Investigating hallucination-proneness, dissociative experiences and trauma in the general population

ABSTRACT

Background

There is evidence to suggest trauma significantly predicts risk for psychosis. Dissociation might be a key causal variable, mediating the relationship between trauma and hallucination-proneness in clinical samples.

Objective

To investigate the associations between dissociation, trauma, and hallucination-proneness in the general population sample.

Methods

The study design was correlational and cross-sectional, using a convenience sample (N = 227). Pearson’s correlation coefficients were conducted to investigate the relationship between measures of hallucination-proneness and dissociative experiences, and subjective trauma and interpretations of voices. A mediation analysis was conducted to investigate whether dissociative experiences mediate the relationship between trauma and hallucination proneness.

Results

There was a significant association between trauma and hallucinations, the severity of subjective trauma exposure was associated with increased hallucination-proneness. There was a significant correlation between hallucination-proneness and dissociative experiences. Dissociative experiences significantly mediated the relationship between subjective trauma and hallucination-proneness. Subjective trauma positively correlated with all measures of the interpretations of voices inventory.

Discussion

Previous findings from clinical samples were replicated in this general population sample, providing support to traumagenic and continuum models of psychosis, which may have implications for clinical practice.

KEYWORDS:

• Hallucinations
• dissociation
• trauma

Introduction

The association between hallucinatory and dissociative experiences were explored in early psychiatric literature but later ignored until relatively recently due to the prominence of biogenetic paradigms in psychiatry (Longden et al., 2012). However, these associations were explored in the context of trauma and abuse in clinical and non-clinical samples (Longden et al., 2012), with dissociative experiences potentially mediating the relationship between trauma and psychosis (Moskowitz & Corstens, 2007). Additionally, hallucinations, like hearing voices, can be understood as being on a continuum within the general population and not indicative of psychopathology (Baumeister et al., 2017). As such, hallucinatory and dissociative experiences can be understood as emerging along a continuum within the general population (Baumeister et al., 2017; Guloksuz & Van Os, 2018). The present paper builds upon this past research by investigating the relationship between trauma, dissociation, and hallucinatory experiences within a general population sample.

Psychosis or psychotic-like experiences such as hallucinations have been conceptualised as lying on a continuum within and between the general and clinical population (Guloksuz & Van Os, 2018). Hallucinations can be defined as perceptual experiences across sensory modalities, despite the absence of objective input to the relevant sense organ (Waters et al., 2012). Although hallucinations are often regarded as signs of psychopathology, hallucinatory experiences are relatively common in the general population. For example, McGrath et al. (2015) found a 5.2% lifetime prevalence of hallucinatory experiences in a large cross-national sample. The phenomenology of hallucinations is incredibly diverse, often studied through their form and content. Form can be defined as the descriptive perceptual characteristics of hallucinations, including location, loudness, number of voices in the case of auditory verbal hallucinations (AVH) and clarity of the percept-like experiences (Johns et al., 2014). Content is what is seen or heard in the case of visual and auditory hallucinations, respectively, and often includes the emotional valence of the experience (Baumeister et al., 2017). An example of a commonly experienced AVH might be hearing a loved one after their death or hearing your name called when no one is present (Longden et al., 2012).

There are several key distinctions between healthy voice-hearing and clinically distressing voice-hearing (Baumeister et al., 2017; Johns et al., 2014). For example, compared to healthy voice-hearers, distressed voice hearers are more likely to experience negative voice content (Baumeister et al., 2017; Johns et al., 2014). However, there is evidence that the cognitive processes underlying healthy and distressing voice-hearing are largely shared (Johns et al., 2014). The exceptions include temporary cognitive disruptions to attentional and memory processes in healthy voice-hearing, contrasted with more chronic disruptions in distressing voice-hearing (Johns et al., 2014). Morrison et al. (2002) found that meta-physical beliefs about voices significantly predicted troublesome voices and increased propensity to experience visual hallucinations and positive beliefs about voices significantly predicted increased hallucination-proneness (Morrison et al., 2002). Overall, this suggests that the distressing hallucinatory experiences are mediated by negative metacognitive beliefs that share processes with anxiety disorders (Morrison et al., 2002).

