Hallucinations are considered a fundamental symptom within psychiatry. ^{Reference Waters, Allen, Aleman, Fernyhough, Woodward and Badcock1} Auditory verbal hallucinations (AVH) in psychosis were richly explored by Jaspers, Bleuler and Kraepelinin in the early 20th century ^{Reference Oyebode2} and from this legacy focus has continued on AVH as a sensory perception in the auditory modality, principally in the form of speech, in the absence of external stimuli. ^{Reference Oyebode2,Reference Anthony3} However, it is increasingly recognised that AVH can occur in otherwise healthy individuals with rates of up to 20% of the population ^{Reference Bak, Myin-Germeys, Hanssen, Bijl, Vollebergh and Delespaul4} and can also occur in a range of non-psychotic illnesses, including post-traumatic stress disorder and borderline personality disorder. ^{Reference Larøi, Sommer, Blom, Fernyhough, Ffytche and Hugdahl5} To what extent AVH in non-clinical and non-psychotic populations is the same phenomena as AVH as a symptom of psychosis is unclear. One recent qualitative study of AVH in a mixed clinical and non-clinical sample, reporting experiences online, showed no clear distinctions between these two groups. ^{Reference Woods, Jones, Alderson-Day, Callard and Fernyhough6} Slotema et al similarly found no clear distinctions of AVH described by patients with schizophrenia as opposed to patients with borderline personality disorder. ^{Reference Slotema, Daalman, Blom, Diederen, Hoek and Sommer7} In neurobiological investigations, there is more evidence of a distinction; Looijestijn et al ^{Reference Looijestijn, Diederen, Goekoop, Sommer, Daalman and Kahn8} investigated sound localisation and demonstrated that only in psychosis were external AVH related to increased activity in the left planum temporale and right middle frontal gyrus cluster. Howes et al, ^{Reference Howes, Shotbolt, Bloomfield, Daalman, Demjaha and Diederen9} in a positron emission tomography study, showed altered dopamine synthesis capacity only underlies hallucinations occurring in patients with psychosis. One difficulty may be that current studies often use standardised measures of AVH or questionnaires with yes/no responses, and these do not capture the rich features that may distinguish subtypes of AVH or unique features in psychosis. Stanghellini et al emphasise this possibility, having investigated the distinction between psychotic and non-psychotic AVH over and above that gained by standard AVH measures. Participants were encouraged to provide further detailed descriptions, and those provided by patients with schizophrenia were very different to that provided by the non-psychotic group. ^{Reference Stanghellini, Langer, Ambrosini and Cangas10} AVH, as currently defined, may also be a poor way to distinguish a prodrome of a psychotic disorder. The occurrence of attenuated or infrequent AVH has been used, in part, to define the risk of developing a psychotic illness. ‘Transition’ to psychosis is made at a point predefined by standard criteria heavily influenced by the frequency of AVH, yet the clinical significance of this transition and role of AVH therein is challenged. ^{Reference O'Connor, Nelson, Lin, Wood, Yung and Thompson11}

Thus, as currently defined, AVH may show poor specificity for psychosis, and we suggest that an increase in primary phenomenological data is needed. The term ‘phenomenology’ has been used to label the detailed description of clinical features, signs and symptoms observed in pathological conditions, as well as being a method of analysing subjective experience that is rooted in a particular philosophical tradition. A phenomenological approach in this latter sense begins with the premise that the analysis of direct description can provide significant understanding and insight into the nature of the experience itself. Here, phenomenology is defined by a commitment to lay bare the ‘essence’ of any experience, independent of its causation. ^{Reference Broome, Harland, Owen and Stringaris12} In the context of a research methodology, phenomenology is characterised by a systematic investigation of subjectivity and a consideration of experience from the first-person perspective and can be viewed as the foundational science for psychopathology. ^{Reference Stanghellini and Broome13} The need to get at a true subjective experience may be particularly challenging in the experience of voice hearing in severe mental illness, whereby other positive symptoms such as persecutory delusions and cognitive and negative symptoms may form barriers to the communication of detailed accounts. Various methods have been used in the past to gain improved access to rich subjective data in psychosis in studies that focus on help-seeking, recovery and the meaning of experience. ^{Reference Lataster, Valmaggia, Lardinois, van Os and Myin-Germeys14–Reference Goldsmith16} When using a phenomenological approach to understand actual symptoms, we have previously used photo-elicitation ethnography in post-psychotic depression, finding this methodology helpful in increasing participants’ ability to communicate their experiences. ^{Reference Sandhu, Ives, Birchwood and Upthegrove17}

