Investigating the Lived Experience of Recovery in People Who Hear Voices

Although there is evidence of both clinical and personal recovery from distressing voices, the process of recovery over time is unclear. Narrative inquiry was used to investigate 11 voice-hearers’ lived experience of recovery. After a period of despair/exhaustion, two recovery typologies emerged: (a) turning toward/empowerment, which involved developing a normalized account of voices, building voice-specific skills, integration of voices into daily life, and a transformation of identity, and (b) turning away/protective hibernation, which involved harnessing all available resources to survive the experience, with the importance of medication in recovery being emphasized. Results indicated the importance of services being sensitive and responsive to a person’s recovery style at any given time and their readiness for change. Coming to hold a normalized account of voice-hearing and the self and witnessing of preferred narratives by others were essential in the more robust turning toward recovery typology.

Voice-hearing, referred to in the clinical literature as having auditory hallucinations, is commonly experienced in the context of schizophrenia and other psychological disorders, as well as in the general population (Beavan, Read & Cartwright, 2011;Choong, Hunter & Woodruff, 2011;Waters, 2010). The experience is highly stigmatised and associated with poor mental health, distress, and isolation (APA, 2000;Thornicroft, 2006). The predominant approach to treating distressing voices in the context of a psychiatric diagnosis is defined by the medical model, which aims to eliminate symptoms, primarily by administration of antipsychotic medication (Drake, Bond, & Essock, 2009;Read, Bentall, & Fosse, 2009;. The definition of recovery varies between dominant medical model discourse versus consumer / survivor discourse, with the former focussing on symptom resolution and a return to a former state of functioning (clinical recovery), whereas the latter emphasises living a "fulfilling and satisfying life, in the presence or absence of symptoms" (Carlton & Bradstreet, 2006, p16) (personal recovery) (Bellack, 2006;). The medical model approach has been criticised for a narrow definition of recovery and pessimism regarding prognosis (Breeding, 2008;Fleming & Martin, 2011;Read et al., 2009). General population surveys indicate that a proportion of people who hear voices are not negatively affected by them (Beavan & Read, 2007;Johns & van Os, 2001;Tien, 1991). Thus, hearing voices in and of itself does not necessarily cause distress (Peters, Williams, Cooke & Kuipers, 2012). In addition, a substantial proportion of people who experience distressing voices functionally recover to a degree which would be considered 'normal' by most people, with approximately 50% meeting criteria for clinical recovery over a 20-year period .
In contrast to clinical recovery, which has quite clear criteria, personal recovery is a deeply personal process and its definition varies from person to person (Brown, 2008). It is possible, however, to identify common threads in recovery processes and identify factors that support well-being (Brown, 2008). Recovery in this broader sense involves developing or regaining a positive sense of self, agency, hope, meaningful occupation, making sense of or reframing experiences, accepting the limitations associated with the illness and creating a satisfying life within those limitations, and improved quality of life (Andresen, Oades & Caputi, 2003;Perkins & Slade, 2012). Others emphasise having the same work and living opportunities as other community members (Kirkpatrick & Byrne, 2009). A pivotal tenet of recovery-orientated approaches is that recovery is possible (Anthony, 1993(Anthony, , 2000White, 2005). Influenced by consumer/survivor groups, interest in recovery-orientated mental health services and treatment has grown over the past twenty years (Anthony, 2000).
One of the most prominent strands of the consumer/recovery movement has been the hearing voices movement. This was born out of collaboration between consumers, clinicians and researchers (e.g. ), leading to establishment of extensive national and state networks which facilitate the running peer-support hearing voices groups (HVGs) and promote connection between voice hearers, referred to collectively as the Hearing Voices Network (HVN; Corstens, Longden, McCarthy-Jones, Waddingham & Thomas, 2014). Principles adopted within the HVN align with both consumer definitions of recovery and the need for a more holistic treatment model which normalises voices and takes into account contextual factors (Fischer, 2003). Distressing voices are understood as a manifestation of solvable emotional problems which may present in a figurative or literal manner (McCarthy-Jones & Longden, 2013). Derived from personal accounts of voice hearing discussed at the first major voice hearer congress in 1987, Escher (1989,1993) proposed three phases of recovery: startling, which involves initial feelings of shock and being overwhelmed by voice-hearing experiences and trying to push voices away, organisation, whereby the person starts to employ some strategies in relation to their voices and place boundaries around their interactions and stabilisation, wherein the hearer comes to integrate voices into their life, recognise them as 'part of me' and no longer wishes to be rid of them.
Although evidence to date indicates that recovery from distressing voicesboth clinical and personalis possible, there a lack of research into the recovery process in relation to hearing voices. Although many first-person accounts of recovery exist and are valuable in and of themselves (e.g. , they were not generated using a qualitative method that allows for analysis of narrative genres or typologies of recovery over time. It is therefore unclear whether recovery processes align with the HVN approach. In addition, diagnoses and stage of recovery were not confirmed. This article addresses this gap in the literature, employing Narrative Inquiry to examine recovery and including quantitative and diagnostic measures.

Method
Narrative analysis is the analysis of text, in this case interview transcripts, in 'storied' form . A narrative text involves the telling of a sequence of events, which, in the act of narrating are organised by the narrator. The narrator interprets the text, selecting information to include, how to describe it, and how various elements are connected .
Therefore, narrative analysis focuses on the narrator's construction of meaning: in other words, not only what happened but how they understand or make sense of these events. It privileges knowledge gained from personal experience as opposed to master narratives or dominant discourse surrounding a given topic. The notion that meaning is partly or entirely socially constructed is implicit in the philosophical underpinnings of narrative analysis. It acknowledges the role of the researcher in the process of meaning-making. Researchers must organise information generated from narrative interviews and interpret it . Research output represents how the researcher has made sense of how the participant has made sense of their experiences. Taking an objective stance is not considered possible; instead, the researcher is required to be aware of and reflect on their subjective response to the research context, process and content (Braun & Clarke, 2008;Oke, 2008). The first author, who undertook interviews, collection of questionnaire data and lead data analysis, was aware of their tendency to understand voice-hearing as meaningful in context of a person's life history, and expectation that active engagement and making sense of voices would be involved in recovery.
Bracketing these assumptions during the research process allowed information which was not in keeping with expectations to be reflected on and acknowledged, rather than obscured (Fischer, 2009;Tufford & Newman, 2012).
In keeping with trends in psychosis research  and because of the advantages it affords, a single complaint approach was employed (Bentall, 2003), examining voice-hearing rather than broader diagnostic constructs. However, it is important to confirm whether or not participants met criteria for a diagnosis during their life-time. This allows comparison with research utilising diagnostic constructs and also confirms that recovery has occurred from a clinically significant degree of distress. In addition to semistructured interviews regarding recovery over time, participants completed the Diagnostic Interview for Psychoses Huxley, Knight & Evans, 1999) is a self-report measure of subjective quality of life which assesses satisfaction with various domains including friendships, finances, leisure activities, safety and health (both physical and mental). The Kessler Psychological Distress Scale (K10; Kessler et al., 2002) is a 10-item self-report measure of global psychological distress based on items concerning depression and anxiety symptoms over four weeks prior to assessment.
Diagnostic interview. The DIP (Castle et al., 2006) is a structured interview employed to generate DSM-IV and ICD-10 diagnoses for research purposes. Responses are analysed to produce a diagnosis.

Semistructured interview. A semistructured interview was conducted with a view to
eliciting information about participants' experiences of voice-hearing and recovery over time. A time-line was used to indicate when voices first started, when they were at their worst and how participants progressed from that point toward recovery.

Procedure
Ethical approval for the research was obtained from the University of Sydney Human Ethics Committee. This research was advertised through the HVNNSW and ASRB. Reimbursement was offered for travel expenses and time spent taking part. Interested participants contacted the researcher and the Participant Information Sheet was provided. Informed consent was obtained from all participants in writing. Inclusion criteria were 1) current or historical experience of distressing voices 2) self-identified as recovered or recovering 3) adequate English language skills. People experiencing acute levels of distress were excluded. Interviews were conducted at a mental health centre in Sydney and took between sixty and ninety minutes, with breaks as required.
Lifetime diagnosis was derived either by administering the DIP or by obtaining DIP data held on the ASRB databank. All participants completed the self-report measures. This data provided triangulation regarding recovery.
Semistructured interviews concerning participants' recovery over time were conducted and audio-recorded. Interviews were transcribed and narratives of recovery over time generated by the first author. Participants were given pseudonyms and identifying details removed or in one case changed to protect participants' confidentiality. Narratives were sent to participants for member-checking to ensure that they were an authentic and accurate representation of participants' experiences. Information gleaned from interviews was used to inform subsequent interview questions (Connelley & Clandinin, 2006).
A preliminary model of recovery over time and typologies (types of recovery over time) was generated. A panel of four consultants who were experts by profession and / or experience (VB, NT, DL and SM) were asked to read the narratives and critique this model, providing feedback and suggestions for improvement. The model was revised incorporating four main critiques: 1) the need to exercise caution in producing models of recovery which suggest linear progression not in keeping with reality and the danger of clinicians applying such models in a prescriptive manner 2) inclusion of disconnection from others 3) noting factors which inhibited as well as encouraged recovery and 4) checking the data to confirm the identified period of exhaustion / despair as well as proximal and distal stressors.

Results
The results of the DIP indicated that during their lifetimes, participants had met DSM-IV diagnostic criteria for schizophrenia (4), schizoaffective disorder (4) or psychosis not otherwise specified (3). Self-report measures showed that nine participants were in stage 5 (growth) of recovery and the remaining two were in stage 4 (rebuilding). Participants reported mixed (5) or mostly satisfied (6) ratings of quality of life, and a broad range of global psychological distress (4 low, 2 moderate and 5 high distress). One participant with a self-reported diagnosis of schizoaffective disorder did not consent to taking part in the diagnostic interview. Their diagnosis is thus based on self-report.
Participants' accounts indicated that recovery was not a neat, linear progression over time. They did not move through each stage progressively. Instead, stages were cyclical in nature, with participants repeating or re-engaging in processes associated with recovery several times or in an ongoing manner. Processes occurred simultaneously rather than discretely and influenced each other. Participants also took time to rest and consolidate the progress they had made.
Given the shortfalls of any model of recovery over time in capturing the essence of individual narratives, as well as the capacity of models to oppress such narratives, one may well ask why production of such a model is advisable or indeed of value. However, to consider what type of story a person is involved in and which stories are being overlooked is to consider how their lives are defined (Frank, 2010). Identifying recover typologies allows us to elucidate and reflect on the (usually obscured) power and function of such narratives and the concrete impact they have on people's lives (Frank, 2010;Madigan, 1992). The description of recovery over time presented in Figure 1 is thus an approximation of what participants described and included to provide a rubric through which to consider recovery.

