Introduction
Cognitive models of auditory verbal hallucinations assert that the beliefs that people hold about the voices they hear mediate associated levels of distress and disability. Two main types of belief have been proposed as important. The first, and more widely studied, set of beliefs was described by Chadwick and Birchwood (Reference Chadwick and Birchwood1994), who highlighted the importance of explanatory beliefs associated with seeing voices as sentient others interacting with the person. These include appraisals of whether the entities responsible for their voices have malevolent or benevolent intent, and of their power and status relative to the person hearing them. In particular, appraisals of voices as malevolent and omnipotent were proposed to mediate distress. Accordingly, measures of these beliefs have been associated with ratings of voice-related distress (e.g. Birchwood and Chadwick, Reference Birchwood and Chadwick1997; Peters, Williams, Cooke and Kuipers, Reference Peters, Williams, Cooke and Kuipers2012) and measures of depression and anxiety (e.g. Van der Gaag, Hageman and Birchwood, Reference Birchwood2003). They have also been shown to predict behavioural responses to hearing voices (e.g. Chadwick and Birchwood, Reference Chadwick and Birchwood1995; Hayward, Reference Hayward2003; Hayward, Denney, Vaughn and Fowler, Reference Hayward, Denney, Vaughan and Fowler2008; Thomas, McLeod and Brewin, Reference Thomas, McLeod and Brewin2009), including compliance with command hallucinations (Barrowcliff and Haddock, Reference Barrowcliff and Haddock2010; Beck-Sander, Birchwood and Chadwick, Reference Birchwood and Chadwick1997; Fox, Gray and Lewis, Reference Fox, Gray and Lewis2004; Shawyer et al., Reference Shawyer, Mackinnon, Farhall, Sims, Blaney and Yardley2008; Trower et al., Reference Trower, Birchwood, Meaden, Byrne, Nelson and Ross2004).
The second set of beliefs about voice experience was described by Morrison (Reference Morrison1998; Morrison, Haddock and Tarrier, Reference Morrison, Haddock and Tarrier1995; Morrison, Wells and Northard, Reference Morrison, Wells and Nothard2002). Morrison highlighted beliefs about the self-related implications of experiencing voices as a mental phenomenon, in a similar way to metacognitive beliefs about experiences such as worry (e.g. Wells, Reference Wells1995). Specifically, Morrison suggested that positive beliefs held about the value of hallucinatory experience may be associated with efforts to engage with and maintain hallucinatory experience (Morrison et al., Reference Morrison, Wells and Nothard2002), reinforcing this phenomenon. Meanwhile, the person may hold parallel negative appraisals that hearing voices is a source of threat, which mediate distress (Morrison, Reference Morrison1998). Developing a measure of these “metacognitive” beliefs, the Interpretation of Voices Inventory (IVI), Morrison et al. (Reference Morrison, Wells and Nothard2002) derived three factors: positive beliefs (e.g. “they make me special”; “I would not cope without them”), plus two types of negative belief, namely metaphysical beliefs (e.g. “they mean I have done something bad”, “they mean I am possessed”) and beliefs about loss of control (e.g. “they will take over my mind”, “they will make me go crazy”). These beliefs differ from those described by Chadwick and Birchwood (Reference Chadwick and Birchwood1994) in focusing on the consequences of hearing voices as an experience of the hearer, rather than on the perceived attributes of voices as social agents. Positive beliefs about voices have been found to be associated with hallucination-proneness (Morrison et al., Reference Morrison, Wells and Nothard2002), whilst negative beliefs predict whether hallucinatory experiences are regarded as problematic and distressing (Morrison et al., Reference Morrison, Wells and Nothard2002; Morrison, Northard, Bowe and Wells, Reference Morrison, Northard, Bowe and Wells2004).
