Source: Journal of Psychosomatic Research
Volume 51, Issue 5, Pages 679-686
Date: November 2001

Information processing in chronic fatigue syndrome*1
A preliminary investigation of suggestibility

Jeannie D. DiClementi(a),(b), Karen B. Schmaling(c) and James F. Jones(d)
a Department of Psychology, University of Colorado at Denver, Denver, CO, USA
b Department of Medicine, University of Colorado School of Medicine, Denver, CO, USA
c Department of Psychiatry and Behavioral Sciences, University of Washington, Seattle, WA, USA
d Department of Pediatrics, National Jewish Medical and Research Center, Denver, CO, USA
*1 This study was conducted at the National Jewish Medical and Research Center, Denver, CO.
Corresponding author. Present address: Department of Psychology, Indiana-Purdue University-Fort Wayne (IPFW), Neff Hall Room 388, 2101 East Coliseum Boulevard, Fort Wayne, Indiana 46805-1499, USA. Tel.: +1-219-481-6403; fax: +1-219-481-6972; email:
Received 4 October 1999; accepted 13 March 2001 Available online 26 November 2001.


This study examines the effects of certain types of information processing on the subjective experience of cognitive deficits in persons with chronic fatigue syndrome (CFS). Two groups of participants, persons with CFS and a group of healthy controls, were administered a symptom inventory and measures of intellectual functioning, memory, automatic processing, and suggestibility. The groups differed significantly on number and severity of reported symptoms and on measures of global suggestibility and automatic processing, but not on measures of intellectual functioning and memory. Suggestibility was related to number and severity of reported symptoms, as well as the inability to inhibit the automatic processing of information. Implications of these findings are discussed, as well as directions for future research and treatment of symptoms associated with CFS.


Chronic fatigue syndrome (CFS) is characterized by severe and debilitating fatigue and includes other symptoms including difficulty concentrating, myalgia, and nonrefreshing sleep [1 and 2], which cannot be explained by the presence of well-characterized medical conditions such as cancer. CFS lacks a known, uniform pathognomic biological marker or pattern of signs; it is a diagnosis of exclusion based on patient symptom report. Occurrence of CFS in community samples has been estimated to be about 1% [3 and 4].

Some of the most commonly reported CFS symptoms are problems with cognitive function. Neuropsychological impairments are subtle in CFS; the best documented difficulties are in information processing, speed, and efficiency [5]. Subjective perceptions of cognitive difficulties are greater than documented impairments; emotional status (e.g., depression) may predict patients' perceptions of cognitive difficulties, but has not been well-associated with objective cognitive performance [5]. These associations warrant further investigation.

Few hypotheses have been postulated to explain the seeming disparity between subjective reports and objective data in CFS. The purpose of this study is to examine the role of three related constructs and phenomena, suggestibility, the nocebo response, and the misinformation effect as potential explanatory mechanisms for symptom reporting in CFS. Placebo reactions occur in the absence of known biological causes or effects. The more specific term, nocebo response, refers to negative reactions to a placebo, such as onset of an illness after receiving an inert substance or intervention [18]. In the literature on placebo response, placebos have been demonstrated to modify both subjectively reported and objectively observed symptoms [6]. For example, placebos have been shown to cause weakness, palpitations, rashes, diarrhea, and edema [7], vomiting, tremor, pallor, hives, and changes in blood pressure [8], as well as anaphylactic reactions [9]. Individual personality variables and expectations have been associated with placebo responding. The so-called placebo reactor has been characterized as more suggestible [10] than the nonreactor.