Context of Voice Onset and Initial Responses
Most participants described poor general mental health, distal and proximal stressors and disconnection from others prior to voice onset. Participants feared the consequences of disclosing their experiences to others, expressing concern that they would be judged negatively because of stigma. They were acutely aware of the risk involved in disclosing that they were hearing voices to mental health professionals, fearing that disclosure may result in invasive treatments or involuntary admission to a psychiatric hospital:

Reaching a Point of Despair / Exhaustion
Overall, strategies initially trialled for dealing with voices including resisting them, avoiding feared situations and people related to voices, were ineffective. Participants often described ongoing poor general mental health. In this context, participants became increasingly overwhelmed by their voices, which typically became louder and more intrusive after initial onset: They gave me some medicine and told me it would get rid of it, but they only just kept getting worse and worse and worse . . . until a week or two later, it was virtually nonstop, these two guys talking to each other about me.
Participants were generally less able to disengage from their voices or resist their instructions, as a result sometimes behaving in ways that were not in line with their values: Medication also eliminated voices in some cases or allowed participants to change their response to their voices, enabling them to hold a more comfortable distance from them. For example, instead of being overwhelmed, participants were able to listen to voices for a few minutes and then distract themselves from them.
Overall, for participants with a dominant turning away narrative, reflecting on their experiences and how they were able to survive and recover was a great deal more effortful than for turning toward participants. They gave sparser, less detailed information and appeared to be unused to telling their stories. These participants, for whom medication was effective, were perhaps unsurprisingly more accepting of a medical model explanation for their voices as symptoms of an illness. These narratives were told stoically, with little or no reflection on the meaning of voice-hearing or curiosity about voices or how to interact with them. Instead, there was a strong sense of wanting to put the experience behind them and get on with their lives.

Turning Toward
Turning toward narratives were characterised by a tendency to turn to face problems, active Overall, the importance of medication in recovery was not emphasised in turning toward narratives and many participants commented that it was ineffective or associated with significant side-effects which out-weighed its benefits. They were characterised by a tone of empowerment, and at times, defiance. There was a commitment to advocacy for the rights of other voicehearers. Participants described significant changes within themselves and their world view as a result of their voice-hearing experiences and recovery.

Common Processes Across Typologies
Across typologies, participants emphasised the importance of learning about how to nourish general mental health. For example, many participants commented that voices amplified existing anxieties or vulnerabilities: The voices always played on what was your vulnerable point.
Learning how to manage anxiety meant that when voices attempted to amplify concerns they had less traction or believability and participants were less distressed by their comments. This was a strong theme across narratives. Participants also learned how to manage their mood, balance activity levels to ensure that they were not over-or under-active, manage stress, ensure they got enough sleep and developed communication and assertiveness skills. Although at times mental health professionals played a role, it was also common for these skills to be developed without external assistance. Many participants used mindfulness and distraction techniques to create some space between themselves and their voices: (Mindfulness) is really helpful . . . even if I think I do hear someone swearing at me, I don't have to take it on board for myself.
As shown in Figure 1, engaging in meaningful activities, connecting with others and (re-) developing a positive sense of self were key recovery processes common to both typologies.
These processes were centred around meeting needs to feel competent, valued, purposeful and connected to others. They were strongly inter-related. Participants' experiences of mental health services and treatment varied. Mental health services were perceived as helpful when medication was effective and professionals worked with participants to develop strategies for dealing with mental health difficulties generally and voices specifically, and who spoke about voices without panicking. Access to housing, financial support, rehabilitation and mental health services as well as anti-discrimination laws facilitated recovery.

Turning Away and Turning Toward
This narrative research aimed to investigate recovery trajectories and critically appraise and further develop Romme, Escher and colleagues' seminal work on recovery from distressing voices, in particular the suggested startling, organisation and stabilisation phases of recovery . In keeping with prior research, participants in the current research understood negative life events to be causally related to voice onset (e.g. . They described an initial response of shock, confusion and fear in response to voices, loss of sense of self, and becoming overwhelmed. This is in keeping with the startling phase. However, from this point, the research identified two typologies of recovery, rather than just one: turning toward and turning away. Turning toward narratives involved a transformation of self, active engagement with voices, adopting a normalised view of voices and integrating them into one's life. This provides support for  organisation and stabilisation phases. These narratives align with McGlashan, Levy and Carpenter's (1975) integrative recovery style, which involves curiosity about psychotic experiences, striving to make sense of them and integrating them into one's life, as well as Frank's (1995) quest automythology narratives, which are characterised by a transformation in identity in response to physical illness. This also resonates with the Māori concept of voices as a 'difficult gift' which are nonetheless an ordinary part of daily life (NiaNia & Bush, 2013).
Participants who were seen to 'turn away', however, had also recovered according to our objective criteria, in terms of symptoms, quality of life and psychological distress. The focus was on the resolution or management of symptoms and leaving these experiences behind, rather than integration. This finding indicates a different type of recovery which does not involve active engagement, negotiation, acceptance and meaning-making characteristic of the suggested organisation and stabilisation phases. Instead, this typology bears parallels to McGlashan et al.'s (1975) sealing over recovery style, characterized by cordoning off psychotic experiences from the rest of one's life, and Frank's (1995) quest memoir narratives, which involve acceptance of illness with trials told stoically and no special insight gained (France, Hunt, Dow & Wyke, 2013). In considering this narrative, it is critical that these participants are not pathologized when compared to those who 'turned towards.' The latter group had the benefit of HNV groups and professionals who supported them to normalise and respond to their voices. Turning away participants can perhaps be seen as more remarkable because they did not have access to such discursive resources.
While taking care not to pathologize either recovery typology, the stories that participants told about themselves were more restrictive in turning away narratives and more liberative for those who turned towards. Indeed, there is evidence that in the long term, an integrative approach towards psychotic experiences confers better psychological health, predicting remission at oneyear follow-up (Staring, van der Gaag & Mulder, 2011). This raises the question of how clinical services should respond to people with different recovery styles. There is evidence that people's recovery style can change over time (Tait, Birchwood & Trower, 2004). Consequently, periodic assessment of a person's recovery mode would allow services to match their approach to the person's natural inclination at that time. If someone persists in using a turning away style and this is perceived as limiting recovery, how should services respond: acknowledge that a limited recovery has sufficient meaning to the individual, or seek to facilitate the adoption of a turning toward style? Longitudinal studies examining recovery style would allow further clarification of this issue.
It is of note that despite significant differences between typologies, many recovery processes were common across typologies. These provided the foundations for further progress.
They were centred around meeting participants' needs for connectedness, positive identity, agency, opportunities to contribute to the community and participate in meaningful activities (Andresen, Oades & Caputi, 2003;Reis, Sheldon, Gable, Roscoe & Ryan, 2000;. This indicates a need to shift from the traditional mode of treatment focussing on the individual to inclusion of family, social networks and broader community and social contexts (Sibitz et al., 2011;Tew, Ramon, Slade, Bird, Melton & Boutillier, 2012;Williams & Collins, 2002).

Voices, Self & Others
Many people experience an initial sense of their identity being subsumed by that of being a Reclaiming a positive sense of self appears critical to recovery (Goodliffe, Hayward, Brown, Turton & Dannahy, 2010;May, Strauss, Coyle & Hayward, 2014). Indeed, in the current research, participants described being less distressed by critical or threatening voices when they had developed a stronger sense of self. Recognition of the importance of people's self-esteem in mediating their affective response to negative voices has lead to this becoming a specific target of psychological interventions, with promising initial results (van der Gaag, Oosterhout, Daalman, Sommer & Korrelboom, 2012).
Reclaiming one's sense of self, however, is a social as well as individual process: the antithesis to personal diminishment which can result from stigma . As participants reconnected with others and developed supportive relationships, they were more able to deal with their voices in an adaptive manner. This is in keeping with research demonstrating significant parallels between how people relate to others and how they relate and respond to their voices (Chin, Hayward & Drinnan, 2009;Paulik, 2011). It is plausible that as people accumulated more positive experiences with others, their beliefs about others and perception of relative social rank changed, positively impacting on their style of relating to their voices. The tendency of people with turning toward narratives to engage with their voices rather than attempt to distance themselves is consistent with research indicating that relating to voices from a position of distance is associated with significantly higher distress (Vaughn & Fowler, 2004;. Engaging with voices and setting appropriate boundaries has been implicated in developing a positive relationship with voices . Indeed, in keeping with recent research, mindfully noticing voices contributed to a less distressing relationship with them (May et al., 2014;Newman-Taylor, Harper & Chadwick, 2009).

Discursive Resources and Making Sense of Voices
The narratives people tell define how they understand themselves and their problems (Lock, Epston, Maisel & de Faria, 2006;Weingarten, 1998). In the context of hearing voices, the dominant narrative, based in Foucauldian terms on 'global' (privileged) rather than 'local' knowledge, is that it is a symptom of a disease best treated by anti-psychotic medication (Madigan, 1992;. Implicitly, this narrative promotes the notion of the person as the problem (Ben-Zeev, Young & Corrigan, 2010;Madigan, 1992;van Os, 2010). Those who 'turned towards' in this research also 'turned away' from the dominant medical model discourse. This act of resistance was conducted in solidarity with others in HVN groups, allowing for the 'thickening' of alternative stories White & Epston, 1990). For example, participants described voices as a manifestation of distress understandable within their life context (e.g. Beavan & Read, 2007;. Although some participants still used aspects of a medical model explanation and vocabulary, they emphasised that voice-hearing is a normal variation in human experience, opening up the possibility of a non-pathologized self. This suggests that exposure to alternative understandings of voice-hearing, which might compliment rather than necessarily replace the dominant medical model narrative, might be helpful.

Parallels in Professional-Driven Treatments
Despite the impact of consumer-driven HVN groups in this research it is important to recognise that promising professional-driven treatments are also coming to similar conclusions regarding the value of direct engagement with voices and a focus on the person, not just the problem, particularly in the field of clinical psychology. Turning towards voices by exploration of continuities with broader life experiences has long been a key element of therapies such as cognitive behavioural therapy for psychosis (CBTp; see Farhall & Thomas, 2013).
Contemporary therapy developments have gone further to more explicitly engage with voices, including role playing interactions with voices (Hayward, Overton, Dorey, & Denney, 2009), the use of computer generated avatars to represent voices  and direct verbal engagement of voices by the therapist (Corstens, Longden & May, 2012).
As a cautionary note, findings of the current research suggest that timing is critical.
Turning toward participants reached a tipping point where doing something differently, although still frightening, was evaluated as a better option than continuing to tolerate their unbearable situation. This is similar to the crisis-induced change noted by Milligan, McCarthy-Jones, Winthrop and Dudley (2013). In addition, those engaged in a turning away style of response might find such treatments unhelpful and require some time to be ready to undertake such work.
These caveats aside, the critical role of interpersonal dialogue and solidarity is also becoming recognised in the network therapy of Seikkulla and colleagues (e.g. Seikkula, Alakare & Aaltonen, 2011). This approach involves intensive engagement with the client, the entire family and any other loved ones and professional stakeholders involved.
Treatment is based on the notion that psychotic reactions are pre-narrative or metaphorical; attempts to make sense of experiences that are so difficult that they have not yet been situated in spoken discourse .
Despite the evidence for the efficacy of these treatments, however, several barriers to accessing them remain, resulting in significant unmet need (e.g. Farhall & Thomas, 2013; Mojtabai et al., 2009;National Collaborating Centre for Mental Health, 2009). It is important to emphasise that not one single participant reported receiving psychological intervention that was specifically aimed at dealing with voices. Those who engaged in behavioural experiments and other CBT strategies did so without the help of a psychologist.