It has been observed that these beliefs about voices involve the person making inferences beyond what is manifest in voice content alone (Birchwood and Chadwick, Reference Birchwood and Chadwick1997; Close and Garety, Reference Close and Garety1998), in fact sometimes being quite incongruent with content (Chadwick and Birchwood, Reference Chadwick and Birchwood1994; Shawyer et al., Reference Shawyer, Mackinnon, Farhall, Sims, Blaney and Yardley2008; Van der Gaag et al., Reference Van der Gaag, Hageman and Birchwood2003). In explaining the formation of such beliefs, a central idea within cognitive models is that appraisals of ongoing events are influenced by more generalized cognitive representations of prior experience, often referred to as schemas. Hence cognitive representations of the personal and social meaning of voices may be understood as influenced by schemas the person holds about themselves and the social world (Chadwick, Birchwood and Trower, Reference Chadwick, Birchwood and Trower1996; Morrison, Reference Morrison2001; Paulik, Reference Paulik2012). Indeed, key psychological models of psychosis have proposed that negative schemas relating to self and others are central to understanding the interplay between environment, emotion and cognition in psychosis (e.g. Birchwood et al., Reference Birchwood, Gilbert, Gilbert, Trower, Meaden and Hay2004; Garety, Kuipers, Fowler, Freeman and Bebbington, Reference Garety, Kuipers, Fowler, Freeman and Bebbington2001; Garety, Bebbington, Fowler, Freeman and Kuipers, Reference Garety, Bebbington, Fowler, Freeman and Kuipers2007; Morrison, Frame and Larkin, Reference Morrison, Frame and Larkin2003). Extreme negative evaluations of both self and others are readily endorsed in people with psychosis (Fowler et al., Reference Fowler, Freeman, Smith, Kuipers, Bebbington and Bashforth2006), and there is developing evidence for the association of these with delusional beliefs (Addington and Tran, Reference Addington and Tran2009; Freeman and Fowler, Reference Freeman and Fowler2009; Smith et al., Reference Smith, Fowler, Freeman, Bebbington, Bashforth and Garety2006). In providing a cognitive representation of early experience, schemas also provide a model for understanding the influence of social environment on the understanding the person forms of their psychotic experiences (Birchwood, Reference Birchwood2003; Garety et al., Reference Garety, Bebbington, Fowler, Freeman and Kuipers2007; Morrison et al., Reference Morrison, Frame and Larkin2003), including the role of aversive environmental factors such as abuse, trauma and marginalization, which are increasingly seen as important in relation to psychosis (e.g. Read, Van Os, Morrison and Ross, Reference Read, van Os, Morrison and Ross2005). This provides a means of conceptualizing connections observed between beliefs about voices and recollections of parental behaviour (Offen, Thomas and Waller, Reference Offen, Thomas and Waller2003), childhood sexual abuse (Offen, Waller and Thomas, Reference Offen, Waller and Thomas2003), and trauma variables (Andrew, Gray and Snowden, Reference Andrew, Gray and Snowden2008).
There is some initial evidence that schemas may be linked with beliefs about voices. Using the “downward arrow” thought-chaining technique, a method used in cognitive therapy to identify core schemas from specific appraisals, Close and Garety (Reference Close and Garety1998) found that the majority of their participants were able to draw out negative cognitions about self that related to the specific appraisals of voices that had been elicited. Other studies have found that measures of the perceived relative power and superiority of voices in relation to oneself correspond to views of one's own lack of power and inferiority in relation to others (Birchwood, Meaden, Trower, Gilbert and Plaistow, Reference Birchwood, Meaden, Trower, Gilbert and Plaistow2000; Birchwood et al., Reference Birchwood, Gilbert, Gilbert, Trower, Meaden and Hay2004). However, whilst power-related dimensions have been compared with broader beliefs using formal measures, there has yet to be a systematic study of how either appraisals of voice intent (malevolence or benevolence) or metacognitive beliefs about voices may relate to broader cognitive representations of self and others.
Representations of both self and others potentially influence appraisals of voice intent. Chadwick et al. (Reference Chadwick, Birchwood and Trower1996) argued that appraisals of voices’ malevolence may arise from negative evaluative beliefs about self. For example, malevolent intent might be appraised if the person has schemas relating to being deserving of punishment or vulnerable to victimization. In support of this, Close and Garety (Reference Close and Garety1998) were successful in eliciting semantically congruent self-related beliefs from targeted questioning. Self-representations might also influence the negative metacognitive beliefs proposed as important by Morrison et al. (Reference Morrison, Wells and Nothard2002). The highest loading items on the metaphysical beliefs scale refer to voices being interpreted as meaning the hearer has done bad things, is a bad person or is possessed, likely to be influenced by negative self schemas. Similarly, negative self schemas are likely to influence the second set of negative beliefs relating to concerns about loss of control when hearing voices.
