![]()
Chapter 1
Understanding command hallucinations
In schizophrenia research, considerable progress has been made in recent years towards understanding the psychological and interpersonal characteristics of hallucinations. In this context, one particular class of hallucinations, namely command hallucinations (CHs) has recently become a focus of theory, research and clinical intervention. Indeed, the importance of CHs has become clear for both theoretical reasons â the light that these symptoms throw on the psychological nature of positive symptoms in general â and practice reasons, since CHs are one of the most high-risk and distressing, but intractable and drug-resistant, symptoms of schizophrenia.
What are command hallucinations? CHs have long been recognised but little understood, with few effective interventions. The key feature that distinguishes them from ordinary hallucinations is that phenomenologically the voice is experienced or interpreted as commanding rather than commenting. The perceived commands range from making a harmless gesture to behaving in ways that are potentially injurious or lethal to self or others. Although there is general agreement on these features in the literature (indeed Bleuler (1924:62) referred to command hallucinations having a âcompulsive powerâ making them difficult to ignore), very little in the way of specific diagnostic guidelines is available, and they are not specifically mentioned in DSMIV (American Psychiatric Association, 1994). There is clearly a need for such guidelines, for example whether the term CHs should be confined to voices where there is an explicit command or whether it should include voices where the client interprets that a command is implied. In our approach we have adopted the latter criterion. As we shall describe later, we have also developed our own rating for severity of command, and a risk assessment grading system for degree of compliance or resistance.
How common are CHs in general? In a recent review, Shawyer et al. (2003) found eight studies reporting the prevalence of CHs in samples of adult psychiatric patients with auditory hallucinations. The median prevalence rate for these studies is 53 per cent, but with a very wide range of 18 to 89 per cent. Equivalent prevalence rates in forensic populations were no higher.
How common are dangerous or harmful CHs? Shawyer et al. (2003) report a median prevalence of 48 per cent for harmful CHs in non-forensic patients with CHs, i.e. roughly half, but again the range was considerable, from 7 to 70 per cent. Forensic groups were almost certainly much higher, with 83 per cent of voice hearers found to have CHs with âcriminalâ content.
How common is compliance with dangerous or harmful CHs? This is the question of most concern to clinicians. A median prevalence of at least partial compliance to harmful CHs is 31 per cent in community samples, though yet again the range is very wide (0â92 per cent (Shawyer et al., 2003)) and therefore the picture is far from clear. In the forensic population results suggested that rates were higher. In summarising, Shawyer et al. said the evidence suggests that CHs may be associated in a complex manner with violence. Nonetheless, harmful CHs to hurt others and to hurt self have been found to be associated with an increased risk of violence toward others and self-harm respectively.
The figures quoted above clearly show that patients donât always comply with their commanding voices, and we know from our own research that they sometimes refuse altogether and sometimes resist but comply in a symbolic or minor way as a gesture of appeasement to the voice. Why they respond in these varying ways is a vitally important issue that we explore in detail later. At this point we simply want to illustrate the kinds of commands that are heard and the range of responses elicited. We obtained an indication of these data from our recent trial (see Chapter 11) in which all 38 patients who entered the trial reported two or more commands â at least one of which was a âsevereâ command â and had recently complied with the command. The most severe commands were to kill self (25), kill others (13), harm self (12) and harm others (14). Less severe commands involved innocuous, day-to-day behaviour (wash dishes, masturbate, take a bath) and minor social transgressions (break windows, shout out loud, swear in public). Further details including incidence and examples of compliance and appeasement of these commands for the sample as a whole are shown in Table 1.1.
What kind of therapy is provided and what is effective? It wonât come as a surprise that this patient group requires and, on the whole, receives a major and inevitably expensive share of mental health service resources. Apart from specialist services such as forensic services and detention in semi-secure units, this group receives input from virtually every relevant professional and type of service, both inpatient and outpatient. We found, for example, that no fewer than 19 categories of service were provided for the 38 patients in our trial, reported in Chapter 10. So provision is expensive, but is it effective? There is little definitive data on this question, and what there is is not encouraging. Sawyer et al. found that patients who complied with their CHs were receiving significantly higher doses of antipsychotic medication than those who did not comply, suggesting that medication may have been ineffective in suppressing their CHs. Indications are that CHs feature strongly in those considered âtreatment resistantâ and even hospitalisation is not necessarily a barrier to
Table 1.1 Prevalence of and types of commands, compliance and appeasement in a sample of 38 individuals with command hallucinations1
compliance (e.g. Jones et al., 1992). Hospitalisation obviously reduces risk to the public but it simply restricts physical opportunities to act on CHs and can hardly be considered an effective âtherapyâ for the individual. In our trial (Trower et al., 2004) we report the very high medication dose that is commonly used, its propensity to creep upwards over time and the absence of a link between drug dose and compliance behaviour. So, in summary, services are expensive but probably not very effective.
The link between CHs and harm to self or others is not straightforward, however. Population studies suggest that people with a diagnosis of schizophrenia are more at risk of harm to self and to others (Brennan et al., 2000) than those without psychosis, but the link between the form of individual symptoms and the risk of individuals acting on these has proved difficult to establish (Milton et al., 2001; Appelbaum et al., 2000; Buchanan, 1993), including the risk associated with the presence of command hallucinations in the MacArthur study (Appelbaum et al., 2000). Some argue that there is no link between the presence of command hallucinations and âdangerous behaviourâ (Rudnick-Abraham, 1999); but the MacArthur study is not without its critics (Maden, 2003). However, we have argued (Braham et al., 2004) that these population studies focus on the form of the command hallucination and ignore the content and the nature of the individualâs relationship with his voice, which we discuss later.