Role of Hearing Voices Groups in Recovery
Results indicate that the hearing voices groups (HVGs) played a significant role in normalising voice-hearing, exposing participants to different interpretations of what it means to hear voices, disseminating strategies for dealing with voices, and providing a sense of community. This is in keeping with other research reporting on the experiences of group members both in the HVNNSW and overseas (Dos Santos, 2014;Ngo Nkouth, St-Onge, & Lepage, 2010;. However, current findings and the research mentioned are based on the experiences of participants who had chosen to attend HVGs and found them helpful. Choosing not to attend an HVG might be part of a turning away response and it is unclear from the current research whether people with this style of response would find attending a group helpful.
In addition, although the flat hierarchy and member-driven agenda of groups are valued by group members (Dos Santos, 2014), this results in significant variation between groups in how they are run (Corstens et al., 2014). It is therefore unclear whether the positive experiences reported by this sample would generalise to all groups and attendees. Finally, the aspects of the HVGs which participants found helpful might also be accessible through peer workers, spiritual groups, structured group therapy or individual therapy. Indeed, some people report that they benefit from the structured nature of group therapy (May et al., 2014). Of course, people might choose to attend both types of groups. The issue is that currently, only limited research (e.g. Dos Santos, 2014; is available on the impact of HVGs, suggesting that further investigation is required to clarify how they influence recovery in people who hear voices, how helpful they are and for whom they are most useful. Some might reject the notion that such forms of evaluation would be helpful. However, the relative paucity of research means that such approaches are not as well-recognised and accepted by clinicians as might otherwise be the case.
More quantitative and qualitative research is required to build a bridge between consumer-and professional-driven approaches, facilitating greater acceptance of HVGs in main-stream services.

Limitations and Directions for Future Research
This research is limited by its retrospective design, because retrospective recall is less accurate than recording events as they occur (e.g. . However, given that narrative research aims to elicit how participants understand their experience, and this involves their subjective recollection of experiences, this is in a sense not problematic. Nonetheless, prospective, longitudinal investigation of recovery style would allow examination of whether there are any people who initially turn away and then turn towards, and if so, what might contribute to this change. It would also be useful to examine whether regular assessment of recovery style and adjustment of treatment style to match patients' needs lead to better treatment outcomes. A second limitation is the nature of the sample, which was small, involved selfselection and was more likely to attract people who had positive experiences of HVGs. Strengths of the study include the rigour employed in generating narratives, which were member-checked by participants, and the focus group's critique of the model of recovery were not able to cope with their voices, with the remaining 150 indicating that they were able to cope . The discovery that such a significant proportion of people were able to function and cope with their voices challenged the notion that experiencing auditory hallucinations was necessarily disabling and associated with mental ill-heath. This motivated Romme, Escher and Hage to organize the first world hearing voices congress. The congress provided an opportunity for voice-hearers to share their experiences and ways of coping with voices. All understandings of voices, whether medical, psychological, spiritual, or trauma-based, were welcomed. Escher (1989, 1993)  Participants who were seen to 'turn away', however, had also recovered according to our objective criteria, in terms of symptoms, quality of life and psychological distress. The focus was on the resolution or management of symptoms and leaving these experiences behind, rather than integration. This finding indicates a different type of recovery which does not involve active engagement, negotiation, acceptance and meaning-making characteristic of the suggested organisation and stabilisation phases. Instead, this typology bears parallels to McGlashan et al.'s (1975) sealing over recovery style, characterized by cordoning off psychotic experiences from the rest of one's life, and Frank's (1995) quest memoir narratives, which involve acceptance of illness with trials told stoically and no special insight gained (France, Hunt, Dow & Wyke, 2013). In considering this narrative, it is critical that these participants are not pathologized when compared to those who 'turned towards.' The latter group had the benefit of HNV groups and professionals who supported them to normalise and respond to their voices.
Turning away participants can perhaps be seen as more remarkable because they did not have access to such discursive resources (Lock, Epston, Maisal & de Faria, 2005). It is also important to note the sense in which withdrawal can be an active means of protecting oneself rather than a passive action, as pointed out by consumers/survivors (e.g. Andresen, Oades & Caputi, 2003).
While taking care not to pathologize either recovery typology, the stories that participants told about themselves were more restrictive in turning away narratives and more liberative for those who turned towards. Indeed, there is evidence that in the long-term, an integrative approach towards psychotic experiences confers better psychological health, predicting remission at one-year follow-up (Staring, van der Gaag & Mulder, 2011). A 'sealing over' style has been associated with poor attachment, depression and low self-esteem (Drayton, Birchwood & Trower, 1998).
This raises the question of how clinical services should respond to people with different recovery styles. There is evidence that people's recovery style can change over time (Tait, Birchwood & Trower, 2004). Consequently, periodic assessment of a person's recovery mode would allow services to match their approach to the person's natural inclination at that time. If someone persists in using a turning away style and this is perceived as limiting recovery, how should services respondacknowledge that a limited recovery has sufficient meaning to the individual, or seek to facilitate the adoption of a turning toward style? Longitudinal studies examining recovery style would allow further clarification of this issue. Assuming that encouraging a turning toward style over the long-term is possible and desirable, identifying mechanisms that support this change would be helpful. For example, Leonard, Jones and Cupitt (2014) found that stronger positive beliefs about worry were associated with an increased tendency to 'seal over'. Addressing beliefs about worry in therapy may facilitate decreased avoidance of difficult emotions and images, encouraging increased emotional processing and therefore recovery.

Role of HVN in Recovery
Five out of six participants who turned towards were recruited from the HVNNSW.
However, before considering the impact of the HVN on recovery, it is worth clarifying its dual role as both a radical social movement focused on human rights and advocacy at a systems level and a user-led organization offering support and therapeutic intervention (Corstens, Longden, McCarthy-Jones, Waddingham & Thomas, 2014). Although related, these will be discussed separately below for the sake of clarity.

Hearing Voices Groups
Results indicate that the HVN groups and approach played a significant role in discussing experiences with other group members, they were exposed to different ideas about how to cope with voices and how voices were experienced by other people. Participation in the group gave them hope and destigmatised the experience. Similarly, dos Santos's (2014) participants reported that they valued the social connections formed in the group as well as the opportunity to receive feedback about themselves. They also described a shift from being too uncomfortable to talk about voice-hearing experiences within the group to gradually gaining confidence within what was experienced as a safe environment. For some, this extended to talking about their experiences outside of the group.
Increased openness on the part of voice-hearers about their voices may contribute to voice-hearing being better understood and less stigmatized in the community generally.
Increased contact between the general public and people with mental health difficulties has been shown to decrease negative attitudes and perceptions of dangerousness (Scheffer, 2003). In addition, psychosocial rather than biological explanations are associated with decreased stigma (Read, Haslam, Sayce & Davies, 2006). Given that in the current study secrecy and fear inhibited recovery, decreased stigma is likely to support recovery. Of course, decisions regarding disclosure should be left to individual choice, particularly given the risk of negative consequences, including discrimination.
An important issue is that currently, only limited research (e.g. dos Santos, 2014; supports the helpfulness of HVN groups. One randomized controlled trial on peer support groups for psychosis generally (not specifically related to hearing voices) indicated a significant positive effect on social network and support (Castelein et al., 2008). Current findings and the research mentioned are based on the experiences of participants who found HVN groups helpful. Choosing not to attend an HVN group may be part of a turning away response and it is unclear from the current study whether attending a group would be helpful for people with this style of response. While participants in our study reported positive experiences of the HVN, however, it is possible that other people have had a different experience. The aspects of the HVN which participants found helpful may also be accessible through peer workers, spiritual groups, structured group therapy or individual therapy. Indeed, some people report that they benefit from the structured nature of group therapy (May et al., 2014). Ruddle et al. (2009) suggested attempts to identify common helpful processes across group type (structured, HVG, CBT) would be more helpful than attempting to determine which type of group is more effective. Thus, further research into HVGs is urgently required.
Some within the survivor movements may reject the notion that such forms of evaluation would be helpful. However, the relative paucity of research means that such approaches are not as well-recognised and accepted by clinicians as might otherwise be the case. Clinicians, working from a scientist-practitioner model, are more likely to feel comfortable recommending therapeutic groups which have been shown to be effective. More quantitative and qualitative research is required to build a bridge between consumer-and professional-driven approaches, facilitating greater acceptance of approaches such as the HVN in main-stream services.
The efficacy of HVGs is not, however, amenable to being researched using traditional research designs, including randomized controlled trials. Participants in dos Santos's (2014) research emphasized that they valued that the group agenda is determined by group members.
They did not wish for it to become a therapy group led by a therapist, which would result in a loss of power and autonomy. However, while the flat hierarchy and member-driven agenda of groups are valued by group members (dos Santos, 2014), this means that they are not standardised. Standardized delivery of a treatment via session by session manuals which specify content is required in randomized controlled trials. While the essential elements of an HVG group could be identified, it is unclear whether such distillation would capture the true nature and processes of a naturally occurring HVG (Corstens et al., 2014). Another issue is how to define effectivenessin terms of symptom reduction associated with clinical recovery, or with personal recovery outcomes including subjective well-being, the nature of people's relationship(s) with their voice(s), sense of self and ability to function and live a valued life (Corstens et al., 2014).
Inclusion of voice-hearers in the research process would help to address this issue. Collaboration between the HVM and researchers is required to consider which research questions should be asked as a priority and which research methods would be most appropriate to employ. This is discussed further in the 'story-telling rights in research' section.