Cognitive representations of others also seem relevant to the formation of voice-related beliefs. Appraisals of voice malevolence and benevolence involve the voice hearer imposing characteristics upon their voices as if the voice were another person in their social environment. Here, the person is viewing their voices using the same lens through which they view other people. In line with this, there is evidence that voice hearers conceptualize their experience of hearing voices in coherent interpersonal terms (Benjamin, Reference Benjamin1989; Chin, Hayward and Drinnan, Reference Chin, Hayward and Drinnan2009; Hayward, Berry and Ashton, Reference Hayward, Berry and Ashton2011; Hayward et al., Reference Hayward, Denney, Vaughan and Fowler2008; Vaughn and Fowler, Reference Vaughn and Fowler2004), that this influences their responses to voices (Thomas et al., Reference Thomas, McLeod and Brewin2009), and that their relationships with voices mirror broader patterns of social relating (Hayward, Reference Hayward2003). Consequently, schemas about others – cognitive representations of the behaviour of other people –may also be important in influencing the types of appraisals formed. Specifically, it might be expected that appraisals of malevolence would be related to schemas held about people in general as potentially hostile or untrustworthy, whilst appraisals of benevolence would be related to more positive schemas held about others. Additionally, Morrison et al. (Reference Morrison, Frame and Larkin2003) have proposed that immersion in hallucinatory experience may develop as a coping strategy for dealing with experiences and memories of trauma and abuse, for which there is accumulating evidence of an association with predisposition to hallucination (McCarthy-Jones, Reference McCarthy-Jones2011; Read et al., Reference Read, van Os, Morrison and Ross2005). Hence it is possible that the positive beliefs about the value of voice experience described on the IVI will be predicted by negative schemas about other people.
In this study we examined the relationship between beliefs about voices and cognitive representations of self and others in people who experience auditory hallucinations. The schema measure developed by Fowler et al. (Reference Fowler, Freeman, Smith, Kuipers, Bebbington and Bashforth2006) was used, as this had been designed to capture schemas about self and others that are proposed as particularly pertinent to psychosis, and has been validated within this population (Fowler et al., Reference Fowler, Freeman, Smith, Kuipers, Bebbington and Bashforth2006; Smith et al., Reference Smith, Fowler, Freeman, Bebbington, Bashforth and Garety2006). This measure assesses four schema factors: negative-self, positive-self, negative-others, and positive-others. The primary aim was to determine whether relationships between these schemas and beliefs about voices exist, including after controlling for the amount and degree of negative voice content. A second aim was to examine the extent to which beliefs about voices would be associated with schemas relating to self or schemas relating to others. In particular, we predicted relationships between each of the negative beliefs about voices (omnipotence, malevolence, metaphysical and loss of control beliefs) and negative-self schemas, and additionally between malevolence and negative-others, between benevolence and positive-others, and between positive beliefs about voices and negative-others.
Method
Participants
Following ethics committee approval, attendees at a specialist outpatient clinic providing psychological therapy for auditory hallucinations (Thomas, Rossell, Farhall, Shawer and Castle, Reference Thomas, Rossell, Farhall, Shawer and Castle2011) completed measures as part of a baseline assessment during their initial appointment. Of 38 consecutive patients seen at the clinic during the data collection period, 34 met the study inclusion criteria of: (a) a diagnosis of schizophrenia or schizoaffective disorder; (b) current auditory hallucinations in the form of voices; (c) history of hearing voices of at least one year; and (d) sufficient literacy and English to complete self-report questionnaires. Twenty-two (65%) were male, and the overall mean age was 35.4 (SD 8.52). Thirteen (38%) were in paid or voluntary employment. Participants reported hearing voices for a mean of 11.1 (SD 6.91) years. All were taking antipsychotic medication at the time of the study.