Suggestibility is defined as the tendency to respond to suggestions. Synonymous with hypnotizability in the literature, suggestibility is characterized by response to suggestions and by the perception that the response behavior is involuntary. For example, during a typical measure of suggestibility, the Hypnotic Induction Profile (HIP) [11], it is suggested that the subject's arm will float upward like a balloon. Of course, the subject raises the arm, but the suggestible person will experience the arm as floating into the air involuntarily and will report it as such. Dixon et al. [12] explain this phenomenon in terms of automatic versus strategic processing. They argue that the highly suggestible individual engages in the selective and automatic processing of a suggestion that the arm is getting lighter, i.e., focusing so selectively on the arm to the exclusion of other stimuli and therefore experiencing it as involuntary. Hilgard and Hilgard [13] reported, for example, that during hypnotic analgesia, highly suggestible subjects showed physiological reactions to painful stimuli in the absence of pain perception. The less-suggestible person, on the other hand, is more strategically engaging in reality-testing procedures by consciously attending to and processing the source of the suggestion. Crawford et al. [14] point out that automatic processing by the suggestible person is due to increased attentional focus involving disattention to extraneous stimuli along with a decrease in reality testing.

Response expectancy is the hypothetical explanatory phenomenon in both placebo effect and hypnotic response [15]; research has shown that nonvolitional responses can be generated through expectancy of their occurrence. In these expectancy conditions, alterations of such responses as nausea, pain perception, sexual arousal, and anxiety, for example, have been observed and supported by corresponding physiological changes [15]. Expectations of enhanced performance on motor skill tasks due to caffeine intake was correlated with actual improved performance for a group of college students receiving no caffeine [16]. In the same study, subjects expecting caffeine-impaired performance after receiving the placebo actually performed more poorly than the control group receiving no beverage and no expectation instructions [16]. Other studies have demonstrated adverse side effects of medications secondary to expectations of such [17] and patient improvement following the administration of a homeopathic treatment. There is also a growing body of evidence connecting placebo responding to onset of illness, referred to as a nocebo response [18].

Along these same lines, research has shown that suggesting erroneous information subsequent to an event often results in erroneous reporting about that event [19]. The research on this `misinformation effect' has indicated that the erroneous information is incorporated into the subject's memory of the event. The typical approach of such research has been to have the subjects experience a stimulus event, often in the form of slides of an accident or crime scene, then receive false information about the event. On a recall measure designed to assess the subject's acceptance of the erroneous information, many subjects in the misinformation condition report the phony information as having been part of the event. In fact, some studies have reported differences in reporting accuracy between misinformed and control subject groups as large as 30% or 40% [20].

Two variables enhance the effects of the false information on recall [20]. Length of the delay between presentation of the event and the presentation of the misinformation is correlated with the degree of erroneous reporting of the event: the greater the delay, the greater the likelihood the subject will accept the false information as accurate. In addition, the degree of subtlety of the erroneous information also is positively related to the degree of accepting it as accurate, i.e., the more subtle the presentation, the more likely it will be incorporated into the subject's recall of the event.

The misinformation effect may be similar to the placebo response, particularly with regard to physical health concerns. If an individual is predisposed to being concerned about physical status, s/he may be more susceptible to suggestions about physiological events, no matter how minor the event or inaccurate the suggestion. For example, people with panic disorder appear to have heightened somatic focus. Interoceptive cues such as a detectable, albeit normal, change in heart rate can trigger a vicious cycle of fear, leading to further changes in heart rate, that may culminate in a full blown panic attack. According to Barlow [21], the individual then rapidly becomes increasingly hypervigilant concerning somatic cues. Such sensitivity to somatic cues demands much of the individual's attentional focus as well as serves to precipitate further physiological changes [21].

CFS patients typically focus a great deal of attention on their internal states [22]. When somatic information, e.g., perceived muscle tenderness, is consistent with the belief that such sensations result from disease, it is more readily, or automatically, processed [22]. The misinformation effect [19 and 20] may inadvertently occur during standard medical evaluations of poorly understood conditions, such as CFS, which lack objective (i.e., biological) markers. For example, after a delay between the onset of symptoms and the visit to the doctor's office, the patient is presented with questions about the specific symptoms that may or may not apply (a potential misinformation condition). Subsequently, during the course of treatment, the patient is often asked to recount symptoms of the illness. Repeated recall of the information (rehearsal) may serve to strengthen the encoding of the information into long-term memory regardless of the accuracy of the information.

