Suppressing unwanted thoughts can lead to an increased occurrence of the suppressed thought in dreams. This is explainable by the ironic control theory, which theorizes why the suppression of thoughts might make them more persistent. The present study examined the influence of thought suppression on dream rebound, dream distress, general psychiatric symptomatology, depression, sleep quality and perceived stress. Thirty healthy participants (good sleepers) were investigated over a period of 1 week. Half were instructed to suppress an unwanted thought 5 min prior to sleep, whereas the other half were allowed to think of anything at all. Dream content was assessed through a dream diary. Independent raters assessed whether or not the dreams were related to the suppressed target thought. The results demonstrated increased target-related dreams and a tendency to have more distressing dreams in the suppression condition. Moreover, the data imply that thought suppression may lead to significantly increased general psychiatric symptomatology. No significant effects were found for the other secondary outcomes.
Suppressing unpleasant thoughts can be tempting, as associated negative emotions may be avoided. However, ironic control theory (Wegner et al., 1987) proposes that suppressing thoughts leads to an increased occurrence of the suppressed content in waking states. According to Wegner et al. (2004), the interplay of two processes is responsible for this effect: first, a conscious operating process that is responsible for mental deflection by concentrating on content other than the target thought, and secondly, an unconscious monitoring process that controls for unwanted thoughts or intrusions, in order to test whether the operating process is needed. The operating process requires greater cognitive capacity. As soon as capacity is reduced, for example by cognitive demands (or sleep), the operating process is ‘weakened’ and unwanted thoughts become more accessible, because the monitoring process continues its activity.
Previous studies on psychology students found that any attempted suppression of unwanted thoughts prior to sleep leads to increased dreaming about that target; the so-called dream rebound effect (Schmidt and Gendolla, 2008; Taylor and Bryant, 2007; Wegner et al., 2004).
This implies that ironic control theory might be applicable to explain the occurrence of bad dreams or nightmares. A recent study on first-year psychology students showed that suppression-induced dream rebound prior to sleep can even be enhanced by cognitive load, such as learning a nine-digit number (Bryant et al., 2011). In accord with the ironic control theory, these findings suggest that cognitive load weakens the operating process and unwanted content becomes more accessible.
There is convergent evidence to support the dream rebound effect and the proposal that thought suppression leads to increased occurrence of the suppressed content in dreams. For example, patients with insomnia report having insomnia-related dream contents while trying to suppress worries about their sleep quality prior to falling asleep (Riemann et al., 2012), or people with post-traumatic stress disorder (PTSD) report dreams on their traumatic experience (for reviews see Mellman et al., 1995; Phelps et al., 2008) and abstaining smokers dream frequently of smoking (Hajek and Belcher, 1991). The aim of the present study was to replicate and extend previous findings on thought suppression prior to sleep, and to investigate the effect of thought suppression on dream content, dream distress, general psychiatric symptomatology, depression, sleep quality and perceived general distress in participants other than psychology students. In order to detect whether the effect is replicable over a longer period of time, participants were investigated over 1 week, instead of 1 or 2 days (as before). We hypothesized that after 1 week the suppression group would have dreamt more often of the target thought and would have scored higher on dream distress compared to a control group. Secondly, we expected a deterioration of general psychiatric symptomatology, depression, sleep quality and perceived stress.
The study was conducted at the Department of Clinical Psychology at the Goethe University of Frankfurt. The study protocol was approved by the local ethics board of the Faculty of Clinical Medicine Frankfurt; written informed consent was obtained from all participants.
A total of 30 mentally healthy volunteers [good sleepers, including 19 women, mean age 39.5; standard deviation (SD) = 14.31] were recruited via general media. Each received €20 as compensation. Exclusion criteria were: chronic nightmares, sleep disorder, mental retardation, schizophrenia, bipolar disorder, PTSD, body mass index (BMI) < 17, substance abuse, studying psychology or being a psychologist and intake of psychotropic drugs within the past 6 months. Participants were assigned to the groups using block randomization.
The study included two assessment points within 1 week, one prior to the intervention (T1) and one afterwards (T2).
During the first session, participants were run through the clinical interviews SCID I and II to secure compliance with inclusion and exclusion criteria (Structured Clinical Interview for DSM-IV Axis I and DSM-IV Axis II Personality Disorders; German versions: Fydrich et al., 1997; Wittchen et al., 1997). A psychologist with a bachelor degree (the second author) had been trained to carry out the clinical interviews.