Measures
Psychotic Symptom Rating Scales Auditory Hallucinations Subscale (PSYRATS; Haddock, McCarron, Tarrier and Faragher, Reference Haddock, McCarron, Tarrier and Farragher1999) is an interview-based set of clinician rating scales, comprising 11 5-point items assessing dimensions of voice experience such as frequency, content and distress. The amount of negative content of voices (item 6) and degree of negative content (item 7) subscales were used to assess negative content. Inter-rater reliabilities of these PSYRATS items are excellent, and they show good test-retest reliability and validity (Haddock et al., Reference Haddock, McCarron, Tarrier and Farragher1999; Drake, Haddock, Tarrier, Bentall and Lewis, Reference Drake, Haddock, Tarrier, Bentall and Lewis2007).
Interpretation of Voices Inventory (IVI; Morrison et al., Reference Morrison, Wells and Nothard2002) is a 26-item questionnaire assessing three types of metacognitive belief about voices: positive beliefs, negative metaphysical beliefs, and loss of control beliefs (see introduction for examples of items). Scales show good internal consistency and test-retest reliability and there is evidence of validity from predicted correlations between scales and hallucination-proneness, hallucinations and hallucination-related distress (Morrison et al., Reference Morrison, Wells and Nothard2002, Reference Morrison, Northard, Bowe and Wells2004).
Revised Beliefs About Voices Questionnaire (BAVQ-R; Chadwick, Lees and Birchwood, Reference Chadwick, Lees and Birchwood2000). The BAVQ-R was administered, assessing beliefs about voices on three scales: malevolence (e.g. “my voice is punishing me for something I have done”), benevolence (e.g. “my voice wants to protect me”), and omnipotence (e.g. “my voice seems to know everything about me”). Each scale has six items that are rated on a 4-point scale ranging from 0 (disagree) to 3 (agree strongly), producing scales of 0 to 18 points. Scales show good internal reliability and validity (Chadwick et al., Reference Chadwick, Lees and Birchwood2000; Mawson, Cohen and Berry, Reference Mawson, Cohen and Berry2010).
Brief Core Schema Scales (BCSS; Fowler et al., Reference Fowler, Freeman, Smith, Kuipers, Bebbington and Bashforth2006). The BCSS is a 24-item questionnaire assessing both negative and positive schemas about self (e.g. “I am vulnerable”, “I am successful”) and others (e.g. “Other people are devious”, “Other people are supportive”), forming four subscale scores, each with a possible range of 0 to 24 points. The measure shows good internal consistency, test-retest reliability and validity within a psychotic population (Fowler et al., Reference Fowler, Freeman, Smith, Kuipers, Bebbington and Bashforth2006; Smith et al., Reference Smith, Fowler, Freeman, Bebbington, Bashforth and Garety2006).
The Positive and Negative Syndrome Scales (PANSS; Kay, Opler and Lindemayer, Reference Kay, Opler and Lindenmayer1987) and Schedule for the Assessment of Insight (David, Buchanan, Reed and Almeida, Reference David, Buchanan, Reed and Almeida1992) were also administered as part of the initial assessment, and scores on these measures were used to examine positive symptoms, negative symptoms and insight as potential confounding variables.
Data analysis
Skewing on the negative metaphysical beliefs and benevolence scales was corrected with square root transformations, after which all variables conformed to a normal distribution, allowing parametric analyses to be used. Initially, correlations between beliefs about voices and voice content, schemas and other clinical and demographic variables were calculated. Studies of relationships between appraisals of voice power and schemas (Birchwood et al., Reference Birchwood, Meaden, Trower, Gilbert and Plaistow2000, Reference Birchwood, Gilbert, Gilbert, Trower, Meaden and Hay2004) have observed large effect sizes equivalent to r ≥ .5, which the sample of N = 34 had 87% power (α = .05) to detect in this analysis. A series of hierarchical linear regression analyses was then conducted for each of the six beliefs about voices. In each analysis, the amount of negative content and degree of negative voice content scales of the PSYRATS were entered as predictors in a first step, and then any of the four schema scales identified as holding a bivariate correlation with that belief at p < .10 (two-tailed) were entered as predictors in a second step. The sample size provided 80% power to detect incremental changes in R 2 of 0.198 with one additional predictor and 0.239 with two additional predictors. Regressions were then rerun to examine whether inclusion of potential confounding variables changed results. For each regression analysis, data were screened for multicollinearity (no predictor variables were intercorrelated r > .7) and residuals plots checked for normality, linearity and homoscedacity. Two-tailed significances are reported throughout.