In this study, we aim for a modern description of AVH in psychosis based on a ‘naive’ exploration of the ‘concrete life worlds’ of those who experience this symptom. We aim to understand AVH phenomenology using an approach that does not attempt to relate an individual's experience to known or existing diagnostic or explanatory factors, and thus acknowledge the possibility of an entirely new framework emerging.

Results

Twenty-five participants with first-episode psychosis met inclusion criteria and consented to participate in the study. At this point of recruitment, data saturation was reached. See Table 1 for demographic and clinical description of the sample. In this paper, we report findings related to the quality of AVH experience. Two overarching themes captured participants’ descriptions of the quality of their voices: entity and control. Within each theme are codes and subcodes.

Discussion

This extensive qualitative study of AVH in first-episode psychosis from participant's primary descriptions reports on the complex and intricate nature of the quality of experiences. We have grouped these into two overarching themes of entity and control. Our results provide not only confirmation but also additional information of the phenomenology of AVH experienced in psychosis.

Participants reported AVH along dimensions of clarity and volume, that voices had an experience of utter ‘realness’, were experienced often as many voices and spoke in complex sentences. From first-hand report, hallucinations thus resemble true perceptions and have all the vividness of real experience. They were accompanied by a subjective sense of a lack of control, authority and were disruptive by their interfering, constant and compelling nature. Our results thus confirm previous literature dating back to early psychopathology. Reflecting Jaspers, we have shown a pattern of complex and veridical AVH in early psychosis, from subjective descriptions and before repeated questioning and framing of experience within healthcare settings. Our results suggest complexity is evident from the first episode of psychosis. In 1999, in their seminal paper, Nayani & David reported complexity increased with severity, and our results would concur with this within a first-episode psychosis sample. ^{Reference Nayani and David25} Many other previous studies using standardised measures report volume, controllability, commenting and number of voices as key aspects. These latter features were also highlighted as those with better ability to distinguish from clinical and non-clinical groups. ^{Reference Daalman, Boks, Diederen, de Weijer, Blom and Kahn26} Alternatively, voice location was described in our results as a confusing dimension; participants not only reported internal, external and mixed locations but also that this was unclear, difficult to describe and a changeable part of their experience. This suggests that this may indeed be the least clinically useful aspect and one that would not lend itself to robust research.

Novel findings from our study include the extent of AVH trying to control behaviour and emotions, the experience of AVH as an entity with complex social interactions and voices being imposed on individuals and occurring with multisensory additional factors.

In terms of AVH as controlling experiences, our results add additional evidence to a perceived lack of control and voice power in clinical populations. ^{Reference Upthegrove, Ross, Brunet, McCollum and Jones27} Many studies also focus on the power patients ascribe to AVH, and this is an aspect with clinical relevance, in terms of AVH as a traumatic experience, suicidal ideation, acting on commands and persistence of voices, that our results also support. ^{Reference Upthegrove, Ross, Brunet, McCollum and Jones27,Reference Birchwood, Iqbal and Upthegrove28} A number of participants reported the voice making demands, which may be likened to commands, but the assumption of instruction with power implied by a commanding voice should not be assumed. Voices were also often perceived as threatening and commenting on actions. These codes, which we have themed control of behaviour, are reflective of that already documented in descriptive psychopathology. ^{Reference Oyebode2} Although voice authority was a feature from our primary data, this was a nuanced and more complex experience whereby voices were trying to gain authority and power through extensive and complex social interactions: manipulating comments, mocking, tricking and blaming. Emotional content is suggested as a discriminating feature of AVH in psychotic disorders by Daalman et al. ^{Reference Daalman, Boks, Diederen, de Weijer, Blom and Kahn26} Our theme of emotional control concurs with this as a key area of future research and one that may have further clinical relevance in terms of risk of suicidal ideation and acts.