Hearing Voices Movement
All participants recruited from the HVNNSW had attended a group at least once. Some did so regularly and three were actively involved in advocacy and peer work. Thus, the degree of participants' political involvement and alignment with the hearing voices movement (HVM) varied. However, they all tended to take a human rights stance and offered critiques of the mental health system. They viewed themselves through a political lens and questioned the lack of power they and other service users had in the mental health system. This is in keeping with the key concerns of the HVM (Corstens et al., 2014). Participants' political stance, values regarding issues of power and rights, and advocacy of others was deeply meaningful to them and formed part of their identity. In contrast, in the context of a traditional service framework, five out of six people received diagnoses. This formed part of how professionals, who held an expert position in relation to participants' experiences at the time, interpreted the experience of distressing voices. Therefore, they also influenced how participants initially understood their experiences, with implications for their sense of self and identity. In most cases, treatment involved being given a diagnosis and prescribed medication. The potentially harmful effects of diagnosis, for example, encouraging a reductionistic approach and contributing to stigma, continue to be debated (Van Os, 2010). Some argue that the diagnostic system should be changed to incorporate categorical and dimensional classification, which is in keeping with evidence from epidemiological research, and to emphasise that it is a variation in normal human experience (Van Os, 2010). Others have focused on how to minimize the negative impact of communicating a psychotic-spectrum disorder diagnosis (Outram et al., 2014).
Certainly in the current study, while diagnosis brought short-term relief for some participants, recovery generally involved coming to understand voice-hearing as a meaningful response to life events. Indeed, most turning towards participants described their treatment within a traditional medical model to be insufficient. They felt that it did not provide an adequate explanation for their experiences in context of their life history and current situation.
One participant commented that not being diagnosed, even though she may have met criteria for diagnosis, meant that her experiences were not pathologised and that this was essential to her becoming well. Indeed, the assumption that effective treatment can only be provided in context of diagnosis bears revisiting. In addition to potentially causing stigma and internalised stigma, there are significant concerns regarding the validity of the schizophrenia and schizoaffective disorder diagnoses (e.g. Bentall, 2003;Heckers, 2009). Further, as pointed out by Johnstone (2008), it is feasible to provide formulation-based treatment and avoid diagnosis altogether.
Approaching distressing voices as potentially meaningful in the context of one's life history, which is by definition provided in a clinical formulation, is certainly what turning toward participants in the current study wished to receive during treatment.
Participants' shift away from a purely medical model explanation and toward the HVM was also therefore an empowering shift toward reclaiming their right to define their own experiences and critique the treatment which with they were provided. This stance represents a shift from a passive patient role, wherein the person feels they have little valid knowledge, to one of expert by experience, which involves viewing knowledge gained by experience as valid and valuable. This is in keeping with  participants' description that rethinking diagnosis meant relegating their problems to the status of being a part of themselves rather than defining who they were as human beings: the difference between "I am schizophrenic" and "I have schizophrenia." Similarly, as participants recovered, they were more active in decision-making regarding their treatment. They emphasised the value of psychiatrists who were willing to work collaboratively with them and, in some cases, agree to changes in medication (e.g. lowered dosage) which were not their preference.
There is concern that these key aspects of the recovery-orientated paradigm have been mistranslatedeven colonised or hijackedfrom their consumer/survivor and recovery movement roots to formal mental health policies and practice (Perkins & Slade, 2012;Slade, 2014). However, while integrating evidence-based and recovery-orientated services remains challenging, it is not impossible. For example, Frese, Stanley, Kress and Vogel-Scibilia (2001) argue that greater autonomy should be afforded to people in mental health services as they recover. Similarly, whichever approach is adopted regarding diagnosis and other relevant clinical practices, it is essential from a consumer rights and advocacy perspective that consumers are included in the debate and that their contributions are given weight and value in such processes. This will be further discussed in the 'story-telling rights' in therapy and research sections below. The notion of 'story-telling rights' is adopted from narrative therapy (e.g. Dulwich Centre, 2014).

Story-telling rights in clinical services: Implications for practice
Results suggest that people have their own resources and expertise regarding what they need to recover and when they need it. Readiness for change emerged as a critical factor in recovery. It is not necessarily something which is likely to occur on the therapists' or hospital's schedule. This is at odds with the demands on funding-pressured services with targets to meet in terms of diminishing length of stay and highlights the importance of out-patient services. In combination with the observed difference in recovery styles, results suggest that a collaborative approach offering different treatments available and encouraging patients to, at the very least, have input into which services they would like to use and when would be advisable. The concept of 'personal budgets'which allow consumers to choose which services they would like to access -is an example of an approach which recognises patients' expertise regarding what they need (Perkins & Slade, 2013). However, there is some discomfort with this approach among professionals who argue that they, rather than patients, are in the best position to determine optimal treatment (Perkins & Slade, 2013). This position is based on a scientist-practitioner orientation, whereby knowledge of the research literature combined with clinical assessment and formulation skills are used to inform treatment (Jones & Mehr, 2007). Although from a clinical psychology perspective formulation is by definition collaborative, there is still some acknowledgement that the clinician has access to expert knowledge which gives them the right to recommend and in some cases subtly determine what treatment will be undertaken. An important aspect of the scientist-practitioner model is that the evidence-base for a particular treatment is derived from aggregate outcome data rather than individual cases. In contrast, a recovery-oriented way of working would give equal weight to the individual's values and goals (Perkins & Slade, 2013). Indeed, people who have recovered from severe mental health difficulties report that taking responsibility for recovery is an essential part of the process (Andresen et al., 2003;Anthony, 1993). In keeping with this, a longitudinal, prospective study found that an internal locus of control was positively associated with recovery and negatively associated with depression and psychosis (Harrow et al., 2009).
The question from a clinical perspective is how to support this sense of agency and internality in people distressed by voices while also meeting one's legal responsibilities. This 'dual-role' affects mental health workers, with considerable tension created between institutional demands and the desire to work collaboratively (Borchers, Seikkula & Lehtinen, 2014). The NSW Consumer Advisory Group (2009) identified the following barriers to recovery: consumers becoming passive in relation to their treatment, service providers taking the expert role and tension between clinical and personal views of recovery. However, they also acknowledged the difficulty in following recovery-oriented philosophy, which emphasises increased agency and choice, when faced with a situation where a service user poses a risk to themselves or others.
It is worth bearing in mind that no treatmentwhether it is medical, psychological or socialworks for everybody or to the same extent (Thomas et al., 2014). The idiosyncratic nature of recovery across participants, differences in recovery style and importance of timing in the current study emphasises differences between people in their recovery and treatment needs.  (Seikkula et al., 2003). It has resulted in excellent outcomes, with 81% of first episode participants reporting no residual psychotic symptoms and 83% having returned to work at two-year follow-up (Seikkula et al., 2003).
The issue may be more about how, as much as what, is done. Indeed, how treatment is organised and how decisions about the best course of action are made appear to have a very significant impact. Traditionally, professionals would assess the person in distress and create a treatment plan, which is then communicated to the person and possibly their family. In contrast, the Open Dialogue approach mentioned above involves the whole team meeting together with the person and their family from the first point of contact. They collaboratively assess the situation, come to a shared understanding thereof and then decide what to do. As pointed out by , traditional forms of treatment delivery trap people in a monological mode, which keeps them stuck in the problem. In contrast, encouraging dialogue, which is done drawing on the principles of systemic and family therapies, encourages the mobilisation of the family's resources to deal with the problem. Open Dialogue also focuses on understanding the person's distress within their life context and situates the problem as separate to the person, as in narrative therapy. This avoids the profoundly harmful effects of diagnosis as discussed above. In contrast, paternalistic services obscure these resources and inhibit recovery by encouraging people to remain in a passive role regarding their recovery. Further, while the impact of life experiences and society generally may be considered, the problem is conceptualised as being within the person.

Story-telling rights in research
The question of who holds the expert position or has access to expert knowledge in  (Corstens et al., 2014). Given the constitutive power of language in defining how we understand and interpret the world, it is perhaps unsurprising that it emerged as a critical topic at the World Hearing Voices Congress (2013). Indeed, the NSW Consumer Advisory Group (NSWCAG, 2009) review specifically recommended a shift away from deficit-based language and toward a strengths-based mode of practice.
More broadly, in order for voice-hearers to have story-telling rights in research, they need to be included in the broader research agenda; in other words, they need to have some power in determining which research questions are asked and how research is conducted (Corstens et al., 2014). Historically, psychiatric and psychological research into voice-hearing has typically not included voice-hearers or service users in the research agenda. This issue is complicated by the reluctance of some voice-hearers to be involved in research given its questionable approach historically. Voice-hearers and their loved ones are in business terms 'key stakeholders' in the research process and should therefore be involved in research (Griffiths, Jorm, Christensen, Medway & Dear, 2002). Their inclusion in research makes it more likely that the research will result in a practical benefit to the group it is intended to benefit. For example, research which aims to identify which medication will be effective and less harmful (i.e. have fewer side-effects) for a particular individual has a clear practical benefit for people who hear voices (e.g. Adkins et al., 2013). The same applies to research by Tsuboi et al. (2013) concerning what range of dosage of antipsychotic medication is required for therapeutic effect while avoiding changes in the dopamine system associated with harmful metabolic changes. Consumers and carers are generally in agreement with other stakeholders regarding the importance of research into psychological and social interventions, as well as prevention of mental health difficulties (e.g. Griffiths et al., 2002). Given that various stakeholder groups may rate research priorities differently, it is important that independent consumer and carer reviews be undertaken. For example, the NSWCAG review (2009)  knowledge not only in what they know, but how they know it. Their inclusion would be a step in the right direction in terms of shifting power toward a disempowered group (Kruger, 2000). It is important that sufficient training in research design and methods is provided if this involvement is to be meaningful rather than tokenistic (Corstens et al., 2014). The concern within a Participatory Action Research (PAR) paradigm on effecting change through research is in keeping with an agenda of empowerment and ensuring benefits to the group being studied (e.g. Khan, Bawani & Aziz, 2013). A broader issue, of course, is that regarding funding, potential conflicts of interest and public availability of research results. The Clinical Antipsychotic Trial of Intervention Effectiveness (CATIE) trial is an example of an attempt to address these issues (Lieberman et al., 2005;Stroup et al., 2003).
The benefits of taking a participatory research action stance are apparent in Neil et al.'s (2013) study. The aim of the research was to create and trial a measure of recoverythe Questionnaire about the Process of Recovery from Psychosis (QPR)which was in keeping with voice-hearers' definitions and was easy to administer and score. The value of service users' involvement at all stages of the research process was apparent. They had direct input into the wording of items and suggested new items to be included. Further, speaking from the perspective of someone who might be asked to complete the measure, they queried whether filling it in might elicit distress. After discussing this issue with the research team, it was decided to include a visual analogue scale of distress in the trial. Research team members with lived experience also disseminated findings in service user forums which would have been inaccessible to non-service user researchers.

Reflections on the research process
In the context of the current research, there was some tension between traditional clinical psychology and narrative or consumer-driven paradigms when designing, implementing and interpreting this research. The diagnostic interview, for example, was included to address concerns of the research coordinator at the University of Sydney that without confirmation of a diagnosis, it was unclear what participants were recovering from, as hearing distressing voices does not in and of itself constitute a psychological disorder. Inclusion of diagnostic information was intended to address this concern for those readers whose work is informed by this paradigm.
However, it is also worth noting that the utility of diagnosis is constantly under debate and some have questioned whether it does more harm than good (e.g. Breeding, 2008).
The current research privileged the knowledge of people with lived experience of recovery from hearing distressing voices. It was designed in collaboration with people who hear voices. Although a power differential will perhaps always exist between researchers and participants, narrative research represents a collaborative endeavour. As part of the research process, participants told their preferred story, which was heard and witnessed by the researcher.
Member-checking and co-construction of narrative recognises researcher and participant as equals in meaning-making and in the research process. Narrative method is by its nature in keeping with recognising the expertise of the interviewee. However, given additional time and resources, there are some improvements that could have been made to the method. In this case, the primary researcher collected data and conducted a preliminary analysis. They were the only person in the research team who had been immersed in the data to that extent. The focus group, including people who were experts by experience and profession, was helpful in terms of considering alternative perspectives on the data. However, consulting with people with lived experience at various points of the research process does not confer the same benefit as their active involvement throughout would. It is possible that other interpretations of the data would have emerged had someone with lived experience been immersed in the data to the same extent that the primary researcher was. This may not always be practical; however, given the demonstrated benefits of this approach, it would be worth considering in future research designs.