In order to assess primary outcomes (mean number of target-related dreams; mean value of distress caused by dreams, both per week), participants had to complete a dream diary during 1 week, where they recorded their dream content in note form and the extent of perceived dream distress on a 10-point rating scale (0 = ‘not at all’; 10 = ‘very much/often’). Whether or not dream contents were related to the target thought was coded by the mean of two independent raters, who were blind to the participants group and any other personal information (0 = ‘no relation’; 1 = ‘accordance between dream content and intrusive thought’). Furthermore, participants were asked whether they had completed the task during the night before and to rate their task compliance (4-point rating scale; 1 = ‘not at all’; 4 = ‘totally’). Also, they were required to keep records about problems falling asleep, night-time awakenings and any physical reactions (0 = ‘no’; 1 = ‘yes’), as well as the amount of distress caused by the dream (10-point rating scale; 0 = ‘not at all’; 10 = ‘very much’).
For secondary outcomes (impact of thought suppression on general psychiatric symptomatology, depression, sleep quality and perceived distress), participants were required to complete the following self-report questionnaires prior to and after the trial. General psychiatric symptomatology was measured by the Symptom Checklist-90 Revised (SCL-90-R; Derogatis, 1986). Depression was assessed using the German version of the Beck Depression Inventory, 2nd edition (BDI–II; Beck et al., 1996). Sleep quality was measured by the Landecker Inventar für Schlafstörungen (LISST; Weeß et al., 2002), which assesses sleep disorders. The Perceived Stress Questionnaire (PSQ; German version: Fliege et al., 2001) was used to assess subjectively experienced distress independent of a specific and objective occasion. All questionnaires fulfil classical test theory values and show appropriate values regarding quality criteria.
To what extent a participant will or will not follow the instruction to suppress a target thought might be influenced by his general individual tendency to suppress unwanted thoughts. To assess this tendency and control for its effect in the trial, we measured the individual tendency of thought suppression using the White Bear Suppression Inventory (WBSI; Wegner and Zanakos, 1994).
After the intake assessments, participants were asked to identify an idiosyncratic negative and distressing intrusive thought. According to Taylor and Bryant (2007), an intrusive thought was defined as: ‘…thought or image that you do not intend to think about, but pops into your head sometimes without you wanting it. It may be a memory, thought or image about a particular person, object, place, past event, imagined future event or even about yourself. It must be a thought or image that you have had before on more than one occasion. For the purpose of this study, it should be a thought or image that you do not like, and one that you do not enjoy having intrude into your mind'. Participants were then asked to rate the level of distress caused by this thought on a 10-point Likert scale (0 = ‘no distress’, 10 = ‘extreme distress’). The selected thought had to range between 8 and 10 on this scale. Subjects were asked to write down that thought. With regard to content, participants chose topics such as insufficient job performance (‘boss is criticizing me’); loss of job; end of relationship (‘my husband betrays and breaks up with me’); illness or loss of beloved person (‘cancer/death of mother’); argument with partner; indebtedness/financial ruin (‘can't pay my house off anymore’); or altercation with parents (‘my parents will be furious as soon as they find out about me dropping out of my law studies’). Participants received the study materials (instructions, dream diary) in sealed envelopes and were requested to open them directly before going to bed.
The suppression group was instructed as follows: ‘Focus your mind deliberately on the intrusive thought you selected before and concentrate on the negative feelings it triggers. Then, for the next 5 min, think about anything but the intrusive thought. Do not think about it, even for a fleeting moment, not even for a second, and do whatever it takes to keep that thought out of your mind. Then, go to bed’.
The control group was instructed to think about whatever they wanted for approximately 5 min, after focusing their mind on the intrusive thought.
Both groups were required to perform their task every evening and to complete the dream diary every morning for the following week.
PASW version 18 (SPSS, Hong Kong, China) was used for all statistical analysis. A significance level of P <0.05 was used throughout the study. High inter-rater reliability of dream content ratings was achieved (intraclass coefficient: 0.96). A power analysis showed that with an alpha level of 0.05, the study has a power of 90% to detect small to medium effects (0.44).
In order to control for the effect of the individual tendency towards thought suppression, we conducted a multivariate analysis of covariance (mancova) with experimental group as fixed factor and WBSI variable as covariate. As dependent variables we used the difference in questionnaire scores between T1 and T2.
A control of the experimental variation showed that the mean score for task performance for both groups was 2.65 (equating to ‘good’) and did not differ significantly between the groups. The mancova revealed a significant overall multivariate effect of the experimental condition (F6,22 = 2.44; P =0.044) and the covariate WBSI (F6,22 = 4.03; P =0.007).
