[Page 515 continued]
Gleaves (1996) was correct that role-playing studies (e.g., Spanos et al., 1985; Spanos, Weekes, Menary, & Bel1rand, 1986) do not by themselves demonstrate that DID is produced iatrogenically. Nevertheless, his assertion that "to conclude that these studies prove that DID is simply a form of role-playing is unwarranted" (Gleaves, 1996, p. 47) represented a misreading of these studies' purpose. These studies were designed not to reproduce the full range or subjective experience of DID symptoms, including multiple identity enactments, but rather to demonstrate the ease with which cues and prompts can trigger participants without DID to display the overt characteristics of this condition.
The SCM asserts that
As a consequence, the findings of role-playing studies provide a sufficiency proof that many of the overt features of DID can be reproduced following interviewer prompting. For example, Spanos et al. (1985) reported that most participants provided with suggestions for DID (e.g., "I think perhaps there might be another part of [you] that I haven't talked to") spontaneously reported amnesia for their alters following hypnosis, whereas no control participants did so. In addition, Spanos et al. found that many role-playing participants spontaneously adopted a different name, referred to their primary personality in the third person, and exhibited striking differences between their primary and alter "personalities" on self-report measures. All of these characteristics are commonly exhibited by DID patients (Ross, 1997). When situational demands are conducive, normal participants can readily role-play a number of characteristics of DID, including reports of physical, sexual, and satanic ritual abuse (Stafford & Lynn, 1998).
These findings demonstrate that the behaviors and reported experiences of DID patients are familiar to many members of the general population. Were this not the case, the SCM would not be able to account for a number of the features of DID. Analogue studies thus provide corroboration for one important and potentially falsifiable precondition of the SCM, although they do not provide dispositive evidence for this model.
Motivations for Developing DID
Gleaves (1996} asserted that an assumption of the SCM is that patients with DID enjoy having this disorder. According to
Gleaves, this assumption stems from the SCM's proposition that DID is largely maintained and in some cases partly caused by social reinforcement, such as attention from others. Nowhere, however, have Spanos (1994) or other proponents of the SCM posited that patients with DID "find having DID enjoyable or rewarding" (Gleaves, 1996, p. 45). To the contrary, proponents of the SCM emphasize that patients with DID often experience profound suffering. Spanos ( 1996), for example, described patients with DID as "chronically disturbed, unhappy, poly-symptomatic ... people who are emotionally needy" (p. 259).
Gleaves (1996) further maintained that the intense suffering experienced by many or most individuals with DID implies that reinforcement processes are largely irrelevant to the etiology and maintenance of this condition. Both behavioral and social leaming theorists, however, have long recognized that individuals often engage in pathological and psychologically painful behaviors as a consequence of reinforcement (e.g., see Hayes, Wilson, Gifford, Follette, & Strosahl, 1996, and Mowrer's  classic discussion of the "neurotic para-ox").
For example, many theorists have argued that a variety of forms of psychopathology can be conceptualized as resulting from short-term reinforcement at the expense of long-term suffering (see, e.g., Ullman & Krasner, 1975). To contend that reinforcement plays little or no role in the genesis of DID because the symptoms of DID are deeply distressing is no more logically defensible than to contend that the etiology of obsessive-compulsive disorder (OCD) is independent of reinforcement because OCD is intensely painful to its sufferers. In fact, there is compelling evidence that OCD is maintained and perlaps partly caused by reinforcement processes (Rachman & Hodgson, 1980).
Gleaves ( 1996) also did not discuss the hypothesis that much of the suffering of DID patients is iatrogenically induced. Indeed, a number of individuals who retracted reports of child abuse have reported that their condition deteriorated as they became increasingly dependent on their therapists and alienated from friends and relatives (de Rivera, 1997; Lief & Fetkowitz, 1995).