The placebo response condition may be an appropriate model to understand some aspects of CFS. Patients with CFS resemble the profile of placebo reactors: They are concerned about physical symptoms and exhibit vigilance to internal sensations [22]. Patients with poorly understood physical conditions and a seemingly unpredictable course, may be more likely to be vigilant than patients with more consistent, predictable conditions. Further, patients with CFS complain of concentration and memory difficulties. These two factors may interact in important ways: First, when a person is unsure of her/his own ability to recall details of an event, accepting information from another source, particularly a high status figure [23] such as a physician, may be a way of compensating for the perceived memory deficits. Second, with increased attention selectively devoted to monitoring internal states, the patient with CFS is limited in her/his ability to attend to extraneous stimuli [14] that may offer alternative explanations for changing internal states. Third, there is ample evidence that patients with CFS have difficulties processing complex cognitive information [5], a characteristic consistent with research on hypnotizability and the selective attentional focusing of highly hypnotizable subjects [14 and 24].

The present study, then, will examine the associations between suggestibility and cognitive functioning among patients with CFS, as compared with healthy patients. The hypotheses are that (1) patients with CFS will show more suggestibility than healthy persons; (2) patients with CFS will perform more poorly on neuropsychological measures that require inhibition of automatic processing but will not perform differently from healthy persons on memory tasks. Research has shown that, in general, memory per se is not affected in CFS [5], therefore, we predict that the groups will be similar in terms of overall memory functioning. Previous research has shown the Stroop test to be a reliable measure of the automatic processing associated with suggestibility [12], and we predict that the CFS group will perform more poorly on the Stroop than the healthy controls; (3) automatic processing will be positively related to suggestibility for patients with CFS as compared to healthy persons; and (4) greater suggestibility will be associated with more numerous and severe psychological symptoms.



Twenty-one patients with CFS were recruited from the outpatient CFS clinic at the National Jewish Medical and Research Center (NJMRC). Selection of the participants began with a standardized review of medical records of CFS clinic patients for those meeting the 1988 case definition of CFS [1]: Utilizing a checklist of CFS symptoms, records were examined for evidence of major CFS symptoms, at least 8 of 11 minor symptoms, and absence of any exclusionary diagnoses. Participants were administered the Diagnostic Interview Schedule for DSM-IIIR [25] to screen for psychiatric diagnoses that would exclude them from consideration of CFS as a diagnosis. The exclusions included any current or preexisting Axis I diagnosis. All participants also met the 1994 case definition of CFS [2]. The 1988 CFS case definition was applied because it is more conservative in terms of psychiatric exclusions, which would lead to increased homogeneity within and between the groups of subjects, thereby also reducing the likelihood that any group differences in the dependent variables would be attributable to or confounded by group differences in psychopathology.

Twenty-one healthy control subjects were also recruited from hospital research and nursing staff and screened in a similar manner. Healthy controls had no current or recent history of medical illnesses, including a history of CFS symptoms. Healthy controls were also administered the Diagnostic Interview Schedule for DSM-IIIR [25] to screen for psychiatric diagnoses, and excluded on the basis of any Axis I diagnosis.


Participants completed two suggestibility measures, three tests of cognitive functions and a global measure of psychological symptoms. The measures of cognitive function were used to characterize premorbid intellectual function and assess current cognitive functioning.


The suggestibility task was patterned after studies designed by Loftus et al. [19 and 26]. Participants were read a neutral story that was presented as a standard story memory test. After hearing the story, subjects were given a questionnaire purportedly measuring their recall of the story (Recall I). The questionnaire actually contained false information about the story embedded in the questions. The false information was divided into half health-related and half non-health-related items. After a delay, during which several of the other measures were administered, subjects were given a recall measure (Recall II) that consisted of forced-choice questions (real vs. false items) to measure to what extent they incorporated the false information into their reports of events depicted in the neutral story. There were eight total misinformation items (four health-related and four non-health-related items), so the participants' misinformation scores ranged from 0 to 8 for the total score, and 0 to 4 for the health- and non-health-related scores.