One proposed mechanism by which voice-hearing experiences are thought to occur is dissociation. Dissociation can be defined as a “disruption of and/or discontinuity in the normal integration of consciousness, memory, identity, emotion, perception, body representation, motor control, and behaviour” (American Psychiatric Association, 2013, p. 291). Dissociation can also be understood as an evolved mechanism for coping with overwhelming stress (Schalinski et al., 2015). For example, the defence cascade model (Schauer & Elbert, 2010) suggests that dissociation occurs when fight or flight responses are not sufficient to counter danger and overwhelming stress (Schalinski et al., 2015). Although often a response to severe trauma, dissociation lays on a continuum where moderate dissociative experiences occur in the general population to cope with moderate stress (Michal et al., 2009). An example might be when confronted with job loss an individual may feel out of place in a familiar office, finding their work colleagues unfamiliar. An extreme example can include feeling outside one’s body or watching oneself from afar, this kind of detachment is frequently reported by sexual assault survivors (Schalinski et al., 2015).

Dissociation has been theorised to play an important role in hallucinatory experiences (Longden et al., 2012), potentially due to the neurocognitive disruptions dissociation produces (Schalinski et al., 2015). For example, hallucinating and remitted hallucinating service-users experienced more dissociation compared to non-hallucinators and controls, with hallucinating service-users experiencing significantly more trauma compared to all other groups (Varese et al., 2012). Furthermore, the hallucinating groups experienced significantly greater response bias, detecting voices when none were present in background noise on an signal detection task (Varese et al., 2012). This suggests that highly dissociated service users are more likely to make source monitoring errors and misattribute internal events to external events (Varese et al., 2012). This could be explained by alterations in attention and self-recognition involved in hallucination-proneness (Waters et al., 2012), which the neurocognitive disruptions of dissociation seem to produce (Schalinski et al., 2015). Overall, this suggests that dissociation mediates the relationship between trauma and hallucination-proneness because traumatised service users experience more dissociation that facilitates the cognitive disruptions associated with increased hallucination-proneness.

The relationship between trauma, dissociation and hallucination-proneness is not limited to clinical populations. For example, voice-hearers without psychiatric history report similar childhood adversity to distressed voice-hearers compared to controls (Baumeister et al., 2017). Dissociation also predicts both increased delusional ideation and anomalous perceptual experiences in non-clinical samples (Humpston et al., 2016). Overall, this suggests that trauma and dissociation increase hallucination-proneness and interact through trauma, producing increased dissociation which mediates the relationship between trauma and hallucination-proneness by further facilitating the cognitive disruptions found in both clinical and non-clinical populations. This could suggest that early life trauma provides some risk towards experiencing more severe psychosis, with trauma increasing moderate and severe dissociative experiences that facilitate hallucination-proneness.

The first hypothesis is that increased dissociative experiences will be associated with increased hallucination-proneness. The second hypothesis is the associations between subjective experiences of early life trauma (STR) and hallucination proneness will be mediated by dissociative experiences, as previously demonstrated in clinical samples (Varese et al., 2012). Lastly, an exploratory hypothesis was formulated to test whether STR was associated with the formation of negative metacognitive beliefs about voices. It was predicted that increased STR will be associated with increased metaphysical beliefs, interpretations of loss of control but not positive beliefs about voices.

Method

Participants

Participants included volunteers from the general population (N = 227) who responded to online social media adverts or first-year psychology undergraduates using the University of Liverpool’s Experiment Participation Requirement (EPR) scheme. Participant had a median age of 23 (range = 18–64 years). Participant gender was 38.5% male, 58.8% female and 2.7% non-binary. Information on race or ethnicity was not collected. The inclusion criteria included being over 18 years old and being able to speak fluent English. The exclusion criteria included having no history of a psychosis diagnosis.