Results
Mean scores on the main measures are given in Table 1. High rates of endorsement of negative-self and negative-other schemas were observed, similar to those reported by Fowler et al. (Reference Fowler, Freeman, Smith, Kuipers, Bebbington and Bashforth2006) for their psychosis sample.
Table 1. Mean (SD) of scores on main measures
Notes: N = 34. PSYRATS, Psychotic Symptom Rating Scales; BCSS, Brief Core Schema Scales; BAVQ-R, Revised Beliefs About Voices Questionnaire; IVI, Interpretation of Voices Inventory
Beliefs about voices were unrelated to age, gender, overall severity of either positive or negative symptoms, or to insight. Malevolence was correlated with voice loudness, r = .36, p = .035, but no other relationships were observed between beliefs about voices and voice frequency, duration, loudness or location. However, several of the beliefs about voices were correlated with the number of years the person had heard voices for: omnipotence, r = .34, p = .050; malevolence, r = .35, p = .042; negative metaphysical beliefs, r = .44, p = .009; and positive beliefs, r = .41, p = .017.
Negative schema scales were relatively independent from corresponding positive schema scales for self, r = –.25, p = .16, and others, r = –.12, p = .49, supporting their separate examination. On the other hand, negative-self and negative-others were moderately correlated, r = .45, p = .008, as were positive-self and positive-others, r = .70, p < .001. None of the schema scales were correlated with amount or degree of negative voice content.
Correlations between the beliefs about voices and the predictor variables are presented in Table 2. This shows that five of the six beliefs about voices were correlated with either the amount, or the degree, of negative content of voices, the exception being positive beliefs about voices. All but one of the beliefs about voices also showed correlations with one or more of the schema scales. In line with predicted relationships, the BAVQ-R scales omnipotence and malevolence were each positively correlated with negative-self, and malevolence showed the expected additional positive correlation with negative-others. However, contrary to predictions, benevolence was unrelated to positive-others, instead showing a trend, p = .051, to be associated with positive-self. On the IVI, metaphysical and loss of control beliefs were also each associated with negative-self, again in line with predictions. However, loss of control beliefs showed an additional unexpected correlation with negative-others. The predicted relationship between positive beliefs about voices and negative-others approached significance, p = .078, although positive beliefs about voices also showed an unpredicted relationship with positive-self.
Table 2. Bivariate correlations between beliefs about voices and both negative voice content and schemas
Notes: N = 34. PSYRATS, Psychotic Symptom Rating Scales; BCSS, Brief Core Schema Scales; BAVQ-R, Revised Beliefs About Voices Questionnaire; IVI, Interpretation of Voices Inventory. † p < .10; * p < .05; ** p < .01; *** p < .001
Hierarchical regression analyses examining voice content and schemas as predictors of beliefs about voices were then conducted, each entering the two negative voice content variables as a first step, and the schemas scales identified above as correlated with that belief as a second step (see Table 3). The estimated proportion of variance explained by negative voice content alone ranged from 7% to 25%. Entering schemas in the second step explained statistically significant additional variance for five of the six beliefs, with the variance explained for the remaining belief (benevolence) falling just below significance. The estimated proportion of additional variance explained by schemas ranged from 9% to 35%, with the total variance explained by negative voice content and schemas combined ranging from 24% to 54%.
Table 3. Hierarchical multiple regression analysis predicting beliefs about voices from negative voice content (Step 1) and schemas (Step 2)
Notes: N = 34. BAVQ-R, Revised Beliefs About Voices Questionnaire; IVI, Interpretation of Voices Inventory. † p < .10; * p < .05; * p < .01; ** p < .001
Table 3 also presents standardized regression coefficients for each of the schema scales. In considering these, it should be noted that the small sample size limits how precisely the parameter estimates of the regressions could be derived, so these should be regarded as approximate. On the basis of the observed bivariate correlations, more than one schema scale had been entered as a predictor in three of the belief about voices regressions: malevolence, loss of control, and positive beliefs about voices. For both the malevolence and loss of control regressions, the two predictors entered were negative-self and negative-others. Only negative-self showed an independent relationship with the belief about voices in each of these regressions. For positive beliefs about voices, positive-self emerged as an independent predictor, with the hypothesized relationship with negative-others approaching significance, p = .058.