The HIP is a brief structured technique that measures behavioral, perceptual, and cognitive responses to suggestions. Briefly, subjects are asked to perform an eye roll and then once their eyelids are closed, the hypnotic induction is begun in the form of an imagery technique guided by the examiner. The induction involves suggestions of floating and arm levitation. Postinduction ratings include subjects' reports of physical sensations, perceptions of dissociation, and involuntariness of actions. An induction score is obtained based on the number of reinforcers required for arm levitation, the arm levitation, reported experience of dissociation and floating, amnesia to the cut-off signal, all of which contribute to an overall experience of involuntariness of actions. Induction scores range from 0 to 12. An eye roll score is also obtained, the eye roll being a relatively stable indicator of suggestibility [11]. Briefly, the participant is instructed to gaze upward as far as possible while holding the head in a horizontal position. The participant is then instructed to slowly close the eyelids while holding the upward gaze. Eye roll scores are calculated by measuring the amount of sclera visible between the lower edge of the iris and the lower eyelid and by measuring the amount of squint that occurs while closing the eyelid. Eye roll scores range from 0 to 7. Both the induction score and eye roll score are indicators of suggestibility, i.e., the higher the score, the greater the level of suggestibility.


This test assesses the ability to shift perceptual sets to conform to changing demands, suppressing an automatic response while producing a controlled response, and completing the task while distracted. This test involves two identical stimulus sheets, which consist of 112 color names arranged in columns, and printed in one of four colors. No word is printed in its matching color. For the first trial, subjects are timed while reading aloud the printed words. For the second trial, subjects are timed while naming the color of ink in which the word is printed, and not reading the word. The Stroop yields two scores: a raw score and a percentile rank for same-aged peers. Research using the Stroop test has also been used to differentiate between low-hypnotizable and high-hypnotizable subjects. Highly hypnotizable subjects show a significantly greater automaticity of processing information [12 and 24].

NORTH AMERICAN ADULT READING TEST (NAART) [28] This test was given to assess intellectual functioning and yields a full-scale IQ score. The test consists of 61 irregularly spelled words printed in two columns on both sides of an 8.5-11 in. card, which is given to the subject to read. Blair and Spreen [28] reported that the correlation between the WAIS-R FSIQ and NAART-predicted FSIQ is .75.


Research using the WMS-R has shown it to be sensitive to memory disturbances in a number of patient groups [29] and thus is a useful screening tool. The Logical Memory Subtest assesses the ability to recall ideas in two orally presented stories and has extensive scoring instructions that result in few scoring ambiguities [30]. The first story of this subtest of the WMS-R was used as a screen for memory difficulties to control for gross between-group differences in memory function.


This is a 53-item self-report measure of emotional, somatic, perceptual, and cognitive symptoms with well-established psychometric characteristics. For purposes of this study, two summary scores were used: the Positive Symptom Index (PSI), a measure of the number of symptoms reported, and the General Severity Index (GSI), which measures the depth of distress experienced.


After being selected according to CFS case definition criteria, each person was contacted by the investigator and asked to participate in the study after the procedures were described. Those who agreed were then scheduled to come into the laboratory. Each subject took about 1 hour to complete the tests. All participants were tested in the afternoon. Procedural steps were as follows:

1. Subjects reported to NJMRC and were given the informed consent form to read and sign; the forms explained the nature of the study with the exception of the suggestibility manipulation.
2. Subjects were then administered the following:
(a) Suggestibility (misinformation) story; (c) Stroop test;
(d) NAART;
(e) Brief Symptom Inventory;
(f) HIP;
(g) Suggestibility Recall II; and
(h) WMS-R, Logical Memory Subtest (Immediate Recall Only).
3. At the conclusion of their participation, subjects were debriefed in both written and verbal form. The debriefing provided full disclosure of the purpose of the study and included an explanation of the deception involved.
4. Subjects were paid for their participation.