Study design

The study was of a cross-sectional design, correlations were performed to assess the relationship between hallucination-proneness, dissociative experiences, subjective ratings of trauma, and interpretations of voices. Data was collected through a survey which included four separate measures and an ad-hoc question. The ad-hoc question about the subjective impact of trauma was used to test the second and third hypotheses regarding the associations with trauma, dissociation, hallucination-proneness and beliefs about voices.

Measures

Revised dissociative experiences scale (DES, Carlson & Putnam, 1993)

This measure consists of 28-items measuring increasingly pronounced dissociative experiences. Endorsement of these items was measured using an 11-point Likert scale from 0% to 100% of the time (0–10 score). The internal consistency of the DES was excellent (α = .94),

Revised hallucination scale (RHS, Morrison et al., 2002)

This scale measured proneness to both visual and auditory hallucinations, consisting of 20-items. Endorsement of items was measured by a 4-point Likert scale (0 = never, 1 = sometimes, 2 = often, 3 = almost always). Two example questions include “I daydream about being someone else” and “I hear a voice speaking my thoughts aloud” prototypical of the two extremes of hallucination-proneness, respectively. The internal consistency was excellent for the RHS (α = .92).

Shutdown dissociation scale (shut-D, Schalinski et al., 2015)

This scale consisted of 13-items measuring Depersonalisation and derealisation experiences of varying intensity based on the defence cascade model of shutdown dissociation associated with various psychological, somatic and perceptual alterations (Schalinski et al., 2015). Endorsement of items was measured on a 5-point Likert scale (0 = not at all, 1 = a little bit, 2 = moderately, 3 = strongly, 4 = very strongly). An example question includes “Have you felt as though you couldn’t move for a while, as though you were paralyzed?” describing the more somatic aspect of dissociation compared to the DES. The internal consistency was good for the Shut-D (α = .89).

The interpretations of voices inventory (IVI, Morrison et al., 2002)

This scale measured beliefs about voices, consisting of 22-items measured on a 4-point Likert scale (0 = not at all, 1 = somewhat, 2 = moderately so, 3 = very much). This questionnaire was worded hypothetically (“if I were to hear sounds or voices that other people could not hear I would probably think that … ”) and split into three subscales for based on the subscales in Morrison et al. (2002). These included 14-items assessing metaphysical beliefs about voices, 8-items assessing positive beliefs about voices and 5-items assessing interpretations of loss of control. The internal consistency was excellent for the IVI (α = .90) and for the IVI subscales it was good for meta-physical beliefs (α = .85), positive beliefs about voices subscale (α = .84) and Interpretations of loss of control (α = .86).

Lastly, an ad-hoc question was developed to test the subjective rating of early adversity or trauma participants may have experienced. The question was worded to capture subjective experiences opposed to objective events (“to what extent do you feel you have experienced early life trauma or adversity (before the age of 16”)?) and was measured using a visual analogue scale (0–100) with the percentage obscured so participants placed their answer based on subjective feeling. This scale was abbreviated to Subjective Trauma Rating (STR).

Procedure

Participants were invited to take part in the survey through online adverts with an anonymous link to a Qualtrics platform hosting the survey. Once accessed, participants were given an electronic copy of the participant information sheet (PIS), with the contact details of the principal investigator (PI) included. Participants were asked to confirm they read the PIS before proceeding to the next webpage to provide informed consent. Participants were then asked to give details about age and gender before proceeding to fill out the survey in the order of the DES, RHS, Shut-D, IVI and ad-hoc trauma question.

Participants were then presented with an electronic copy of the debrief form that included contact information of the PI, relevant resources regarding voice-hearing, advice, guidance and wellbeing services if distress was caused by the study.

Data analysis

All variables were not normally distributed, and SPSS 24 and later 27 were used to compute the correlational analyses. Due to the large sample size, parametric statistics were used to test all the hypotheses, including Pearson’s correlation coefficient to investigate the first and the last exploratory hypothesis.