The suggestion of voice entity, over and above a realness of voice hearing, is also a novel finding which closely ties in with work that suggests voices are more than hallucinated sound. Bell proposes a social cognitive model of voice hearing in hallucinated social identities and one that is primarily a social relationship, with social environment predisposing to voice hearing from early childhood and later development. ^{Reference Bell29} We have previously shown that within psychosis, childhood trauma may have a symptom-specific risk for hallucinatory experiences; however, the aetiological process whereby people who hear voices acquire what amounts to internalised social actors is under-researched. ^{Reference Stanghellini, Langer, Ambrosini and Cangas10,Reference Upthegrove, Chard, Jones, Gordon-Smith, Forty and Jones30} In terms of neurocognitive models of AVH experience in psychosis, self-monitoring deficits may explain positive symptoms including hallucinations and passivity phenomena in schizophrenia with sensory confusion of self from non-self. ^{Reference McGuire, David, Murray, Frackowiak, Frith and Wright31} deWeijer et al and others suggested AVH could be related to an error in ‘corollary discharge’. This is the process by which the initiation of a thought or act, such as speech, is accompanied by an alteration in neuronal discharges that alert the individual to the fact that such a thought or act is self-generated. ^{Reference de Weijer, Neggers, Diederen, Mandl, Kahn and Pol32} With a self-generated thought or movement, accompanying signals to the sensory and motor cortex inform parts of the brain about an intended action, allowing the sensory areas to predict this possibility and code the sensory consequences as the expected consequence of this action. Allen et al ^{Reference Allen, Modinos, Hubl, Shields, Cachia and Jardri33} suggested that in schizophrenia there is a predictive coding error that accounts for the perceived externality or misattribution of internally generated speech to an external source. Thus, this may explain the realness and auditory quality of voices. The development of complex social interactions with voices are highlighted in our results and this area remains open to investigation.

Our other novel findings of AVH experience include an emphasis on imposition; the sensation of ‘being done to’ with passivity. Traditionally, ‘true hallucinations’ occurring in psychotic disorders are conceived of as indistinguishable from real perceptions except that there is no stimulus. Jaspers’ view is that normal perceptions and true hallucinations are received with an attitude of passivity. By this he meant that the experience could not be altered by force of will. ^{Reference Jaspers34} Our data further suggest that for patients with psychosis and AVH, this aspect of experience is accentuated. With ordinary speech, there is the option to ignore what someone is saying, to block it out, to change focus to other things or simply leave the room. With AVH in psychosis one cannot escape the experience and the verbal hallucination has a compelling and controlling dimension that renders it inescapable. Again, the role of corollary discharge may be relevant in the generation of this imposed AVH experience. ^{Reference Heinks-Maldonado, Mathalon, Houde, Gray, Faustman and Ford35} Alternatively, within a psychosis sample, positive symptoms that are the result of salience dysregulation would be expected to be accompanied by a subjective experience of ‘draw’ or compulsion to a given salient stimuli. This would be in keeping with Howes et al ^{Reference Howes, Shotbolt, Bloomfield, Daalman, Demjaha and Diederen9} and their distinction between psychotic and non-psychotic voices driven by dopamine dysregulation.

Finally, our novel findings include AVH experienced with additional multisensory factors, including pain, pressure and itching. Together with the reported feelings of imposition and control, the suggestion is that in this population there is an overlap between AVH and other positive symptoms of psychosis, such as hallucinations in other modalities or passivity phenomena. This may suggest distinguishing features that could be elicited in the phenomenology of voice-hearing in psychosis. However, recently Woods et al also report phenomenological data from a mixed sample using an online methodology, which included one open question subject to qualitative analysis; voices were also being experienced in a physical dimension, with 18% of their sample reporting AVH occurring with bodily sensations, including being in a dream-like state, feelings of heat or accompanied by tingling sensations. ^{Reference Woods, Jones, Alderson-Day, Callard and Fernyhough6} Thus, this area needs further research and replication.