Gleaves's analysis overlooked the possibility that maladaptive and even subjectively distressing behaviors that might not appear to be reinforcing from the perspective of outside observers (e.g., displays of multiplicity) might nonetheless be reinforcing to clients with weak social support systems who have become intensely dependent on their therapists. Indeed, there is evidence that socially deprived individuals tend to find negative social attention more reinforcing than no attention at all (see, e.g., Gallimore, Tharp, & Kemp, 1969).
Child Abuse and DID
In his analysis of the literature linking child abuse to DID, Gleaves (1996) again cited Carson and Butcher's (1992) opinion: "while it is somewhat amazing that this connection [between DID and child abuse] was not generally recognized until1984, there is now no reasonable doubt about the reality of this association" (p. 208). Scrutiny of the literature reviewed by Gleaves, however, calls this conclusion into question.
Before we examine the child abuse-DID link, it is important to note that recent quantitative reviews raise questions concerning the magnitude of the association between child sexual abuse and later psychopathology. Specifically, the meta-analysis of Rind et al. ( 1998) suggests that the association between child sexual abuse and psychopathology may be
Moreover, Rind et al. reported a low effect size (.09) for the association between child sexual abuse and self-reported dissociative symptoms across eight studies (N = 1,324). The interpretation of Rind et al.'s findings and conclusions is potentially complicated, however, by the fact that their analyses were based on college sampels which were found by Jumper (1995) to yield smaller effect size for the relation between child abuse and psychopathology than do either community or clinical samples. In contrast, a separate meta-analysis by Rind and Tromovitch (1997) found comparably low effect sizes for this association in college and community sample
Despite the findings of Rind et al. (1998), we believe for at least two reasons that the issue of whether child abuse predisposes to DID remains an open question that merits further investigation.
First, it is conceivable that the relation between child abuse and psychopathology is pronounced in magnitude only among individuals who have experienced abuse that is severe, repeated, or both, although Rind et al. found that the frequency, duration, and forth of sexual abuse did not moderate the association between early abuse and later psychopathology.
Second, there is some evidence that self-reports of physical and sexual abuse may underestimate actual abuse rates (Widom & Morris, 1997; Widom & Shepar( 1996). A nontrivial rate of false negatives for child abuse might have attenuated the reported relations between child abuse and psychopathology in a number of studies. Nevertheless, the formidable methodological difficulties involved in operationalizing and assessing child abuse when it is mild or moderate in severity (see Rind et al., 1998), in corroborating abuse reports (e.g., see Schooler, Bendiksen, & Ambadar, 1997, for an illustration of some of the methodological complexities involved in corroborating child abuse reports), and in determining whether child abuse psychopathology correlations imply causation (DiLalla & Gotteman, 1991) demand a circumspect analysis of the evidence regarding the association between child abuse and DID.
In the following section, we separate our evaluation of the literature concerning the child abuse-DID link into two major issues:
The corroboration of child abuse reports among DID patients
Although Gleaves reviewed a number of studies suggesting a high prevalence of child abuse among DID patients (see Gleaves, 1996 Table 3, p. 53), in none of these studies was the abuse corroborated by independent sources. In Coons et al. (1988), Ross et al. (1990 Boon and Draijer (1993), and Ellason et al. (1996), the abuse reports were based exclusively on patient interviews, and in Putnam et al. (1986), Ross, Norton, and Wozney (1989), and Schultz et al. (1989), the abuse reports were based exclusively on clinicial questionnaires. The absence of corroboration for reported abuse in these studies (see also Scroppo et al., 1998) is problematic in view of recent findings indicating that memory is considerably more malleable, reconstructive, and vulnerable to suggestion than previously believed (Loftus, 1993, 1997a; Malinoski & Lynn, 1995). [*6]
In the therapeutic context, in which clinicians have ample and repeated opportunities to cue the emergence of abuse histories, the possibility of false memories is considerably more problematic.
Moreover, recent evidence suggests that memories of traumatic events (e.g., combat experiences) may not be immune to this problem (Southwick, Morgan, Nicolaou, & Charney, 1997).