The second hypothesis was tested using a mediation analysis computed in R (version 4.3.0) using PROCESS version 4.3 (Hayes, 2022). The independent variable was STR, the dependent variable was hallucination proneness with combined dissociation score (CDS) as the mediator variable. The Shut-D and DES were correlated with each other (r (185) = .72, p < .001) and were combined to create the CDS to avoid collinearity. 95% confidence intervals (CI) were estimated for the indirect effect from 5,000 bootstrapped samples. A significant indirect effect was considered if the 95% CI does not contain a zero (Preacher & Hayes, 2008).

Results

Descriptive statistics

Descriptive statistics for all variables utilising data from the total sample are provided in Table 1.

Table 1. Descriptive statistics of all variables included in the analyses from the total sample.

Measure	Mean (± SD)
STR	42.08 (±34.70)
RHS	12.90 (±9.90)
DES	49.74 (±38.95)
SHUT-D	9.86 (±8.46)
IVI Subscales:
Meta-physical	2.14 (±3.68)
Positive beliefs	2.26 (±3.33)
Loss of control	6.61 (±6.82)

There was a significant positive correlation between the DES total score and RHS total score, r (186) = .81, p < .001, and Shut-D total score and RHS total score, r (218) = .71, p < .001.

There was a significant positive correlation between STR and hallucination-proneness, r (177) = .33, p < .001, and STR and the dissociation scales, DES, r (152) = .35, p < .001, and Shut-D, r (177) = .38, p < .001.This suggests a mediation analysis would be appropriate.

In the mediation model, there was a significant association between STR and Combined Dissociation Score (CDS) (B = .24, SE = .10, 95%CI .04 to .44, p = 0.02) with greater STR associated with increased CDS. The associations between CDS and hallucination proneness (B = 3.76, SE = .21, 95%CI 3.35 to 4.16, P < .001) was also significant with higher CDS associated with increased hallucination proneness. The total effect was significant (B = 1.06, SE = .26, 95%CI .55 to 1.58, p < .001), with a significant indirect effect between STR and hallucination-proneness via CDS (B = .90, SE = .42, 95%CI .15 to 1.77). The direct effect between STR and hallucination-proneness was not significant (B = .16, SE = .46, 95%CI −.75 to 1.06, p = .737). The mediation model is shown in Figure 1.

Figure 1. A mediation model showing the indirect effect between subjective trauma rating and hallucination-proneness via combined dissociation score.

Subjective Trauma rating (STR), Combined Dissociation Score (CDS) (sum of Dissociative Experiences Scale (DES) and Shutdown Dissociation Scale (Shut-D)), Hallucination-proneness (RevisedHallucination Scale (RHS). p <.05*,p <.01**, p<.001***.

Lastly, there was a significant positive correlation between subjective trauma score and total metaphysical beliefs score, r(177) = .33, p < .001, total positive beliefs about voices score, r (177) = .24, p = .001 and interpretations of loss of control, r (177) = .36, p < .001.

Discussion

Overall, the first hypothesis was supported with increased dissociative experiences highly correlating with increased hallucination-proneness. The second hypothesis was supported because the association between subjective trauma ratings and hallucination-proneness was mediated by CDS. Lastly, the third hypothesis was not supported because high subjective trauma significantly positively correlated with all subscales of the IVI.

Theory implications

This study builds upon previous research that explores the relationship between trauma, dissociation, and psychotic experiences in clinical samples (Varese et al., 2012) and dissociation and hallucination-proneness in non-clinical samples (Humpston et al., 2016). As such, this study replicated that dissociation likely play a significant role in hallucination-proneness in the general population, and extended the role of trauma from clinical samples to a general population sample. This may have important implications for how psychosis is conceptualised. For example, this suggests that psychotic experiences likely lie on a continuum in the general population (Guloksuz & Van Os, 2018). This was supported in this general population sample by the relatively high levels of reported hallucinatory experiences which suggest these experiences are on a continuum of frequency and intensity that extends beyond clinical samples.