McCarthy-Jones et al ^{Reference McCarthy-Jones, Thomas, Strauss, Dodgson, Jones and Woods36} call for subtyping to be a key advance needed in AVH research. Cutting across diagnostic groups, antecedent and treatment responses, and neurological categories, they propose that phenomenological subtypes include ‘hypervigilance’, ‘inner speech’, ‘autobiographical memory’, ‘epileptic’ and ‘deafferentation’. Our direct patient reports offer confirmation that some aspects of AVH in psychosis may be captured by these proposed groupings; however, additional factors would need further expansion. Primary research from our qualitative data reinforces the proposal that integration of methods in investigating AVH need stronger qualitative phenomenological focus. ^{Reference Upthegrove, Caldwell, Ives, Oyebode and Wood37} One limitation highlighted in the past literature includes the repeated use of standardised measures to record voice-hearing phenomenology. ^{Reference Stanghellini, Langer, Ambrosini and Cangas10} These include the Brief Psychiatric Rating Scale and the Positive and Negative Syndrome Scale, which measure frequency and disruption, ^{Reference Hedlund and Vieweg38,Reference Kay, Flszbein and Opfer39} and the Psychotic Symptom Rating Scale, which rates location, loudness, distress, controllability and beliefs. ^{Reference Haddock, McCarron, Tarrier and Faragher40} In terms of content and power, the Beliefs About Voices Questionnaire rates benevolence, malevolence, omnipotence, resistance and engagement and the Voice Power Differential Scale rates voice power. ^{Reference Chadwick, Lees and Birchwood41,Reference Birchwood, Meaden, Trower, Gilbert and Plaistow42} However, even in their totality, the novel phenomenological features we report, including the extent of social relationship and emotional control, imposition and multisensory additional features would not be captured.

This study has a number of strengths. Our use of novel, qualitative techniques allowed a more reflexive and direct aspect to AVH experience, and a mobile interview allowed participants to describe their experience in a non-threatening, open setting. By using diaries and photos collected prior to interview, the information we have gleaned was less ‘tainted’ by the researcher's questioning and should, therefore, be a credible representation of the nature of such experiences. Methodologies used were in-depth and aimed at gaining true first-hand accounts in a sample that is large in terms of qualitative research. Photo elicitation is a technique that is particularly useful in mental health research as it affords individuals the opportunity to explore feelings and experiences that may otherwise be difficult to articulate. ^{Reference Kantrowitz-Gordon and Vandermause43} Without the use of a priori dimensions, standard questionnaires or structured interviews, we have been able to capture AVH aspects that participants themselves feel are most important. We were also able to continue interviewing until data saturation (i.e. no new themes emerged), giving confidence that the essence of the experience was captured within our sample.

The study does however have clear limitations. The sample of patients with psychosis, although our intended focus, could reflect a group identified with diagnostic criteria that already include frequent and complex voices, and as such our results only reflect this limited nosology. However, it is clear from literature that the diagnostic outcome from first-episode psychosis is broad, including affective and non-affective psychoses. ^{Reference Heslin, Lomas, Lappin, Donoghue, Reininghaus and Onyejiaka44} Replication of our extensive qualitative methodology in non-psychotic groups is needed before any firm conclusions can be drawn. Although our sample contained participants from both genders, participants were predominantly male and from a White ethnic group. Future research is also needed using this methodology in non-help-seeking populations, and other cultures, to ascertain whether AVH experience is similar in these differing populations.

However, our results should be of clear interest to clinical, translational and biological researchers. Studies aiming to understand the underlying neuropathology of auditory hallucinations have been challenging. Neuroimaging studies looking to distinguish psychotic from non-psychotic AVH produced mixed results, ^{Reference Upthegrove, Caldwell, Ives, Oyebode and Wood37,Reference Diederen, Daalman, de Weijer, Neggers, van Gastel and Blom45} and it may be that some aspects of AVH are common across all voice-hearing populations; however, key distinctions in psychotic voices, if present, have not been captured by studies using standardised rating scales. Our results should provide a starting point for future research aiming to address these questions and suggest that the themes of voice entity and control with their associated codes (see Tables 2 and 3) are important aspects to study and ones that may suggest areas of key clinical relevance.