Recovery trajectories
Recovery trajectories were non-linear, cyclical and involved periods of rest and consolidation. This pattern is in keeping with narrative research into recovery from childhood sexual abuse, wherein participants emphasised that recovery was not a linear progression and  Participants' understanding that general mental health and disconnection from others preceded and maintained distressing voices is in keeping with existing research. For example, a lengthy (24+ month) period of mood, anxiety and sleep difficulties as well as a decrease in social functioning is common prior to the onset of psychosis (Birchwood, Spencer & McGovern, 2000;Häfner, 2000, Rietdijk et al., 2009Yung & Jackson, 2004) and these difficulties are frequently co-morbid with psychotic symptoms (Morrison, 2009). An increased degree of depression and lower self-esteem are associated with greater voice severity and negative content (Smith et al., 2006). From a clinical perspective, the rationale for targeting anxiety, mood, sleep, stress difficulties and social difficulties in treatment are two-fold: firstly, they contribute to lowered quality of life and higher distress, and secondly, they appear to exacerbate or act as a trigger for more distressing voices (Eisner, Drake & Barrowclough, 2013;Krabbendam et al., 2005;Morrison, 2009;Soehner, Kaplan & Harvey, 2013). Indeed, learning about how to nourish general mental health was a strong theme across typologies and was understood to be essential to recovery.

Inhibitors of Recovery: Loss & Stigma
Participants described a sense of loss and grief regarding identity and future due to onset of voice-hearing. This is a common experience among people who hear distressing voices (Mauritz & von Meijl, 2009). Voice-hearing is often an overwhelming experience in and of itself (Lieberman & Fenton, 2000). In addition to this, participants were fearful of disclosing their voice-hearing experiences to others, particularly mental health professionals, due to concerns about stigma and involuntary or invasive treatment. Secrecy about voices due to fear of alarming loved ones or concern that others would respond negatively was also reported by dos Santos's (2014) and Pyle and Morrison's (2013) participants. Indeed, secrecy is an adaptive response when one possesses a characteristic which is understood by the community to be shameful (Byrne, 2000). In keeping with existing research, stigma was a barrier to accessing treatment, forming social relationships and gaining employment, negatively impacting participants' sense of self (Corrigan, 2004;Franz, Carter, Leiner, Bergner, Thompson & Compton, 2010;McCann, Lubman & Clark, 2011). Thus, secrecy and stigma inhibited recovery mainly through inhibiting processes that supported recovery, keeping participants isolated and unexposed to non-stigmatising views of the self.

Initial Responses & Becoming Overwhelmed
Given this context, it is perhaps unsurprising that participants typically initially responded to voices by attempting to push them away. This is in keeping with the 'startling' phase of recovery described by , which involves attempts to distance oneself from voices and avoid them. In addition, research indicates that relating to voices from a position of distance is associated with distress (Vaughn & Fowler, 2004). Similarly, Escher, Delespaul, Romme and van Os (2003) found that adolescents who responded defensively toward their voices, were more likely to become depressed. Participants described low self-esteem, low mood and a sense of being overwhelmed and powerless to change their situation for the better.
Some acted upon voice commands in ways which were not in keeping with their values, exacerbating their sense of despair. A sense of entrapment is common among those struggling with their voices and is associated with depressed mood (Birchwood, Spencer & McGovern, 2000). At this stage, attempts to cope with distressing voices had been largely unsuccessful.
The tone of participants' narratives was one of defeat and powerlessness, with a sense that their lives were progressing in accordance with factors outside of their control. Understandably, beliefs that voices are powerful and omnipotent is associated with depressed mood (Chadwick & Birchwood, 1994). In addition, an external locus of control is associated with depression in people diagnosed with schizophrenia (Harrow, Hansford & Astrachan-Fletcher, 2009).

Despair / Exhaustion Leading to Change
Participants described tolerating a state of despair for months and even years before

The Small and the Ordinary: Recovering from an Illness versus Recovering a Life
Despite significant differences between typologies, many recovery processes were common across typologies. They were centred around meeting participants' needs for connectedness, positive identity, agency, opportunities to contribute to the community and participate in meaningful activities (Andresen, Oades & Caputi, 2003;Reis, Sheldon, Gable, Roscoe & Ryan, 2000;. These 'small and ordinary' things were essential to rebuilding a life beyond simply eliminating symptoms (Perkins & Slade, 2012;Weingarten, 1998, p4 & p7). They formed the foundation for further progress and recovery processes. For example, for participants who engaged in challenging their voices, it was only possible for them to do so once they had regained or started to regain a positive sense of self and connection to the community. Being able to participate in the broader community, be it through paid, unpaid or voluntary work, meant that participants felt connected to and valued by society. It also provided an antithesis to the 'passive patient' role (Kelly, Lamount & Brunero, 2010;Tew et al., 2012).
Paid work also addressed an important barrier to social inclusion through providing an income (van Niekerk, 2009). Flexible work arrangements or casual/contract work, anti-discrimination workplace laws, vocational services, financial support and access to safe housing were essential to participants' recovery trajectories. Some participants commented that without access to contract work or protection under anti-discrimination laws, their recovery would have been undermined and their eventual outcomes much poorer.
In keeping with these findings, occupational identity has emerged in qualitative research as an essential aspect of the self, underpinning the ability to live with psychotic experiences (Makdisi et al., 2013). Supported employment has been shown to lead to positive outcomes for people with mental health difficulties including higher pay, increased number of days worked and shorter time to securing paid employment (Marshall et al., 2014). Likewise, in terms of access to housing, Kirkpatrick and Bryne's (2009) participants reported that having stable accommodation allowed them to exit the continuous circuit of movement which characterised their lives while homeless. Increased certainty and control regarding this basic need for shelter meant that they were able to reconnect with others. Importantly, they maintained the ability to choose when and with whom to connect. Their accommodation also made engagement in employment and planning for the future feasible.
Poverty is a significant predictor of who will develop psychosis and is associated with increased psychological distress more generally . Higher rates of schizophrenia have been found in countries with greater income disparity (Burns, Tomita & Kopadia, 2014) and there is evidence that greater relative poverty rather than poverty per se is associated with poorer mental health, including psychosis . Whether social policies achieve in practice what they set out to in theory is debatable, with the NSWCAG identifying a significant gap between policy and outcomes (2009). While social inclusion has become the focus of social policies regarding mental health in many countries, there is some debate about how to measure it and thereby provide a baseline from which to evaluate progress (Baumgartner & Burns, 2013). What is clear, however, is that when recovery is considered through a social rather than purely individual lens, a human rights framework is much more commonly evoked and brought into focus. This has partly occurred due to the advocacy of consumer groups and highlights the broader socio-political context within which treatment is provided (Sayce, 2001). Although a broader social perspective and human rights issues are rarely discussed in-depth in the training of most psychologists and psychiatrists,

Meaningful Activities
Participants reported that engagement in activities that held meaning for them was essential to their recovery. This is in line with findings that it is associated with higher life satisfaction in people with mental health difficulties (Goldberg, Brintnell & Goldberg, 2002) as well as lower depressive affect and higher self-esteem in a community sample of unemployed persons (Waters & Moore, 2002). Paid employment, volunteer work, social and solitary activities help to meet the need of human beings to feel competent, autonomous and related to others (Reis, Sheldon, Gable, Roscoe & Ryan, 2000;van Niekerk, 2009). In the current study, three participants were actively engaged in providing mental health services in the context of paid or voluntary work, both within the HVN and externally. Helping others with mental health difficulties is associated with better mental health, less hopelessness and an increased sense of meaning in life (Schwartz, Meisenhelder, Yunsheng & Reed, 2003). It is also the antithesis to the 'passive patient' role. Participants engaged in paid employment within the mental health system were actively engaged in attempts to improve it. They were more likely to talk about mental health issues within a human rights framework (e.g. Perkins & Slade, 2013). Those paid as peer support advocates commented that receiving remuneration for their work indicated to them that they were more than 'token' consumer team-members and that their contributions were valued.
Other participants emphasised the structure and routine which regular activities provided and the enjoyment they derived from their activities. This is Supported employment is an effective intervention for people with mental health difficulties (Marshall et al., 2014) and engagement in paid employment is associated with higher self-esteem and quality of life (van Dongen, 1996). Qualitative research indicates that work, despite its challenges, can support well-being for people with mental health difficulties, particularly through providing a sense of purpose, evidence that one is able to contribute to society, forming a positive identity and providing a distraction from symptoms (van Dongen, 1996;van Niekerk, 2009). Indeed, purpose in life is a strong predictor of self-efficacy in people diagnosed with a major mental health disorder (Scott, 2007). It is of note that despite evidence of the positive impact of supported employment programs, several barriers to their implementation have been observed and availability of these services remains an issue (Marshall et al., 2014).

Voices, Self & Others
Many people experience an initial sense of their identity being overwhelmed and the current study, participants described being less distressed by critical or threatening voices when they had developed a stronger sense of self. This is in keeping with research demonstrating that a reduced capacity to reassure oneself after self-critical thoughts is associated with shaming voice content, while people who report feelings of self-hatred and inadequacy are more likely to experience their voices as controlling and critical (Connor & Birchwood, 2013).
Recognition of the importance of people's self-esteem in mediating their affective response to negative voices has lead to this becoming a specific target of psychological interventions. For example, van der Gaag, Oosterhout, Daalman, Sommer and Korrelboom (2011) trialled the use of imagery techniques targeted at strengthening positive memories and perceptions of the self.
Results were promising, with significant improvements in depression in the treatment condition compared to treatment as usual, mediated by improvements in self-esteem and acceptance of voices. Similarly, participants who undertook person-based cognitive therapy reported that acceptance of voices and developing a positive sense of self were pivotal to their progress (Goodliffe, Hayward, Brown, Turton & Dannahy, 2010;May, Strauss, Coyle & Hayward, 2014).
Re-building one's sense of self, including developing increased self-esteem, self-efficacy, agency and non-pathologised identity, has been implicated in recovery from domestic violence (Oke, 2008) alcohol abuse (Paris et al., 2001) child sexual abuse (Anderson & Hiersteiner, 2007;) and severe or long-term mental health difficulties including psychosis , Song & Shih., 2009, Thornhill, Claire & May, 2004. Redefining experienceswhether they be of abuse or psychosiswere essential to forming a positive, non-pathologised self.
Reclaiming one's sense of self, however, is a social as well as individual process: the antithesis to personal diminishment which can result from stigma .
As participants reconnected with others and developed supportive relationships, they were more able to deal with their voices in an adaptive manner. This is in keeping with research demonstrating significant parallels between how people relate to social others and how they relate and respond to their voices (Chin, Hayward & Drinnan, 2009;Paulik, 2011). It is plausible that as people accumulated more positive experiences with others, their beliefs about others and perception of relative social rank changed, positively impacting upon their style of relating to their voices. The tendency of people with turning toward narratives to engage with their voices rather than attempt to distance themselves is consistent with research indicating that relating to voices from a position of distance is associated with significantly higher distress (Vaughn & Fowler, 2004;. Engaging with voices and setting appropriate boundaries has been implicated in developing a positive relationship with voices .