Dissociation strongly mediated the association between STR and hallucination-proneness, with the direct effect between STR and hallucination proneness not reaching significance. Varese et al. (2012) found dissociation mediated the association between sexual abuse and hallucination proneness in a patient sample but not in the aggregated sample. Instead, dissociation only mediated the association between negative home life and neglect. This could suggest distinct clinical and non-clinical mechanisms, but together with the results of this study, it might also suggest a shared continuum. A liability-threshold model for psychotic experiences (Guloksuz & Van Os, 2018) could elucidate the findings of these mediation analyses because of the increased incidence and severity of abuse reported in psychosis samples (Varese et al., 2012).

From this perspective, the increased severity and types of adversity (for example, sexual abuse or neglect) produce greater dissociative symptoms that facilitate the aforementioned neurocognitive processes (for example, poor source monitoring, self-recognition and attention) of hallucinatory experiences. These experiences may emerge as distressing symptoms in varied phenotypes, such as schizophrenia or dissociative identity disorder (Moskowitz & Corstens, 2007) but share an underlying mechanism. Future research should use a serial mediation analysis to investigate how specific neurocognitive processes that may emerge from dissociation could facilitate hallucination proneness. This research could then be related to specific hallucinatory experiences or clinical outcomes.

Clinical implications

These results highlight the clinical significance of the associations between trauma, dissociation and hallucination-proneness in a general population sample. This suggests a continuum of shared causal traumagenic processes from the general to the clinical population highlighting the need for more trauma-informed care and services. For example, STR scores were relatively high in this sample. This highlights the widespread nature of distressing early life experiences that affect individuals across the lifetime. This is supported by previous research which has found that adverse childhood experiences (ACE) including forms of abuse and familial dysfunction are reported at least once by 50% and twice by 25% of the sample (Felitti et al., 1998). These adverse experiences were found to have a dose-response relationship with increased physical and mental health complaints and represent a leading public health crisis (Felitti et al., 1998).

The relationship between adverse childhood experiences and negative mental health outcomes also extends to psychosis as previously discussed, this association appears to be particularly strong in individuals who experience sexual abuse (Varese et al., 2012). However, this study focused on the subjective sense of experiencing adversity and not specific adverse experiences or victimisation. This had the benefit of capturing a far broader range of adverse experiences not usually included in studies of childhood adversity. There is evidence to suggest that the appraisal of adverse events is what drives post-traumatic stress, including the perceived threat and perceptions of familial support and not the actual severity of the event alone (Trickey et al., 2012). As such, clinical practice should recognise the broad range of adversity that may be causal of psychosis and design interventions and clinical research following a continuum of reported experiences of subjective adversity and not categorical self-reports of assumed traumatic events that may miss sources of adversity.

There was a significant positive correlation between STR and all subscales of the IVI (Morrison et al., 2002). This suggests that as STR increased so did the negative metaphysical beliefs, interpretations of loss of control and positive beliefs about voices. There is evidence to suggest these negative metacognitive beliefs are involved in the maintenance of psychosis (Morrison et al., 2002). This suggests that dysfunctional attitudes and meta-cognitive beliefs are informed by traumatic childhood experiences. STR was also positively correlated with positive beliefs about voices these could represent an adaptive coping strategy. Overall, this suggests that psychotherapies, such as cognitive behavioural therapy (CBT) for psychosis should consider traumatic life experiences when attempting to treat the cognitive appraisals of psychotic experiences that produce distress. There is evidence that trauma-informed CBT for psychosis improved traumatic symptoms caused by experiencing a psychotic episode (Gianfrancesco et al., 2019), it may also be effective for addressing the causal influence of trauma on psychosis.

Psychotherapies that target dissociative experiences may be efficacious for treating distressing psychotic experiences. This has been used in the “Accepting and Working With Voices” Maastricht Approach for treating distressing voices (Corstens et al., 2018). The Maastricht Approach is well-regarded by service user organisations and experts-by-experience as useful for personal recovery (Corstens et al., 2018).