Descriptive statistics were used to describe the demographic characteristics of the groups. t tests were used to test for group differences in demographic variables (age and years of education). If homogeneity of variance assumptions were not met, the results based on unequal variance t values were reported. One-way analyses of covariance were used to test for group differences on the measures of suggestibility and cognition, controlling for any demographic variables that differed by group. Finally, correlations were calculated between the suggestibility measures and measure of automatic processing (Stroop test) and psychological symptoms.



The CFS group (n=21) consisted of 3 male and 18 female subjects. Of the 21 CFS patients, one was of Hispanic origin, the remaining were white, not of Hispanic origin. The ethnic and gender composition of this subject group approximated that of the larger population of CFS patients seen at NJMRC. Mean age of the CFS group was 40.19 (S.D.=7.2). Mean years of education of the CFS group was 14.48 (S.D.=2.04).

The healthy control group (n=21) was recruited from hospital nursing and research staff and consisted of 4 male and 17 female subjects, all of whom were white, not of Hispanic origin. Mean age of the control group was 33.85 (S.D.=6.7). Mean years of education of the healthy control group was 15.52 (S.D.=2.67).

The healthy control group was significantly younger than the CFS group [t(40)=2.95, P<.01]1. Therefore, age was used as a covariate in all subsequent group comparisons. The between-groups difference in years of education was nonsignificant [t(40)=-1.42, ns].


It was predicted that CFS patients would demonstrate greater suggestibility. As shown in Table 1, patients with CFS had significantly higher eye roll scores than the healthy controls [F(1,39)=36.56, P<.001], after controlling for age. In addition, the CFS group's mean eye roll score was notably greater than the expected mean eye roll score of 2.4 for highly hypnotizable persons reported by Spiegel and Spiegel [11]. The induction scores followed a similar pattern, showing significantly higher [F(1,39)=37.57, P<.001] average induction scores for the CFS group than for the control group. Thus, the participants with CFS showed significantly greater suggestibility on a global measure of suggestibility than the healthy control group.

For the misinformation task, between-groups differences in total number of errors and nonsomatic errors were nonsignificant. Inspection of the data in Table 1 suggests that subjects with CFS were more likely to accept somatically related misinformation than healthy control subjects, but this effect could not be tested statistically because of the lack of variability among the scores for the latter subjects.


CFS patients and healthy controls did not differ significantly on either the WMS-R logical memory immediate recall or the NAART estimated FSIQ. Thus, the two groups were equivalent in terms of both intellectual functioning and memory. There was a significant between-groups difference in which the CFS group performed more poorly than the healthy controls on a measure of automatic processing, the Stroop test [27] [F(1,39)=7.03, P<.05], after controlling for age. In addition, the CFS group's mean raw score of 96 fell below the impairment cutoff of 98, placing them in the impaired range of functioning. These data are presented in Table 2.


Automatic processing was significantly related to scores on suggestibility measures. For the entire sample, induction scores were negatively correlated with both Stroop raw scores (r=-.40; P<.05) and with Stroop percentile scores (r=-.26; P<.05), such that the level of performance on the Stroop decreased as induction scores increased. Correlations between eye roll scores and Stroop scores were nonsignificant. Stroop raw scores were negatively correlated with somatic errors (r=-.62; P<.01), such that impaired performance on the Stroop was associated with greater numbers of somatic errors on the misinformation task. Correlations between errors and Stroop percentiles were not significant. Eye roll scores were positively correlated with number of somatic errors on the misinformation task (r=.30; P<.05), such that the greater the eye roll score, the greater the number of somatic errors. These results lend support to the assertion that greater suggestibility is related to the inability to inhibit automatic processing of information.

Both general and specific measures of suggestibility were related to the number and severity of psychological symptoms. As shown in Table 3, the eye roll score was positively correlated with the number and severity of symptoms endorsed (both PSI and GSI) of the Brief Symptom Inventory. Induction score was positively correlated with both the PSI and the GSI of the Brief Symptom Inventory. In addition, somatic errors from the misinformation task were positively correlated with the number of symptoms.


