The Anorexic Voice: A Dialogical Enquiry and Thematic Analysis

Abstract

Although prevalent among those given a diagnosis of anorexia nervosa (AN), the internal anorexic voice (AV) is little understood. This study aimed to explore the AV’s role in the development and maintenance of AN from a new perspective – that of the AV itself. Nine women with a diagnosis of AN participated. Data was collected via dialogical enquiry, which entailed direct, semi-structured interviews with participants’ AVs. Transcripts were analyzed using thematic analysis. Three themes build a picture of a symbiotic yet destructive relationship between AV and individual. Firstly, “Pragmatism” describes the AV’s supposedly invaluable, problem-solving persona. Secondly, “Relationship” depicts the valued yet fraught, unbalanced, and often unwanted bond between AV and individual. Thirdly, “Self-preservation” presents the AV’s drive to retain control for its own survival. In clinical practice, curiosity regarding relationships with AVs could highlight barriers to recovery and inform treatment. Future research should focus on diverse samples and consider ways of working with eating disorder voices to promote recovery.

Keywords:

• Anorexia nervosa
• anorexic voice
• chairwork
• eating disorders
• voice dialogue

Introduction

Anorexia nervosa (AN): the need for improved understandings

Anorexia nervosa (AN) is a potentially life-threatening mental health difficulty (Arcelus et al., 2011), with a lifetime prevalence of up to 4% (Smink et al., 2013). Unfortunately, current treatments are associated with high dropout and relapse rates (Atwood & Friedman, 2020; Lock & Le Grange, 2019). In addition, motivation to recover is often low (Blake et al., 1997; Gregertsen et al., 2017) and attempts to enhance motivation in eating disorder (ED) therapy show limited effectiveness (Knowles et al., 2013; Waller, 2012). Hence, alternative developments in the conceptualization and treatment of AN are needed.

Why research the internal anorexic voice (AV)?

The anorexic voice (AV) appears to be central to some individuals’ experiences of AN (Pugh, 2016; Tierney & Fox, 2010; Williams & Reid, 2012). Eating Disorder Voices (EDVs) constitute an internal and often hostile commentary on one’s eating, weight, shape and worth (Pugh, 2016) and are usually experienced as both part of, and phenomenologically separate from, the self (Fox et al., 2012). While a significant proportion of individuals living with EDs experience an EDV (Noordenbos et al., 2014), it remains poorly understood.

Preliminary research implicates the AV in the development and maintenance of AN (Pugh, 2020). For many individuals, AVs are initially experienced as benevolent but increasingly hostile over time (Tierney & Fox, 2010; Williams & Reid, 2012), affecting the individual’s sense of self and driving disordered eating (Jenkins & Ogden, 2012; Rance et al., 2017; Seed et al., 2016; Williams et al., 2016). Similarly, individuals often ascribe positive functions to the AV during the early stages of AN (e.g., guidance and companionship), followed by dysfunctional and egodystonic qualities during later stages (e.g., coercion and punishment) (Tierney & Fox, 2010). Despite this, some fear losing the AV (Tierney & Fox, 2010; Williams et al., 2016) while (sometimes simultaneously) describing it as an obstruction to recovery (Smethurst & Kuss, 2018; Tierney & Fox, 2011). In addition, quantitative studies have associated the perceived strength of the AV with severity of eating pathology in AN (Pugh & Waller, 2016, 2017; Serpell et al., 2004). Accordingly, changing the ways in which individuals relate to their AV may help promote recovery for some individuals (Aya et al., 2019). Given the challenges of treating AN, as well as some individuals’ attachment to the AV and ambivalence toward severing ties to it (Tierney & Fox, 2010), better understanding the EDV’s perspective and motives may help support engagement, formulation, and positive outcomes in treatments for AN where they feature.

Self-multiplicity and the dialogical self

Self-multiplicity describes a conceptualization of the mind as consisting of multiple “selves”, “modes”, "subpersonalities", or “inner voices” (Lester, 2015; Rowan, 1990). Within psychology, self-multiplicity can be traced back to Friedrich Nietzsche’s descriptions of agentic “drives” (Poellner, 1995), Williams James’ (2021) distinction between the “self-as-knower” and “self-as-known”, and Carl Jung’s (1960) autonomous “feeling toned complexes”. In the last few decades, self-multiplicity has emerged as an important concept in many psychotherapies, most notably voice dialogue (Stone & Stone, 1998), internal family systems therapy (Schwartz & Sweezy, 2019), and schema therapy (Young et al., 2003). Whether these parts of the self-constitute autonomous subpersonalities (the hard multiplicity view), transient states of thought and feeling (the modest multiplicity view), or are purely metaphorical (the soft multiplicity view) remains contested.