Limitations

Although using a subjective scale of trauma has its benefits as previously discussed. For example, providing a more ecologically valid measure of appraisal of trauma. Future research would benefit from comparing this measure to existing validated scales of trauma in order to test the subjective scales concurrent validity and whether trauma findings can be generalised to other research. Although race or ethnicity were not included, it is possible that individuals impacted by social adversity, including discrimination might have included this in their own STR. It would be impossible to tell if this occurred without validation and future research could focus on the interrelationships between adversity, ethnicity and hallucinatory experiences.

The cross-sectional design of this study meant that it was limited in being able to predict hallucination proneness from the traumagenic or dissociative independent variables, it would have been particularly useful to investigate these risk factors over a period of time. For example, whether dissociative experiences predict or emerge from hallucinatory experiences. The lack of measures of general wellbeing, including anxiety might confound some of the findings in this study (Morrison et al., 2002). For example, findings from this study and Morrison et al. (2002) suggest investigating if anxiety mediates the relationship between trauma and beliefs about voices as anxiety is associated with hallucination-proneness (Waters et al., 2012). The lack of anxiety measures meant it could not be controlled when investigating beliefs about voices and the minority of highly hallucination-prone individuals may represent a distressed sub-group analogous to clinical groups which is particularly relevant to the dimensional models proposed by Baumeister et al. (2017).

Although the role of dissociation is highlighted, providing support for dissociation mediating the relationship between trauma and psychosis (Varese et al., 2012). The role of trauma on producing hallucinations is likely more complex. For example, trauma may alter brain development, with glucocorticoid toxicity impairing the integrity of brain areas such as the hippocampus or prefrontal cortex (Read et al., 2014), possibly associated with hallucinatory experience through the impairment of source memory or executive function. This traumagenic pathway is more direct than what this study can capture or control for. This limitation could also relate to the overlap of the constructs of dissociation and psychosis, particularly with regards to voice hearing (Moskowitz & Corstens, 2007). This is a limitation of this study and research in the interface between dissociation and hallucinations in general as they may be the varied expression of the same underlying traumagenic phenomenon.

The strikingly high correlations between hallucination proneness and dissociation could be related to Moskowitz and Corstens (2007) position regarding voice hearing and dissociation to be one and the same experience. Potentially, these correlations display collinearity due to the overlap in the constructs measured in the scales and not because they are independently correlated. This could be interpreted to mean that dissociation is such a nebulous term it cannot be readily differentiated from hallucinatory experience or that they are shared processes. Typically, they are considered independent constructs, and these correlations still provide evidence towards overlapping experiences, even if they are not necessarily independent.

Conclusions

Overall, this study supports previous findings of the associations between trauma, dissociation, and psychosis (Varese et al., 2012, 2012) but extends it to the general population. This has several important theoretical and clinical implications. For example, it suggests the environmental determinants of psychotic experiences occur in non-clinical populations, supporting a dimensional continuum model of psychosis. These findings do suggest a need for widespread trauma and dissociation-informed services, due to the prevalence of trauma in the general population and poor health consequences (Felitti et al., 1998). There is evidence that such measures would prevent a third of psychosis cases (Varese et al., 2012). This study highlights the need for trauma-focused CBT for psychosis as suggested by Gianfrancesco et al. (2019). Overall, this study provides further support for the mounting evidence that necessitates trauma and dissociation-informed services for psychosis.

Ethical Committee Statement

Study design

The study was of a cross-sectional design, correlations were performed to assess the relationship between hallucination-proneness, dissociative experiences, subjective ratings of trauma, and interpretations of voices. Data was collected through a survey which included four separate measures and an ad-hoc question. The ad-hoc question about the subjective impact of trauma was used to test the second and third hypotheses regarding the associations with trauma, dissociation, hallucination-proneness and beliefs about voices. The study was approved by the University of Liverpool Health and Life Sciences Research Ethics Committee (Approval Reference: 5569).

Disclosure statement

The authors have no conflicts of interest to declare in relation to the subject of this study.

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.

Additional information

Funding

None to declare.