The data supported the first hypothesis that the participants with CFS were more suggestible than healthy persons on a measure of general suggestibility, the HIP [11]. Support for the more specific misinformation effect was mixed. The CFS group made more somatic errors than the control group, the controls having made no somatic errors. Group differences in both nonsomatic errors and total errors were nonsignificant. It is possible that the methodology used in the present study attenuated the robustness of the misinformation task. Loftus [20] points to elapsed time between the presentation of the event and presentation of the misinformation as an important variable in the acceptance of the false information. In the present study, the elapsed time between presentation of the story and presentation of the false information was minimal, thus possibly interfering with the misinformation effect.

There were no differences between participants with CFS and healthy controls on both the global screen of memory functioning and the measure of intellectual functioning. While the primary purpose of the study was to investigate suggestibility, not neuropsychological performance per se, our results are consistent with past research that CFS patients typically demonstrate intact functioning on these measures [5] but have difficulties with other tasks, including speed-dependent information processing (e.g., [32]) and verbal memory involving list learning [33]. However, on the measure of automatic processing, the Stroop test, the CFS group performed significantly more poorly than the control group and fell in the impaired range of performance for same-aged peers. This measure is a timed task and as patients with CFS have difficulties with speed-dependent information processing, a goal for future research would be to untangle the relative contributions of speed-dependent processing and automatic processing from these patients' performance difficulties. In summary, these results supported our second hypothesis that subjects with CFS performed more poorly on a specific measure of automatic processing than control subjects, but the groups did not differ on global measures of intellectual or memory function.

In addition, consistent with our third hypothesis, subject performance on the Stroop was significantly negatively associated with a measure of global suggestibility, the induction score of the HIP. That is, the inability to inhibit automatic processing was associated with greater suggestibility. Finally, suggestibility was also associated with increased numbers and severity of psychological symptoms, as had been posited as the fourth hypothesis. These results of the associations among suggestibility, automatic processing, and psychological symptoms, while statistically significant, left the majority of the variance in suggestibility unaccounted for by these other factors. Future research should examine further the role of suggestibility as a potentially relevant information processing style among persons with medically unexplained illnesses such as CFS.

Results of this preliminary investigation appear to offer support for the role of suggestibility among some persons with CFS. The nocebo effect, or the onset of an illness after an inert intervention, may occur among some persons with CFS, although this speculation awaits examination using longitudinal data in future studies. Placebo responding has long been well documented in the literature, particularly with regard to the effects and side effects of medications or medical treatments. However, less is understood regarding the onset or maintenance of illness as the result of placebo response. The participants with CFS in the present study did demonstrate the selective and automatic processing of the suggestions during administration of the HIP, findings also supported by results from the Stroop test. The subjects could not inhibit their automatic processing in the face of the distracting or inaccurate information with the Stroop or the HIP, respectively. Stated alternatively, patients with CFS have relative difficulties attending to relevant, if peripheral, information. This effect could, in turn, be subjectively experienced as memory or cognitive deficits.

It is important to emphasize here that suggestibility is not synonymous with gullibility as some might infer. Rather, suggestibility refers to a type of information processing that is selective and distorted in that relevant information cannot be attended to and incorporated in a balanced fashion. Suggestible persons with CFS are certainly experiencing the symptoms they report. However, a tendency to focus on symptoms, perhaps to the exclusion of other information, may be related to CFS patients' difficulties in processing complex cognitive information, and difficulties with divided attention [33].

The methodological issues in the current study, e.g., sample size, lack of delay between presentation of the misinformation story and the false information, use of an analogue laboratory task rather than a more naturalistic examination, etc., demand that these data be interpreted with caution. However, these results do suggest directions for future research, in addition to those mentioned above. The inclusion and comparison of additional groups with qualities that are hypothetically shared with or distinct from CFS would be an extremely interesting direction for future research, as would be the examination of the longitudinal relationship between suggestibility and illness. The study of suggestibility among patients with CFS as compared to other medically unexplained illnesses or illnesses with a relapsing and remitting course may help elucidate the possible mechanisms underlying our observed associations of suggestibility among persons with CFS: For example, does uncertainty regarding the cause or the course of the illness increase suggestibility, or does suggestibility increase the likelihood of medically unexplained illness? To the extent that suggestibility and automatic processing characterize CFS, treatment of CFS symptoms may be enhanced by cognitive rehabilitation that includes practicing divided attention and focused attention tasks, cognitive reframing strategies (e.g., to suggest alternate, benign interpretations for somatic sensations, such as feeling tired after walking being a sign of deconditioning, rather than of further injury), and/or hypnosis, to refocus the patients' attention to the nonsomatic realm, or employ suggestions for enhanced functioning.