Discursive Resources and Making Sense of Voices
The narratives people tell define how they understand themselves and their problems (Lock et al., 2005;Weingarten, 1998). In the context of hearing voices, the dominant narrative, based in Foucauldian terms on 'global' (privileged) rather than 'local' knowledge, is that it is a symptom of a disease best treated by anti-psychotic medication (Madigan, 1992;. Implicitly, this narrative promotes the notion of the person as the problem (Ben-Zeev, Young & Corrigan, 2010; Madigan, 1992;van Os, 2010). Those who 'turned towards' in this study also 'turned away' from the dominant medical model discourse. This act of resistance was conducted in solidarity with others in HVN groups, allowing for the 'thickening' of alternative stories (White & Epston, 1990). For example, participants described voices as a manifestation of distress understandable within their life context (e.g. Beavan & Read, 2007;. Although some participants still used aspects of a medical model explanation and vocabulary, they emphasised that voice-hearing is a normal variation in human experience, opening up the possibility of a non-pathologized self. This is in keeping with the Hearing Voices Movement (HVM) approach and with findings that a stronger belief that psychotic experiences occur on a continuum is associated with lower stigma (Wiesjahn, Brabban, Jung, Gebauer & Lincoln, 2014). Findings suggest that exposure to alternative understandings of voice-hearing, which might compliment rather than necessarily replace the dominant medical model narrative, may be helpful.

Effective Medication
Participants with a dominant turning away narrative emphasised the importance of effective medication becoming available, in helping them to survive, think clearly, function and eventually recover. Medication was essential in enabling participants to meet their basic needs.
For example, medication was understood by participants to keep them calmer and enable them to manage difficult emotions better, to communicate and build relationships with others, and to engage in activities, including paid work. The benefits of medication for some voice-hearers in terms of decreasing the severity of distressing voices and the intense negative emotions accompanying them has been documented elsewhere (Moncrieff, 2009;NICE, 2009). It is important to note, however, that three out of five turning away participants still experienced hearing voices. However, they described medication as enabling them to disengage their attention from voices when they wished to do so and decreasing the intensity of their emotional response. The benefits of medication appeared to be intertwined with other recovery processes.
For example, participants used mindfulness skills in combination with medication to deal with distressing voices. In addition, they did so in context of generally improved mental health, which in turn was contributed to by medication. Thus medication formed an essential part of recovery in combination with other processes. The importance of factors above and beyond a decrease in clinical symptoms has been demonstrated in the CATIE findings. While emerging as significant predictors of subjective quality of life (SQOL), depressive, positive and negative symptoms collectively explained only 20% of variance in SQOL (Fervaha, Agid, Takeuchi, Foussias & Remington, 2013). In turning away narratives, there was less motivation to seek alternative meanings of voice-hearing. Once symptoms had resolved or became manageable, there was little or no perceived need to reflect any further on the meaning of the experience.
While some participants were uncomfortable with a medical model explanation, they had difficulty generating or making visible alternative interpretations.
In contrast, those with turning toward narratives either found medication to be ineffective in decreasing symptoms, or found it somewhat helpful but associated with significant sideeffects which outweighed its benefits. Potential side-effects of antipsychotic medications are well-documented and include weight-gain and increased lipids, glucose and heart-rate However, there is also the idea that medication should be given sparingly, at low doses and for a limited time only (Corstens et al., 2014). Indeed, from the HVM perspective, given an understanding of voices as messengers about an important emotional issue, eliminating them entirely by use of medication is viewed as potentially counter-productive. Similarly, occasions when in medical terms people's 'symptoms' increase, this can be and often is viewed as a learning opportunity within the HVM framework. This highlights some essential differences and possible points of tension between the two approaches and the need for increased communication among proponents of both views.
The heterogeneity in response to antipsychotic medication is well-documented (e.g. Adkins et al., 2013). It is therefore not surprising that there was a great deal of variability in effectiveness reported by participants in the current study. A relatively new and promising development is the trend in research toward attempts to identify which medication is likely to be effective for a particular individual, or alternatively which individuals are unlikely to respond to any medication (e.g. Ramsey, Liu, Massey & Brennan, 2013).

Parallels in Professional-Driven Treatments
Across typologies, participants described mindfulness skills as an effective way of dealing with distressing voices. Given findings that a resistant style of relating to voices is associated with increased distress, intentionally noticing voices without being drawn in and caught up with them holds promise (Farhall, Greenwood & Jackson, 2007;Thomas et al., 2014).
There is evidence that responding mindfully to distressing voices is associated with decreased Coyle & Hayward, 2014). In the current study, participants described the effect of mindfulness skills as providing some space and distance from voices. Participants described being more able to disengage from voices or not take their comments to heart as much as they did before learning how to mindfully notice voices as transient events. This is very much in keeping with the 'decentred' relationship with distressing emotions and events which is aimed for in ACT and mindfulness-based therapies.
Despite the impact of consumer-driven HVN groups in this study it is also important to recognise that promising professional-driven treatments are coming to similar conclusions As a cautionary note, findings of the current study suggest that timing is critical. Turning toward participants reached a tipping point where doing something differently, although still frightening, was evaluated as a better option than continuing to tolerate their situation. This is similar to the crisis-induced change noted by Milligan et al. (2013). In addition, those engaged in a turning away style of response may find such treatments unhelpful and require some time to be ready to undertake such work. These caveats aside, the critical role of interpersonal dialogue and solidarity is also becoming recognised in the network therapy of Seikkulla and colleagues (e.g. . This approach, involves intensive engagement with the client, the entire family and any other loved ones as well as professional stakeholders involved. Treatment is based on the notion that psychotic reactions are prenarrative or metaphorical; attempts to make sense of experiences that are so difficult that they have not yet been situated in spoken discourse . Despite the evidence for the efficacy of psychological treatments, however, several barriers to accessing them remain, resulting in significant unmet need (Farhall & Thomas, 2013;Mojtabai et al., 2009;National Collaborating Centre for Mental Health, 2009). For example, up to 40% of people diagnosed with schizophrenia reported receiving no psychological intervention over a six to twelve month period (Mojtabai et al., 2009). Barriers to implementation of evidence-based therapies for psychosis include organisational factors such as managerial focus on acute needs, and therapist factors including lack of specific training and supervision in psychological interventions for psychosis (Mojtabai et al., 2009;Farhall & Thomas, 2013). It is important to emphasise that not one single participant in the current study reported receiving psychological intervention that was specifically aimed at dealing with voices. Those who engaged in behavioural experiments and other CBT strategies did so without the help of a psychologist.

Limitations and Directions for Future Research
This study is limited by its retrospective design, as retrospective recall is less accurate than recording events as they occur (e.g. Yoshihama & Gillespie, 2002). However, given that narrative research aims to elicit how participants understand their experience, and this involves their subjective recollection of experiences, this is in a sense not problematic. Nonetheless, prospective, longitudinal investigation of recovery style would allow examination of whether there are any people who initially turn away and then turn towards, and if so, what might contribute to this change. It would also be useful to examine whether regular assessment of recovery style and adjustment of treatment style to match patients' needs leads to better treatment outcomes. A second limitation is the nature of the sample, which was small, involved self-selection into the study and was more likely to attract people who had had positive experiences of HVN groups.
Strengths of the study include the rigour employed in generating narratives, which were member-checked by participants, and the focus group's critique of the model of recovery Psychotherapy Research,, 583-590. Fischer, D. (2003). People are more important than pills in recovery from mental disorder.    Three female voices. The first was friendly and alleviated the hearer's sense of isolation. The second appeared later and was more interested in intellectual conversations. The third was more childlike and mostly made comments related to bodily needs, for example, encouraging the hearer to eat or go to sleep when needed. A group of male voices appeared during a time of increased stress and life transition. They increased the hearer's anxiety and were associated with a strong but vague sense of threat.

Yes 11
Male and female voices, insulting, bossy and threatening at times and agreeable at others. Amplified hearer's anxieties, for example worries about life choices such as which degree to pursue at university.

APPENDIX G: Additional information regarding the development of the model of recovery over time
The preliminary model of recovery was developed by the candidate through reading, rereading and reflecting upon narratives generated from interviews. Common threads in recovery processes were identified and included in the model. Following this, the model was reconsidered in terms of whether it accurately represented each individual narrative. The model was also discussed with primary supervisor, Dr. Paul Rhodes and associate supervisor, Dr. Mark Hayward and changes made accordingly. The resulting preliminary model is presented in Figure 3. The notion of 'protective hibernation' was chosen after careful reflection to indicate an active withdrawal in order to survive rather than 'giving up'. The notion of active withdrawal has been commented upon in literature by consumers and other authors (see Andresen et al., 2003). The preliminary model of recovery over time was presented at the World Hearing Voices Congress, 2013, held in Melbourne, Australia.
Once all narratives had been member-checked, they were sent to the four members of the focus group, who were all experts by profession, research, clinical experience, lived experience, or a mixture of the above. The group included Dr. Vanessa Beavan, Debra Lampshire, Dr.
Simon McCarthy-Jones and Dr. Neil Thomas. They were also sent the preliminary model of recovery and asked to comment upon it in relation to the narratives. The aim was to change or refine the model to more accurately and authentically represent participants' narratives. The candidate also met separately with Dr. Vanessa Beavan to discuss this process. This meeting resulted in a clarification of the aim of the focus group. Group members were unfamiliar with narrative method and thus there was a need to explain it and idea of typologies or types of recovery story. The notion of 'the boxer' and 'the rock-climber' as narrative typologies were briefly discussed. However, they were ultimately not adopted. Group members did not reside in the same area, being spread across Sydney, Melbourne and Auckland. Therefore, meeting faceto-face was not logistically feasible. An appointment was made with group members to discuss the model via Skype. Feedback from the meeting was very useful. The most important point made was that the model made recovery appear far too neat and linear. It did not therefore authentically represent the lived experience of participants and others with lived experience of recovery. This raised concerns that the model could be applied in a rigid manner in clinical settings, with people distressed by their voices and experiencing mental health difficulties expected by mental health staff to follow a linear process of recovery in a prescriptive manner.
Group members emphasised the messy, non-linear, cyclical nature of recovery. It was agreed that this should be emphasised when presenting the model. It was also noted that inhibitors of recovery such as stigma, isolation and fear of disclosure of voice-hearing experiences were not included in the model. Group members also noted that there were several processes implicated in recovery which were common across typologies related to meeting basic human needs. These included connection with others, coming to hold a positive sense of self and engagement in meaningful activities. It was also noted that isolation and poor general mental health formed the context in which voices appeared to commence. One group member felt that stressors preceding voice onset was not a factor which was clear in the narratives. This was checked by the candidate following the group meeting. Several other points were made. Following the meeting, changes were made to the model incorporating the main points raised by the focus group. The revised model is presented in Figure 4.
The candidate met with Dr. Beavan to discuss the revised model. The model had become almost too detailed and therefore unclear. Changes were made to simplify unnecessarily detailed aspects of the model. For example, stigma, disconnection from others and fear of disclosure were collapsed into one category, while poor general mental health was included as a contextual factor. It was also decided that proximal and distal stressors should be included in the model as this more authentically represented participants' experiences. It was also noted that the two recovery typologies should be made to stand out more in the model, as this was the main concept to be communicated. Further iterations of the model were discussed with the candidate's primary and associate supervisors. Finally, through discussions with Dr. Mark Hayward, it appeared that the concept of 'empowerment/leaning in' and 'protective hibernation/batten down the hatches' narratives was similar to the notion of 'turning toward' and 'turning away' developed in his and colleagues' research. It was agreed that in order to make the research most useful clinically it would be preferable to build some consistency in the terms used in research rather than opting for idiosyncratic terminology which may dilute the overall impact of these ideas. The tension between remaining true to one's participants and results on the one hand and making research of most benefit to clinicians and voice-hearers was noted. Ultimately, as these terms reflected very similar if not identical concepts, it was decided that the 'turning toward' and 'turning away' terminology would be adopted. The final model of recovery developed is presented in Figure 5.