Although the research of Pezdek, Finger. and Hodge (1997) indicated that memory implantation may be likely to occur only when the event being implanted is plausible and accords with script-relevant knowledge existing in memory, the relevance of their findings to early abuse reports requires clarification. Pezdek et al.'s findings might suggest that unintentional implantation of child abuse memories in DID patients can occur only when these patients possess implicit causal theories regarding the association between early abuse and DID, although this possibility has not been examined.
In addition, abuse memories recovered in therapy may be less likely to be veridical than abuse memories recalled continuously since childhood (Loftus, 1993), although there is little empirical evidence directly relevant to this assertion. Because none of the studies cited by Gleaves (1996, p. 53) provided information on whether the reported abuse was recalled continuously or recovered in treatment, this potentially important distinction cannot presently be addressed.
In addition, the phenomenon of "effort after meaning" whereby individuals interpret potentially ambiguous events (e.g., hitting, fondling) in accord with their implicit theories regarding the causes of their disorders, further renders some reports of relatively mild or moderate physical and sexual child abuse difficult to interpret without independent corroboration (Rind et al., 1998).
Finally, it is difficult to exclude the possibility that the same unintentional cues emitted by therapists that may promote the creation of alters might also promote the creation of false memories of abuse (Spanos, 1994), although little is known about the prevalence of suggestive practices among DID practitioners. As a consequence, it is not known whether the reported association between child abuse and DID might be at least partly spurious and contaminated by therapists' methods of ascertaining information.
Several investigators have, however, attempted to corroborate the retrospective abuse reports of DID patients. Gleaves (1996) cited the findings of Coons and Milstein ( 1986) and Coons ( 1994), who claimed to provide objective documentation for the abuse reports of a number of DID patients, as offering especially compelling support for the child abuse-DID link.
Close inspection of these studies, however, reveals various methodological shortcom-ngs. In neither study were diagnoses of DID made blindly of previous abuse reports. This methodological shortcoming is problematic because certain therapists might be especially likely to attempt to elicit features of DID among patients with a history of severe abuse. In the Coons (1994) study, DID diagnoses were made only after medical histories and psychiatric records (many of which may have contained information regarding abuse histories) had been reviewed. Moreover, because standardized interviews were not administered in Coons and Milstein (1986) and were administered only to an unknown number of participants in Coons (1994), the possibility of diagnostic bias is heightened. Finally, the patients in Coons (1994) "were diagnosed personally by the first author over an 11 year period" (p. 106). Because there is no evidence concerning whether these patients had DID prior to treatment, the possibility of iatrogenic influence is difficult to exclude. [*7]
Gleaves ( 1996) neglected or underemphasized several pieces of data that appear to call into question the veracity of some reports of child abuse in studies of DID and that underscore the importance of corroborating these reports.
In the study by Ross et al. (1991 ), 26% of DID patients reported being abused prior to age 3, and 10.6% reported being abused prior to age 1. Similarly, Dell and Eisenhower ( 1990) noted that 4 of 11 adolescent patients with DID reported that their first alter emerged at age 2 or earlier, and 2 of these patients reported that their first alter emerged between the ages 1 of 2. Memories reported prior to age 3 are of extremely questionable validity, and it is almost universally accepted that adults and adolescents are unable to remember events that occurred prior to age 1 (Fivush & Hudson, 1990). It is possible that the memories reported in these studies were accurate but that they were dated incorrectly. Nonetheless, the nontrivial percentages of individuals in Ross et al. (1991) and Dell and Eisenhower (1990) who reported abuse and the emergence of alters at very young ages raise concerns regarding the accuracy of these memories.