This study was supported in part by a grant from the National Institute of Allergy and Infectious Diseases, U01-AI32244. The authors would like to thank C. Munro Cullum, PhD, for his assistance in the development of this study.


Table 1. Group differences on measures of suggestibility
Hypnotic induction profile Misinformation task
Eye roll* Induction* Somatic Nonsomatic
M S.D. M S.D. M S.D. M S.D.
CFS 3.43 0.75 9.10 2.73 0.19 0.51 0.38 0.58
Controls 1.33 1.11 2.33 3.17 0.00 0.00 0.52 0.68
* P<.001.

Table 2. Group differences in global memory, intellectual functioning, and automatic processing
WMS memory FSIQ Stroop raw *
M S.D. M S.D. M S.D.
CFS 12.81 3.19 107.91 7.77 96.15 17.75
Controls 14.71 3.36 112.52 6.91 106.95 8.86
* P<.05.

Table 3. Correlations between the measures of suggestibility and psychological symptoms
Eye roll .409* .597*
Induction .475* .405*

Misinformation task
Somatic errors .160 .524*
Nonsomatic errors ^B .085 .004
* P<.01.


1. GP Holmes, JE Kaplan, NE Gantz, AL Komaroff, LB Schonberger, SE Straus, JF Jones, RE Dubois, C Cunningham-Rundles, S Pahwa, G Tosato, LS Zegans, DT Purtilo, N Brown, RT Schooley and I Brus, Chronic fatigue syndrome: a working case definition. Ann Intern Med 108 (1988), pp. 387-389.
2. K Fukuda, SE Strauss, I Hickie, MC Sharpe, JG Dobbins and AL Komaroff, The international chronic fatigue syndrome study group. Chronic fatigue syndrome: a comprehensive approach to its definition and study. Ann Intern Med 121 (1994), pp. 953-999.
3. WJ Gunn, DB Connell and B Randall, Epidemiology of chronic fatigue syndrome: the Centers for Disease Control study. In: Chronic fatigue syndrome. Ciba Foundation Symposium vol. 173, Wiley, Chichester (1993), pp. 83-101.
4. D Buchwald, D Garrity, R Pascualy, P Kith, RL Ashley, MH Wener, PG Kidd, WJ Katon and JE Russo, Chronic fatigue syndrome. Toxicol Ind Health 8 (1992), pp. 157-173.
5. LA Tiersky, SK Johnson, G Lange, BH Natelson and J DeLuca, Neuropsychology of chronic fatigue syndrome: a critical review. J Clin Exp Neuropsychol 19 (1997), pp. 560-586.
6. AK Shapiro, The placebo response. In: JG Howells, Editor, Modern perspectives in world psychiatry, Oliver and Boyd, Edinburgh (1968), pp. 596-619.
7. MP Jensen and P Karoly, Control theory and multiple placebo effects. Int J Psychiatry Med 15 (1985), pp. 137-147.
8. L. Lasagna, The placebo effect. J Allergy Clin Immunol 78 (1986), pp. 161-165.
9. SW Perry and G Heidrich, Placebo response: myth and matter. Am J Nurs (1981), pp. 720-725 April.
10. RM Steinbrook, MB Jones and JD Ainslie, Suggestibility and the placebo response. J Nerv Ment Dis 140 (1965), pp. 87-91.
11. H Spiegel and D Spiegel. Trance and treatment: clinical uses of hypnosis, American Psychiatric Press, Washington, DC (1978).
12. M Dixon, A Brunet and J-R Laurence, Hypnotizability and automaticity: toward a parallel model of hypnotic responding. J Abnorm Psychol 99 (1990), pp. 336-343.
13. ER Hilgard and JR Hilgard. Hypnosis in the relief of pain (2nd ed ed.),, William Kaufman, Los Altos, CA (1983).
14. HJ Crawford, AM Brown and CE Moon, Sustained attentional and disattentional abilities: differences between low and highly hypnotizable persons. J Abnorm Psychol 102 (1993), pp. 534-543.