In line with previous observations (Chadwick et al., Reference Chadwick, Lees and Birchwood2000; Peters et al., Reference Peters, Williams, Cooke and Kuipers2012; Simms, McCormack, Anderson and Mulholland, Reference Simms, McCormack, Anderson and Mulholland2007; So and Wong, Reference So and Wong2008), omnipotence and malevolence were positively correlated, r = .68, p < .001. To isolate the effects of schemas on voice malevolence from those already observed in the literature on the voice power dimension, the regression for malevolence was rerun including the omnipotence scale as a covariate. Doing this, findings changed, with negative-others becoming the sole significant predictor, β = .29, t = 2.097, p = .045, in place of negative-self, β = .02, t = 0.132, p = .90.
Given the observed correlation between beliefs about voices and length of illness, the regressions were also rerun, entering length of illness as an additional predictor. Length of illness was a significant independent predictor of only one of the six beliefs (positive beliefs about voices, β = .33, t = 2.218, p = .035), and entering it did not alter any of the other observed results. The regression of malevolence was also rerun entering the identified potential confound of voice loudness, but it did not emerge as a significant predictor or alter the pattern of results.
Discussion
Overall, the findings suggest that schemas are meaningfully related to the full range of beliefs, which have been described as important in mediating adjustment to hearing voices, including both the beliefs about voice intent (malevolence and benevolence) described by Chadwick and Birchwood (Reference Chadwick and Birchwood1994) and the metacognitive beliefs about voices described by Morrison et al. (Reference Morrison, Wells and Nothard2002).
We examined whether beliefs about voices were related to both negative voice content and to schemas. In line with previous observations (Birchwood and Chadwick, Reference Birchwood and Chadwick1997; Chadwick and Birchwood, Reference Chadwick and Birchwood1994; Close and Garety, Reference Close and Garety1998; Van der Gaag et al., Reference Van der Gaag, Hageman and Birchwood2003), beliefs about voices appeared to be influenced by negative voice content: a significant proportion of variance was predicted by the combined negative voice content scales in the regressions for omnipotence, metaphysical and loss of control beliefs, and there were trends for the malevolence and benevolence regressions. These relationships arose even though beliefs about voices appeared mostly unrelated to other aspects of voice experience such as frequency, duration, loudness and location. Our study extends previous findings of the relationship between content and appraisals of voices by providing data on relationships with the two aspects of negative voice content assessed by the PSYRATS: the amount and the degree of negative content. The results suggest that all of the negatively valenced beliefs about voices (omnipotence and malevolence on the BAVQ-R; metaphysical and loss of control beliefs on the IVI) tend to be predicted by the degree, rather than amount, of negative content. In other words, it was not how often voice content was negative but how negative it was when it occurred that predicted these beliefs. Conversely, benevolence was inversely correlated with the amount of negative content, suggesting that positive appraisals of voice intent are less likely to arise when there is a significant proportion of negative voice content. Positive content was not assessed, which may additionally contribute to the formation of benevolence beliefs (Van der Gaag et al., Reference Van der Gaag, Hageman and Birchwood2003) as well as positive beliefs about the value of voice experience. Overall, these observations reinforce that there are meaningful associations between voice content and beliefs about voices, that studying these may be of value in understanding how appraisals of and responses to voices develop, and that controlling for voice content is an important consideration in future studies of beliefs about voices.
After controlling for voice content, five of the six beliefs about voices were significantly predicted by schema scales, with the sixth approaching statistical significance. The proportion of additional variance explained by schemas was at least as much as that explained by negative content. Whilst this cross-sectional study does not provide data on the direction of causality, these results support the key tenet of cognitive theory that core schemas play a role in formation of beliefs. Our results suggest that general schemas influence the formation of not only appraisals of voice social rank and power (e.g. omnipotence), as has been previously demonstrated, but also those of voice intent (e.g. malevolence) and metacognitive beliefs about voices (e.g. fears of loss of control). This provides a means of accounting for observed individual differences in adaptation to hearing voices: the person's broader schemas influence the extent to which they interpret the phenomenon of hearing voices as a personal or social threat, in turn impacting upon adaptation and distress.