In this context, it is worth noting that Ross and Norton (1989) found that DID patients who had been hypnotized reported significantly higher rates of sexual and physical abuse than DID patients who had not been hypnotized. Because there is little evidence that hypnosis enhances memory (Lynn, Lock, Myers, & Payne, 1997), this finding is consistent with the possibility that hypnosis produces an increased rate of false abuse reports. Nevertheless, this conclusion must remain tentative in view of the absence of independent corroboration of the abuse reports and the correlational nature of Ross and Norton's data. [*8]
Gleaves ( 1996) contended that "there have been no cases in the scientific literature where the alleged abuse in a patient with DID was found to be totally fabricated" (p. 54). To maintain this position, Gleaves would be forced to argue that most or all of the memories of satanic ritual abuse that have been recovered by a large proportion of DID patients (estimated by Mulhern, 1991, to be 25% as of the mid-1980s) are veridical. Nevertheless, federal law enforcement officials have been unable to detect the existence of satanic cults (whose purported crimes involve multiple murders, cannibalism, and bizarre human sacrifices) despite years of intensive investigation (Bottoms, Shaver, & Goodman, 1996; Hicks, 1991; Lanning, 1989). Although it is conceivable that a subset of satanic ritual abuse reports represent the memory overlay of actual abuse incidents (Loftus, 1997b), the burden of proof would appear to rest on Gleaves and others, rather than on critics of the PTM, to provide documentation of such incidents.
Interpretation of the child abuse-DID association
Even if the child abuse reports of most DID patients could be corroborated, several important questions arise concerning the interpretation of these reports. In particular, it remains to be determined whether a history of child abuse is
With respect to the first issue, base rates and referral biases pose potential difficulties for Gleaves's interpretation of the abuse data. Because the prevalence of reported child abuse among psychiatric patients in general tends to be high (see, e.g., Pope & Hudson, 1992), these data are difficult to interpret without a psychiatric comparison group. Moreover, the co-occurrence between reported abuse and DID could be a consequence of several selection artifacts that increase the probability that individuals with multiple problems seek treatment.
Either or both of these artifacts could lead to the apparent relation between child abuse and DID discussed by Gleaves. Indeed, Ross ( 1991 ) found that nonclinical participants with DID reported substantially lower rates of child abuse than did patients with DID recruited from a clinical population. This finding is consistent with the hypothesis that selection biases account at least partly for the high levels of co-occurrence between reported child abuse and DID.
Moreover, Ross, Norton, and Fraser (1989) reported that American psychiatrists reported a substantially higher prevalence of child abuse among DID patients (81.2%) than did Canadian psychiatrists (45.5%). This finding suggests the possibility of biases in the assessment or elicitation of child abuse reports and raises questions concerning the claim that child abuse is necessary for most cases of DID (Spanos, 1994).
Gleaves (1996) dismissed [*3] Spanos's (1994) argument that the relation between child abuse and DID, even if shown to be genuine, is correlational in nature and could be a product of unidentified third variables, such as adverse family environment.
Gleaves likened the literature concerning the relation of child abuse and DID to the literature concerning the relation of trauma to PTSD: "The empirical support for the relationship between PTSD and trauma is also correlational. However, such a state of affairs would not set to be a convincing argument that PTSD is not a posttraumatic condition" (Gleaves, 1996, p. 53). But this analogy is questionable
Many studies have revealed dramatically increased rates of PTSD shortly after objectively documented events, such as Hurricane Andrew (Garrison, Bryant, Addy, & Spurrier, 1995) and the 1988 Armenian earthquake (Goenjian et al., 1994). Thus, although the relation between trauma and PTSD is correlational, the
provides compelling support for the assertion that this relation is causal, at least some cases.
The relation between child abuse and DID is markedly different: the traumatic event is often neither clear-cut nor readily corroborated by objective evidence. Nor are there data demonstrating that this event is unambiguously followed almost immediately by the signs and symptoms of DID.
Moreover,( Gleaves's assertion (1996, p. 55) that most patients with DID met criteria for PTSD borders on being tautological and begs the very question that is at issue: Is the child abuse genuine? If not, the diagnostic criteria for PTSD would not be satisfied, as this diagnosis requires exposure to a life-threatening or otherwise extremely dangerous event (APA, 1994).