo Alto, Calif.: Consulting Psychologists Press, 1959. Weitzenhoffer, A. M., & Hilgard, E. R. Stanford Hypnotic Susceptibility Scale, Form C. Palo Alto, Calif.: Consulting Psychologists Press, 1962. Received March 9, 1981