This examination of the respective contributions of cognitive representations of self and of others was preliminary, given the sample size limitations of the study. However, it is notable that there was a general pattern in which negative-self schemas were more widely associated with beliefs about voices than were negative-other schemas. Negative self-schemas were linked to malevolence, loss of control and metaphysical beliefs in addition to omnipotence, although the relationships with malevolence and negative metaphysical beliefs were no longer significant when partialing out variance shared with omnipotence. The association between omnipotence and negative-self concurs with findings by Birchwood et al. (Reference Birchwood, Meaden, Trower, Gilbert and Plaistow2000, Reference Birchwood, Gilbert, Gilbert, Trower, Meaden and Hay2004) that the rated power and status differential between self and voice is associated with appraisals of one's own relative power and status in the social world. The metaphysical and loss of control scales have been found to be associated with intolerance of uncertainty (White and Gumley, Reference White and Gumley2010), but have not otherwise been significantly studied. Finding an association with negative-self schemas suggests a potential broader relationship with negative self-evaluative beliefs.
When examining the role of cognitive representations of others, the anticipated bivariate relationship between malevolence and negative-others was found, but, as this was confounded with omnipotence, further examination with a large sample would be required to clarify the relative contributions of negative schemas about self and about others in explaining variability in appraisals of voice malevolence. Even so, the predicted association between benevolence and positive-others did not emerge, and instead benevolence appeared primarily related to an absence of negative content. This suggests that positive representations of others are not significantly involved in the formation of this belief, or have only a small effect. Finally, the predicted relationship between positive beliefs about voices and negative schemas about others was partially supported by this predictor approaching statistical significance, lending credence to the proposal of Morrison et al. (Reference Morrison, Frame and Larkin2003) that engagement with voices may develop as response to negative real-life relationships. However, this predicted relationship was overshadowed by positive-self unexpectedly being a stronger predictor of positive beliefs about voices. Positive views of voices have often been regarded as more adaptive because they are more readily endorsed by nonpsychiatric voice hearers (e.g. Sorrell, Hayward and Meddings, Reference Sorrell, Hayward and Meddings2010), which may account for an association with positive self-representations. It seems possible that endorsement of the positive metacognitive belief scale more often reflected people applying an optimistic view to their voice experience than positive self-regulatory beliefs about engagement with voice experience.
Few other studies have examined potential determinants of these beliefs about voices, although some attention has been paid to early experiences. Offen, Thomas and Waller (Reference Offen, Waller and Thomas2003) found that malevolence was predicted by recollections of paternal (but not maternal) overprotection, and in another study the same authors reported a correlation with early childhood sexual abuse (Offen, Waller and Thomas, Reference Offen, Thomas and Waller2003), whilst Andrew et al. (Reference Andrew, Gray and Snowden2008) reported that higher ratings of malevolence and omnipotence and lower ratings of benevolence were associated with past trauma. Schemas may provide a pathway by which these early experiences impact upon the formation of beliefs about voices and, in turn, the person's resulting relationship with their hallucinatory experience. Future study including measures of early experiences in conjunction with schemas may clarify a possible mediating role.
In addition to schemas influencing appraisals of voices, some models have proposed that schemas may also directly influence voice content (e.g. Beck and Rector, Reference Beck and Rector2003; Paulik, Reference Paulik2012). Data in our current study did not support this, with no correlations observed between any of the schema scales and either of the negative voice content scales. This is in contrast to reports that voice content very often appears meaningful in the context of the person's history (e.g. Reiff, Castille, Muenzenmaier and Link, Reference Reiff, Castille, Muenzenmaier and Link2012). This discrepancy may, however, reflect voice content arising from relatively specific mental representations: this is consistent with models of hallucinations as intrusions of specific intrusions from memory not recognized due to a failure to encode contextual cues (e.g. Steel, Fowler and Holmes, Reference Steel, Fowler and Holmes2005; Waters, Badcock, Michie and Maybery, Reference Waters, Badcock, Michie and Maybery2006). Hence voice content may reflect relatively specific cognitions, whilst appraisals of their meaning and development of ideas about their identity are influenced by more generalized self and other representations. Indeed, as schemas also appear unrelated to the overall severity of voices (Smith et al., Reference Smith, Fowler, Freeman, Bebbington, Bashforth and Garety2006), the findings of this study suggest that schemas relate primarily to adaptation to voice experience rather than to mechanisms involved in voice formation.