Univariate analysis demonstrated that participants in the suppression condition reported significantly more target-related dreams than those in the control condition, with large effect sizes for both raters (R1: Cohen's d =1.41; R2: Cohen's d =1.41). Moreover, they displayed a tendency to score higher in dream distress.
Results, adjusted means, standard deviations and test statistics are displayed in Table 1. Fig. 1 shows the mean extent of target-related dreams per night for each condition.
Table 1. Results of the multivariate analysis of variance (mancova)
Dream diary outcomes
Mean number of target related dreams per week
Problems falling asleep
SD, standard deviation; d1–d7, assessments on days 1–7; T1, assessment 1; T2, assessment 2; BDI–II, Beck Depression Inventory; SCL-90-R GSI, Symptom Checklist-90 Revised global score; LISST, Landecker Inventar für Schlafstörungen; PSQ, Perceived Stress Inventory.
Sleep quality (LISST)
Univariate analysis indicated a significant main effect for the experimental condition on general psychiatric symptomatology (Cohen's d =0.89), with the participants in the suppression group scoring higher at T2. There were no significant main effects, and accordingly only small or no effect sizes for ratings of depression (d =0.36), sleep quality (d =0) or perceived distress (d =0.36).
The results yielded no significant group differences for task performance, night-time awakenings, problems falling asleep and physical reactions.
The aim of the current study was to replicate and extend the findings of previous studies demonstrating that suppression of target thoughts prior to sleep leads to increased dreaming about that content (Schmidt and Gendolla, 2008; Taylor and Bryant, 2007; Wegner et al., 2004). In accordance with research on thought suppression in waking states, which explains the effect by processes that increase the target's accessibility (for a meta-analysis see Abramowitz et al., 2001), our results show increased dreaming of target thoughts, increased dream distress and increased perceived stress in the suppression condition. In extension of previous research we showed that the effect (Wegner et al., 2004) can be found over a longer period of time (1 week), and in participants without any psychological background.
Moreover, the data show that thought suppression may lead to significantly increased psychiatric symptomatology, as measured with the SCL-90-R. Further research on this is needed. No significant effects were found for sleep quality or depression ratings, night-time awakenings, physical reactions or problems falling asleep.
There are limitations to our study. First, the extent of dream relatedness was estimated by two raters which might have been biased. However, the raters were blind to experimental conditions and showed high inter-rater reliability. The fact that we found no significant effects on many of the secondary outcomes might be due to insufficient power because of the small sample size. In future studies it would be important to conduct a dream baseline prior to the thought suppression manipulation in order to detect the extent of spontaneously occurring disturbing dream contents.
The results may contribute to our understanding of the mechanisms underlying thought suppression, distressing dreams and nightmares. Bad dreams or nightmares may be a result of thought suppression. This hypothesis is in line with research regarding psychophysiological insomnia. Riemann et al. (2012) report that patients with insomnia are frequently experiencing presleep concerns, i.e. worries about poor sleep and its consequences, which they try to suppress mainly unsuccessfully and which dominate their dream content. Specifically, it is supposed to lead to an increased occurrence of insomnia-related topics in the dreams. Consequently, giving up thought suppression might help to reduce bad dreams and nightmares. This would conform to the results of randomized controlled trials on the treatment of nightmares, which demonstrate that exposure to (e.g. Burgess et al., 1998) and imaginary modification of nightmare contents (e.g. Krakow and Zadra, 2006) both help considerably in reducing nightmares (e.g. Hansen et al., 2013). Our results are also in line with existing research concerning the psychopathology of nightmares. Levin and Nielsen (2007) developed a neurocognitive model that integrates former theories regarding nightmare development. They propose that different types of dysphoric dreaming (from bad dreams to nightmares) result from dysfunction in a network of affective processes triggered by stressful and emotionally negative events (e.g. loss of job). As concentrating upon and then avoiding negative thoughts and memories causes distress, it is assumed that these processes may provoke nightmares. Additionally, Spoormaker (2008) developed a cognitive model that explains recurrent nightmares by activation of a script (memory of nightmare storyline) that is thought to become activated when elements in a neutral dream resemble elements of the script. It is possible that recurrent unwanted thoughts are apt to activate a nightmare script, provided that they are associated with the content.
None of the authors received financial or personal support or is involved with organization(s) which have a financial interest in the subject matter of the paper.