"I was at the shopping centre and I had an out-of-body experience where I came out of my body and I was watching myself and I walked down to the rail-way line and started lying on the railway line -I couldn't stop myself, I was just doing it and I was watching myself do it."
Fortunately, David came to and got up off the tracks before a train arrived: "That's the scariest thing that's ever happened to me by far, because if there was a train coming, I'd be dead." David had had no suicidal thoughts and did not have a history of depression.
"It was like it wasn't me. I've never been depressed or anything." Afterwards, he was worried that he might do something similar again and be seriously injured or killed. David had had no suicidal thoughts and no history of depression: He went straight to the emergency department of the nearest hospital and asked to be committed for a little while. The staff agreed to admit him to hospital for a week or so. That turned into just over a month. His black-outs stopped while he was in hospital:

"It stopped happening while I was in hospital, I wasn't stressed out anymore, I didn't really have anything to do so I really wasn't too worried."
It was in this context that David first heard voices. At first, he thought he was just over-hearing people talking in the ward. He heard two male voices talking to each other about him, criticising him: "They'd say that I was lazy and that I'm worthless and things like that. They'd talk to each other, like one guy would go "isn't he lazy, he's just sitting there watching television." And the other would go "yeah, that's pathetic." And then they'd just talk back and forth to each other." At first, it was sporadic. He would hear voices positioned just behind his right shoulder and when he turned around, no one was there. Then it started happening more frequently. David didn't tell the hospital staff because he wanted to get out of hospital and was worried that if he talked to them about his voices that he wouldn't be allowed to go home. He was discharged. When he went back for his follow-up assessment, the voices were still there and he decided to tell medical staff about them: "I told them what was going on, I said there's two guys and they talk to each other. It wasn't happening all the time at that stage, maybe happen once every couple of days for five minutes at a time or something along those lines." They did a number of tests to try to ascertain what was causing his experiences. David was assessed by a psychiatrist, who told him he was experiencing the early stages of schizophrenia, which was a degenerative brain disease. The first thing David did was read everything he could find about it. Everything he found was very negative: "Everything I read was like 'ah, the drugs will take it away' it wasn't, like everything was pretty negative, it was a brain disease that you got and so I got pretty depressed after that. And of course I wasn't able to do the stuff I used to be able to do. I got really depressed." The explanation of his experiences provided by the psychiatrist and in text-books about schizophrenia was not helpful to David. The idea that he had a degenerative brain disease made extremely anxious about what might happen to him in the future: "It was really scary too because I didn't know -the voices weren't threatening me or anything, which was good, but it was just, this is going to get worse and worse and worse until I'm like one of those guys on the street yelling at people or something. So I was really worried about that and scared it was going to get worse and worse. Which it did get, it just kept getting worse and worse, the voices and that…(they got) louder, more frequent, louder especially. They started off just like people whispering and after say a month they were yelling and it was going pretty much all day. I would get maybe five minutes at a time if that, without them…The first month or two it got worse and worse and worse and worse." David took the anti-psychotic medication he was prescribed, but it was not effective: And so then they gave me some medicine and told me it would get rid of it, but they only just kept getting worse and worse and worse…until a week or two later, it was virtually non-stop, these two guys talking to each other about me." The medication had no effect on his voices, if anything, they became more frequent. It did two things which weren't helpful. Firstly, David put on a lot of weight as a side-effect of the mediation: "The doctor said the drugs would work in 4-6 weeks and all they did was like make me fat and that's the only thing they did." Secondly, apart from hearing voices, David had always heard music. This was a very normal experience for him and he didn't realise that other people didn't experience the same thing: he thought that is what people meant when they said 'I have a song in my head': "They did one thing though which I didn't know at the time. When I went into the original time…I said I hear music, it's like I've got an iPod on. And they said that's not what everyone else has. Cos I started hearing music when I was about 14 or 15 and I just assumed that everyone had it and when someone said 'I've got a song in my head' I thought that's what they meant...it had never worried me or anything like that. And the drugs did get rid of the music, which was bad, because I liked the music….and I just had the voices left." He went back to the psychiatrist, who increased the dosage of his medication. He trialled the increased amount for a couple of weeks as advised by the psychiatrist, but unfortunately, it still didn't work for him. He went through the same process with a second drug, which also didn't work. Eventually, David put his foot down and said he would agree to take only a low dose of a third drug. This was also not effective; however, he continued to take this low dosage in case it was doing something. David put a great deal of effort into trying to make sense of what was happening to him. He pushed himself to read information about hallucinations and schizophrenia despite finding it very difficult to concentrate: "I read absolutely everything I could. I couldn't concentrate, so it was really hard, but I found if I put in ear-phones, it blocked it out, so I could read a little bit." All the information he found was very negative in terms of prognosis and focussed on treatment by medication: "I bought every book on Amazon I could find on schizophrenia and went googling and everything and all the drugs and stuff. And it was just all, ah you've got a brain disease and drugs fix it." David put a great deal of effort into trying to understand his experiences and wanted to be able to talk to someone about it. In his case, his need to talk about his experiences was not met through seeing his psychiatrist:

"He didn't talk to me much about what was going on. I say I've got voices, there's two guys talking and he sort of said ah yeah, that's early stage of schizophrenia that you've got and take these drugs and that'll fix it. And that's about all he said. He didn't really ask what was going on with anything, just said take these drugs and they'll fix you up."
David also only saw his psychiatrist for ten to fifteen minutes every two to three weeks. Overall, David found going to a psychiatrist to be unhelpful. He also questioned whether he really had schizophrenia or not, because from reading the diagnostic criteria, he felt that he only experienced auditory verbal hallucinations. He explained that while he developed paranoias, these were related to worrying that he was acting weird because of his voices, for example, talking back to them without realising, or that other people could tell he was hearing voices and judging him negatively for it. The idea that David's experiences were due to a degenerative brain diseasewhich was the explanation provided by the psychiatrist and in textbooks he readincreased his anxiety and made him feel hopeless for the future: "It made me really depressed and really scared 'cos it's like I thought, this is just the start, it's some degenerative thing that's going to keep getting worse and worse." It was particularly difficult for David given that the medications he trialled hadn't worked, because this is the only treatment for his voices that he had read about at that time.
David started seeing a psychologist and a social worker, both of whom he found helpful. Importantly, he was able to talk to both of them about his experiences. The social worker helped him to get back to work, which was very important to him. Through working with her, he developed a lot of diversional techniques, which helped him cope on a day-to-day basis. This was especially important during the first year after his diagnosis. He used relaxation to help him cope at work. If he noticed he was becoming tense, he would take a couple of minutes' break and breathe deeply. David started pushing himself to travel by train to work. He would stay on the train for one or two stops and then get off and wait for the next one, then get on again for another couple of stops and get off again. His anxiety about people judging him negatively gradually subsided. Eventually, he was able to stay on the train for the whole journey.
David's psychologist also recommended that he start practicing meditation, which relaxed him and helped to improve his concentration: During these initial months after he became unwell, David was working part-time. But with the increase in the frequency and volume of his voices, it became almost impossible for him to concentrate: "And so once it started getting to the point that it was virtually non-stop, I was incapacitated, I couldn't concentrate or anything. It was becoming really scary and really really annoying." David took a less stressful position at same workplace and cut down to part-time hours. This allowed him to maintain an income and some routine in his life. He didn't know anyone in his new department. Some of the people he used to work with knew what was happening in his life, but his new colleagues and boss were unaware of exactly what had been going on for him, just that there was something "wrong in (his) brain." David was worried that he was doing strange things without realising it. On top of that, because his new colleagues hadn't known him from before, he was worried that they were judging him negatively, thinking that he was a weirdo. He felt that he had to hide things from people, which he didn't like doing. All of these things created an immense emotional burden: "For the first six months or something at the new job, some of the old IT guys knew what was happening, but at my new job, no one knew I why I was sick. I felt like I was hiding things and I wasn't sure if they were thinking I was acting strange or -because like I said I couldn't really tell. So I thought I might be acting strange, no-one's going to say anything, because I've been, probably since I've started with them I've probably been acting strange, so they probably think I'm a weirdo or something." Because of David's worries about how he was acting and how others saw him, he tended to interpret ambiguous situations which could have been benign or not related to him as potentially indicating negative judgement from others: "I was really worried…like I said if someone's laughing, I thought that someone from HR had told them I had a problem and they were laughing because I had a problem." He was really worried that he would lose his job because he was acting strangely and people didn't know why: "I kept thinking ah they're going to sack me, they're going to get rid of me." Despite being reassured by his psychologist that he had shown no signs of becoming violent, David still feared that he would harm someone: It also meant that he slowly started shifting away from an interpretation of his voices as a meaningless phenomenon caused by neurons firing abnormally to a view that they were meaningful and potentially helpful: "At first I thought it was just nonsense they're talking about, I don't know, I just think they're talking nonsense. At first I didn't pay attention to what they were saying, I thought it was just a bunch of neurons in my brain firing and they're just making them say gobbeldy-gook, just stuff that doesn't matter. I went to the group for a little while and then they had a couple of presentations with people and they all talked about their own experiences and I read a couple of the books published on it and people were talking about having meaning in the voices and things like that." As David progressed in his recovery, he came to interpret his voices as trying to communicate something potentially useful to him, but in a strange way. He first started to think this way when he started attending HVN groups. During group meetings, the possibility of voices communicating something helpful, albeit in a sometimes frightening way, was discussed: "When I thought about it I thought there probably is meaning there, it's just they're saying differently in the way that you would normally say it…. I think they're trying to tell me things, they just tell me in a strange way."