Clinically, identifying schematic beliefs related to beliefs about voices potentially provides an additional therapeutic target for psychological intervention. They may provide a parallel focus of intervention, or an alternate target when beliefs about voices are held with too strong conviction to provide a workable focus. Schemas associated with malevolence may be particularly important in providing a potential target for cognitive restructuring because no trials of cognitive-behavioural therapy (CBT) for psychosis that have included voice malevolence as an outcome have observed changes in this dimension (Peters et al., Reference Peters, Landau, McCrone, Cooke, Fisher and Steel2010; Trower et al., Reference Trower, Birchwood, Meaden, Byrne, Nelson and Ross2004; Valmaggia, van der Gaag, Tarrier, Pijnenborg and Slooff, Reference Valmaggia, van der Gaag, Tarrier, Pijnenborg and Slooff2005). Indeed, even though malevolence beliefs are robustly associated with distress, methods for working with these beliefs have not tended to be included in descriptions of CBT, which primarily focus on modifying beliefs about voice power, and control over the experience and the origins of voices (e.g. Trower et al., Reference Trower, Birchwood, Meaden, Byrne, Nelson and Ross2004; Chadwick et al., Reference Chadwick, Birchwood and Trower1996; Morrison and Renton, Reference Morrison and Renton2001). Beliefs in voice malevolence are not easy to directly modify into an alternative belief without either colluding with the idea that voices are sentient others, or challenging the person's overall explanatory model. Hence schemas that may be supporting these beliefs about voices represent an important alternate avenue for intervention. In addition to traditional cognitive restructuring methods, methods of working with voices are emerging that adopt a more experiential approach, which may be well-suited to intervening at a schematic level (e.g. Chadwick, Reference Chadwick2003; Mayhew and Gilbert, Reference Mayhew and Gilbert2008; Van der Gaag, Van Oosterhout, Daalman, Sommer and Korrelboom, Reference Van der Gaag, Van Oosterhout, Daalman, Sommer and Korrelboom2012). Other suitable approaches include methods from acceptance and mindfulness-based therapies that can defuse the impact of negative self-related ideation (Thomas, Morris, Shawyer and Farhall, Reference Thomas, Morris, Shawyer, Farhall, Morris, Johns and Oliver2013), and methods described for working with voices in a broader interpersonal context (Hayward, Overton, Dorey and Denney, Reference Hayward, Overton, Dorey and Denney2009). Further study on identifying the specific content of schemas associated with voice beliefs may inform more precise targets of these developing interventions.
This study was a preliminary investigation. Descriptions of voice content were limited to scales capturing overall levels of negativity in content that would not have been sensitive to variation in different content themes, which may have had closer relationships with schemas. Likewise, there was no specific assessment of positive or neutral voice content. Conversely, ratings were dependent upon participant report during interview, so participant reports of the amount and degree of negative content may have been influenced by broader beliefs about voices inflating their potential relationship. The schema scales also focused on overall negative versus positive themes, so would not have been sensitive to nuances of meaning or emotional themes that may have greater congruity with aspects of voice experience. Further study examining the relationships between beliefs about voices, content and schemas may focus on this in more detail. In particular, the role of negative-other schemas is demanding further clarification, given the sample size limitations of this present investigation. A further limitation of the study is the use of a treatment-seeking sample that will not be representative of the full spectrum of voice hearers.
Nonetheless, the study suggests that schemas offer a promising means of conceptualizing the formation of beliefs about voices, and possess potential as a therapeutic target in the future development of interventions for voices. When working clinically with people who hear voices, broader schemas may be worth incorporating into formulations of the person's experience and interpretation of their voices.
Acknowledgements
This research project was supported by funding from the William Buckland Foundation.
Comments
No Comments have been published for this article.