A cognitive model for CHs
It is often assumed that symptoms such as hallucinations and associated affect and behaviour such as compliance are directly causally related within the syndrome of schizophrenia. However, as we have already seen, there is considerable variation in CHs between the command and the emotional and behavioural response to the command, such that there has to be some mediating variable between the two. In a critical review of the literature, Braham et al. (2004) found that the relationship between command and compliance was mediated by a number of factors, including beliefs about the voiceâs identity, familiarity, power and intent. The studies reviewed support the view that the beliefs that an individual holds about their voice will influence compliance, and that a cognitive model is required to explain the relationships and guide clinical intervention. In a cognitive model of hallucinations, Chadwick and Birchwood (1994) and Birchwood and Chadwick (1997) showed empirically that the distressing affect and behaviour arising from hallucinations may be understood as a function not simply of the content or topography of voice activity but also of voice hearersâ appraisal of their meaning. One of the key insights in this model, drawn and adapted from rational emotive behaviour therapy, is to view the hallucination as an activating event (A), whose significance is appraised by the individual in terms of their belief system (B), and which largely gives rise to characteristic emotional and behaviour
consequences (C). This is illustrated in Figure 1.1. Here we show voice activity as the activating event that triggers the key beliefs of power and dominance/subordination (explained below). These beliefs give rise, according to our cognitive theory, to the consequences â principally the emotions of fear, guilt and depression and sometimes elation, and the behaviours of compliance or appeasement. These behaviours can be construed as âsafetyâ behaviours â part of a cognitive mechanism that maintains the belief in the power of the voice, which we explain below. Chadwick and Birchwood (1994) found evidence for this type of cognitive mediation in the maintenance of beliefs about voices: in many cases, belief content was âat oddsâ with voice content, suggesting that meanings are constructed by individuals rather than directly voice-driven. Indeed, participants in this study disclosed what was for them compelling evidence for their beliefs, which only occasionally drew upon voice content. These results were replicated by Van der Gaag et al. (2003).
A second discovery from this body of research has, we believe, thrown radically new light on the nature of the appraisal of voices. This is the finding that the distress and behaviour linked to voice activity may be understood in terms of the nature of patientsâ perceived relationship with their voices, in particular their personification of them and their appraisal of voicesâ âpower and omnipotenceâ and whether the voice is malevolent or benevolent (Chadwick and Birchwood, 1994). The key variables in this relationship can be summarised under the headings of power, identity and meaning. Power refers to beliefs the individual makes about how much they can or cannot control the voice and how much they believe they need to comply (see Figure 1.1). Identity refers to beliefs about âwhoâ the voice is: often this is a supernatural force, the Devil or God or a spirit. Meaning refers to beliefs about the voiceâs intent â that the voice must be obeyed or it will punish the person.
Recent research has found empirical support for the centrality of the power differential between voice and voice hearer in the level of distress and also coping strategies (Birchwood et al., 2000). This model has explanatory value when one considers the risk of compliance with command hallucinations. Junginger (1990) found that recent compliance was more likely where the individual personified, or attributed an identity to, the voice. Over 85 per cent of voice hearers see the voice as powerful and omnipotent whereas, by contrast, the hearer is usually perceived as weak and dependent, unable to control or influence the voice (Birchwood and Chadwick, 1997). Thus, it was found that the greater the perceived power and omnipotence of the voice, the greater the likelihood of compliance (Beck-Sander et al., 1997). This relationship is not linear and was moderated by appraisal of the consequences of resisting the voice on one hand, and the consequences of social transgression on the other. Results showed that those with benevolent voices virtually always complied with the voice, irrespective of whether the command was socially âinnocuousâ or âsevereâ (Beck-Sander et al., 1997). Furthermore, we have argued that the relationship with the voice is a paradigm or mirror of social relationships in general, such that individuals who feel subordinate to the powerful voice also feel subordinate to others in their social world (Birchwood et al., 2000, 2004).
From these and other studies we can now assert with some confidence the following findings.
1 Voice hearers construct the link between themselves and their voice as having the nature of an intimate interpersonal relationship and often one that is inescapable (Benjamin, 1989). In a cross-sectional study, Junginger (1990) found that recent compliance was more likely where the individual personified the voice (i.e. attributed it to an identity).
2 More than 85 per cent of voice hearers see the voice as powerful and omnipotent, whereas the hearer is usually weak and dependent, unable to control or influence the voice (Birchwood and Chadwick, 1997). This is particularly the case among those seen in psychiatric services.
3 More than two-thirds of voice hearers were at least moderately depressed, which was directly attributable to the interpersonal appraisal of power and entrapment by the voice (Birchwood and Chadwick, 1997) and to social appraisals of powerlessness and low social status (Birchwood et al., 2004).
4 The greater the perceived power and omnipotence of the voice compared to the voice hearer, the greater was the likelihood of compliance (Beck-Sander et al., 1997), though this relationship is not linear and is moderated by appraisal of the voiceâs intent and consequences of resisting.
5 Voice hearers perceive the voice as omniscient (e.g. as knowing the personâs present thoughts and past history, able to predict the future), and this was seen as proof of the voiceâs power.
6 Some voice hearers construed their voice as benevolent; others as malevolent and persecutory (Chadwick and Birchwood, 1994; Birchwood and Chadwick, 1997).
7 Those with benevolent voices virtually always complied with the voice, irrespective of whether the command was âinnocuousâ or âsevereâ (Beck-Sander et al., 1997), whereas those with malevolent voices were m...