15. I. Kirsch and JR Council, Response expectancy as a determinant of hypnotic behavior. In: NP Spanos and JF Chaves, Editors, Hypnosis: the cognitive-behavioral perspective, Promethesus Books, Buffalo, NY (1989), pp. 360-379.
16. M Fillmore and M Vogel-Sprott, Expected effect of caffeine on motor performance predicts the type of response to placebo. Psychopharmacology 106 (1992), pp. 209-214.
17. JL Strauss and S Cavanaugh, Placebo effects: issues for clinical practice in psychiatry and medicine. Psychosomatics 37 (1996), pp. 315-326. 18. RA Hahn, The nocebo phenomenon. In: A Harrington, Editor, The placebo effect: an interdisciplinary exploration, Harvard Univ. Press, Cambridge, MA (1997), pp. 56-76.
19. EF Loftus, K Donders, HG Hoffman and JW Schooler, Creating new memories that are quickly accessed and confidently held. Mem Cognit 17 (1989), pp. 607-616.
20. EF Loftus, When a lie becomes memory's truth: memory distortion after exposure to misinformation. Curr Dir Psychol Sci 1 (1992), pp. 121-123.
21. DH Barlow. Anxiety and its disorders: the nature and treatment of anxiety and panic, The Guilford Press, New York (1988).
22. GC Wood, RP Bentall, M Gopfert, ME Dewey and RH Edwards, The differential response of chronic fatigue, neurotic and muscular dystrophy patients to experimental psychological stress. Psychol Med 24 (1994), pp. 357-364.
23. RF Bornstein, The dependent personality: developmental, social, and clinical perspectives. Psychol Bull 112 (1992), pp. 3-23.
24. M Dixon and J-R Laurence, Hypnotic susceptibility and verbal automaticity: automatic and strategic processing differences in the Stroop color-naming task. J Abnorm Psychol 101 (1992), pp. 344-347.
25. LN Robins, JE Helzer, J Croughan and KS Ratcliff, The NIMH diagnostic interview schedule: its history, characteristics, and validity. Arch Gen Psychiatry 38 (1981), pp. 381-389.
26. EF Loftus, DG Miller and HJ Burns, Semantic integration of verbal information into a visual memory. J Exp Psychol: Hum Learn Mem 4 (1978), pp. 19-31.
27. MR Trenerry, B Crossen, J DeBoe and WR Leber. Stroop Neuropsychological Screening Test: manual, Psychological Assessment Resources, Odessa, FL (1989).
28. JR Blair and O Spreen, Predicting premorbid IQ: a revision of the national adult reading test. Clin Neuropsychol 3 (1989), pp. 129-136.
29. O Spreen and E Strauss. A compendium of neuropsychological tests: administration, norms, and commentary, Oxford Univ. Press, New York (1991).
30. D Wechsler. Wechsler Memory Scale-Revised: manual, The Psychological Corporation, San Antonio (1987).
31. LR Derogatis and N Melisaratos, The Brief Symptom Inventory: an introductory report. Psychol Med 13 (1983), pp. 595-605.
32. J DeLuca, SK Johnson and BH Natelson, Information processing efficiency in chronic fatigue syndrome and multiple sclerosis. Arch Neurol 50 (1993), pp. 301-304.
33. V Michiels, R Cluydts and B Fischler, Attention and verbal learning in patients with chronic fatigue syndrome. J Int Neuropsychol Soc 4 (1998), pp. 456-466.

(c) 2001 Elsevier B.V. / ScienceDirect