APPENDIX I: Caroline's narrative
Caroline is the youngest of ten children born when her mother was 47. She was very close to her mother as a child and did not see a lot of her father, who was often at work. Caroline's mother died from cancer when she was ten years old. This had a big impact on Caroline, particularly given how close they had been. From the age of ten to fifteen she lived with her father, who she trusted and felt safe with. Sadly, her father died when she was fifteen. Caroline missed her father very much. She remembers feeling fearful after he passed away: After her father died, she went to live with an older sister, who never married. Caroline's sister was kind to her: she described her as being like a second mother. She explained though, that she did not feel that her sister was her real mother, and did not regain the sense of safety, security and trust in another person that she had with her parents:

"After I lost (my father) I just couldn't trust anyone and I felt lonely... my sister and brother couldn't be my real father and my real mother. I could live with them and get more trust (in them), but they were just my brother and my sister, and not my real parents."
Apart from the loss of her parents, Caroline's childhood was very happy. After her parents passed away, she put a lot of effort into things she needed to learn, including how to live by herself, look for work, progress in her studies and make a good life. She was a good student and felt that she could be successful in life: However, the voices also gave her instructions to stay away from others, not contact people, not believe what others said to her, which made her feel unsure. For example, they would tell her: "Don't contact that (person)", "don't believe that", "don't go there", "get (away from) other people -don't talk, don't contact (them)....just stay with yourself, don't believe anyone." The voices' comments functioned to isolate Caroline from those around her. Looking back, Caroline feels that it was at times when she felt alone that her voices told her to do "not the right thing….to hurt people or do something to myself." Most frequently, they instructed her to verbally abuse family members. Caroline described a sense of losing herself:

"When I was in trouble with hearing voices, I didn't know myself… I lost my feeling, lost my selfknowledge."
When Caroline started hearing voices, she lost control of herself and behaved in a way that she would not ordinarily wish to behave: "With the voice telling me to do things I just lost my self-control… I behaved toward people…just aggressively." While Caroline was not physically aggressive toward others, she was aggressive in the sense of holding herself above others, not speaking to them, and feeling she was special or better than them: "I was just thinking I learned spirituality…special knowledge, like the Buddhist teachings tell from the Buddhist book, and I was just thinking 'I'm wise, I have good wisdom', and I just looked down on people and just behave aggressively -I didn't talk, I just watched and thought by myself….Just thinking people foolish...just look down them." However, while one of Caroline's main emotions was feeling like she was a special human, "like the Buddha", who knew everything about people and didn't need to connect with anyone, her sadness was always there. Her sense of specialness and rejecting attitude toward others overlaid constant sadness, loneliness and distress: "The thing is I always felt upset, I always felt sad….with my life, with my loneliness." One of the worst thing about hearing voices, however, was not having any knowledge about them and how to deal with the experience. Caroline's lack of knowledge was disempowering: it meant that she could not differentiate between "good" voices which told her to do things which were beneficial to her and those which told her to harm herself or others.
"I couldn't make some consideration about which ones were not good voices, (that) told me to do the wrong thing and I couldn't control myself, I lost control… no one told me about the voices hearing voices -and the problem voices...(they) distressed me and (said) untrue things." After returning home from the monastery, Caroline went to see a psychiatrist and was prescribed a heavy dose of anti-psychotic medication, which led to side-effects, including eye-rolling, difficulty concentrating and feeling sleepy. She saw her psychiatrist fairly regularly, however, did not find the treatment helpful: Caroline started to work but was self-conscious about the side-effects of the medication eyerolling when she started work and felt that her co-workers were thinking negative things about her because of it (although now she thinks this may not really have been the case). With the support of her family, Caroline calmed down and had become less socially isolated. Subsequently, the doctor decreased the dose of her medication. Caroline found the lower dosage helpful in keeping her calm.
Once she had stabilised, Caroline got a full-time job and worked for a couple of years. During this time she continued to feel calm and had no major periods of low mood or distressing voicehearing. Following this, however, Caroline decided to enrol in a Masters degree, which was a very demanding and stressful. During her studies, she was diagnosed with hyperthyroidism and other medical difficulties, including muscle fatigue. She says: "My eyes were yellow, my skin was yellow, and I think I was admitted to hospital." She underwent liquid radiation treatment twice. The cumulative effect of being enrolled in a demanding course in context of medical difficulties took its toll on Caroline, who felt extremely tired: "I had no strength, I couldn't hold on on the bus, I couldn't climb up the stairs...my legs, my arms, they were just tired, you know? (I had) no strength." On top of this, her family discussed her living situation and decided that she could no longer live with her brother in the city. One of Caroline's sisters explained that her brother was too busy to look after her while she was physically unwell and that she would have to go to live with their other sister: "Yeah, she said I can't stay with my brother because my brother is just too busy and can't look after me well...and she just made decision move to stay with my sister…another one." Caroline moved in with her sister in a smaller town. She started to feel down, and this had a negative impact on her studies. She did not pass all of her subjects. In contrast, most of her friends completed the course in two years' time, and after they left university, she lost contact with them and became socially isolated. By the second or third year of her studies, her mood was very low and she had started to hear distressing voices again. She decided to stop her tablets without consulting her doctor. With the negative impact of her health problems, low mood and distress from hearing voices, Caroline considered withdrawing from her course. However, one of her lecturers supported her: "The teacher hadn't had any student who couldn't finish off the degree. She said "No no, not at all. All of my students need to get it"the degree -and they just said to me 'you have to come to my office every day, sit beside me...and do your work. She supported me. At that time she didn't know I had mental health illness, she just know I had a lot of health problems..." With the help of her lecturer, Caroline continued to study part-time and completed her Masters degree. At this time in her life, Caroline felt bad about herself. Although she had always been a good student and felt she had the ability to succeed, she did not achieve the success she had hoped for. She tried and tried to find a way out of her situation, to change her life, but couldn't.
In the end, she felt like everything was hopeless: Immediately after passing her Masters degree, she started a practical placement, however, was unable to pass due to her mental health problems interfering with her functioning. She was failed from the placement after six months: "I just felt down and felt upset...and I was just thinking I've lost my ability and just felt like 'I can't work anymore".
This was in stark contrast to how she felt when she finished her bachelor's degree: She had negative thoughts about herself, struggled with being able to differentiate between 'true' and 'untrue' voices, and experienced significant side-effects from the medication prescribed to her by her psychiatrist: "At that time I still feel mixed with the true and the untrue and a lot of side-effect of medication killed my confidence." She felt disappointed with her occupation because she felt she was unable to succeed after she became unwell:

"I was very upset with my occupation because I'm lost my success after I'm got the mental illness."
Caroline's psychiatrist focussed mainly on medication and did not help her to understand her experiences. As a result, she felt abnormal: It took at least one year for Caroline to find another job. Fortunately, Caroline was happy at her new workplace, and she started to feel better. Caroline explained that after she got a job that she really liked, she "recovered, yeah, and my life it just got better." She continued to work at the same organisation for 4 years, only leaving when she decided to immigrate to Australia with her husband. She recovered faster from distressing voice-hearing the second time she became unwell. Caroline explained that this is due to many things. Firstly, she was living with her sister and her nephew and nieces, two of whom were studying medicine or nursing and "knew about human problems and supported me". Her nephew and niece talked to her and helped her to cope with her voices: "I know a family member is really helpful, because I just tried to believe people around me and I tried to talk about my problem voices and about hearing voices to my sister, and my youngest niece said " don't think too much, don't believe it" or something like that....Yeah, I recovered, I got better." Finally, Caroline's medication was helping her to keep calm. She was still able to seek out activity and do things to make her life good. In contrast to the first time she became unwell due to distressing voices, she felt there was hope for her future, that she could get something valuable of life, and that she was a valuable human being with a contribution to make to others around her. Similarly, she finds it difficult to identify the voices and tell others which ones are 'good' or 'bad': "I can't tell people I hear this one or this one is the good one of them or this one is the bad one of them." Overall, though, Caroline used trial and error to find out how to cope with her voices. As any one of her voices may sometimes say helpful things and at other times advise her to do something harmful, it was not possible to discount everything a particular voice might say. Caroline started to ask herself whether the voice's comments were in keeping with Buddhist practice or not. For example, if it told her to be impolite to others, she would conclude that the voice was lying, as this is against Buddhist principles of not harming others. Today, if a voice is telling her something emotionally harmful, she respond to it saying "you're lying today". She also used mindfulness to help cope with her voices.
She also had discussions with other people in the temple she attended who had similar problems to hers. These group discussions were helpful and allowed her to increase her understanding of voice-hearing. Caroline emphasised, however, that everyone's experience of voice-hearing is different, and her understanding of voice-hearing and strategies that work for her may not apply to other people: After she moved to Australia, Caroline's understanding of her past mental health difficulties increased and she started to tell more people about it: "At the time I didn't tell people much, but after I moved to Australia I understood myself, I understood about my mental health problems in the past... and I just know that hearing voices makes people get lost." In Australia, Caroline attended group discussions through the Hearing Voices Network. These allowed her to deepen her understanding of her own experiences and contribute to helping others. She explained that in a group discussion, people can change their experience by hearing about other people's voice-hearing and learning from their experience and how they have recovered. Caroline said more workshops and dissemination of knowledge was needed to help voice-hearers who are struggling with their experiences. She sees a role for research in this area: "You have knowledgeable people to tell you (about it) and probably some guidelines from the study tell what happen to long treatment or right treatment or a problem with bad voices and good voices and helping them to recover right on track." Looking back, Caroline thinks a lot of things would have helped her to recover better and more quickly. Firstly, she emphasised that knowledge about voice-hearing and having someone to help her discern which voices were telling her the truth and which were not, would have helped her a great deal.
Apart from family members, she also sees medical professionals as a potential source of this kind of help and would have liked to have had a discussion with her doctor or psychiatrist about her voices. She also felt that being prescribed the right medicationi.e. medication which would calm her down but not lead to significant side-effectswould also have helped her.
"I just needed some knowledge, the right doctor, the right medication...to make me calm and make me recover more quickly." Today, Caroline is proud of herself and what she has achieved and sees herself as a valuable human being with a contribution to make to others. What makes her achievements particularly meaningful is the amount of effort that she had to put in to get where she was going. She had to have more patience and fight harder for what she wanted for her life than people who have not faced as many challenges. Having achieved the things she has in her life, she feels not only that there is hope for her, but that there is hope for other people with similar difficulties: "You know something people just have more difficulty in life, (but) they can get something of success of life too….because they have more patience, they just have more effort for their life, because they just feel like they fight a lot in life….to reach to the purpose of life and they just feel like they're wonderful, they're excellent."