Dialogical Self Theory (DST) (Hermans, 2001) has attempted to bridge disparate conceptualisations of self-multiplicity. According to DST, parts of the self (referred to as “I-positions”) are capable of engaging in dialogical relationships with one another and are subject to power dynamics, with certain I-positions dominating or supporting others. Far from being pathological, DST conceptualizes this “multi-mind” as natural and, in many ways, adaptive. However, imbalances or disharmonies within the multiple self may contribute to forms of emotional distress and psychopathology (e.g., Dimaggio et al., 2010). Examples include “tyrannical dialogues” in which the individual’s inner world is dominated by one (often overbearing) I-position (e.g., the “inner critic” in depression) and “cacophonous dialogues” involving rapid and chaotic oscillation between I-positions.

It has been observed that the inner worlds of individuals with EDs are remarkably multi-voiced (Schwartz, 1987), yet few studies have explored this experience directly. Research has tended to explore either personal accounts of living with EDVs (e.g., Tierney & Fox, 2010) or measure the strength of its association with dimensions of disordered eating (e.g., Pugh & Waller, 2017). However, there is a risk that self-reports (whether they be questionnaire or interview-based) yield less informative “reappraisals” of internal events, rather than intuitive and experience-near accounts that accurately reflect lived experience (Safran & Greenberg, 1982). Moreover, if the multi-voiced nature of EDs does contribute to important features such as ED cognitions and ambivalence toward recovery (e.g., Bell, 2013; Pugh & Waller, 2017), applying DST in real-time may provide valuable insights into EDVs as they manifest in real-life contexts, allowing for a more comprehensive understanding of these experiences. For this reason, experiential methods such as chairwork may offer valuable medium for exploring the AV.

Chairwork, voice dialogue, and the AV’s perspective

Chairwork is an experiential medium for psychotherapy, utilizing chairs, movement, and dialogue to directly communicate with parts of the self (Pugh, 2019; Pugh et al., 2021). Voice dialogue (Stone & Stone, 1998) is a form of chairwork and a distinct therapeutic method which has been successfully applied in work with voices, including those arising in the context of psychosis (Corstens et al., 2012; Longden et al., 2021). In summary, voice dialogue entails the individual changing seats and speaking from the perspective of an inner voice. Questions are then put to the voice by the “facilitator”, such as where it originates from, what its role is in the individual’s life, and what its intentions and concerns are. At the end of the conversation, the individual returns to their original chair and reflects on the dialogue from a position of enhanced separateness and self-awareness. By talking directly with voices, insights into the nature of these parts of the self are garnered, both for the voice-experiencer and individuals who support them (Longden et al., 2021; Schnackenberg et al., 2018; Steel et al., 2020). Research indicates that voice dialogue and associated approaches such as “talking with voices” are acceptable interventions for voice-hearers (Longden et al., 2022b), the professionals who work with them (Longden et al., 2022a), and the voices themselves (Middleton et al., 2022).

Initial evidence suggests that chairwork and voice dialogue are also acceptable interventions for individuals with EDs (Chua et al., 2022). However, to the best of our knowledge, these approaches have not yet been used as a research method. In this study, voice dialogue was used to investigate the nature of AVs directly, which we refer to as “dialogical enquiry”. In addition to developing a more accurate and comprehensive understanding of lived experience with an EDV, doing so from the perspective of the voice itself may have clinical implications, such as informing case conceptualisations where EDVs are present, better understanding voice-hearer dynamics in AN, and guiding appropriate interventions. In addition, exploring the AV from its own perspective may go some way toward reducing internal conflict, promoting self-compassion, and increasing self-understanding (Chua et al., 2022).

Objective

This study aims to explore the AV’s role in AN development and maintenance from a novel perspective: that of the AV. Qualitative data were collected using dialogical enquiry, which involved semi-structured voice dialogue interviews with AVs. Transcripts were analyzed using thematic analysis (Braun & Clarke, 2006). Focal topics for the dialogical enquiry included the AV’s functions, needs, and interactions with the participant over the course of AN. The aim of the study was to generate new insights into the nature of AVs and disordered eating, informing how these experiences are conceptualized and addressed therapeutically.

Method

Participants

A sample of nine participants was recruited from four UK ED services. Recruitment was curtailed at this point due to the COVID-19 pandemic. Participants were aged 18 and above, diagnosed with AN (most commonly, the restrictive subtype of AN), and reported an AV. All participants identified as female and were in their twenties or thirties, with a median age of 28 years. The sample was predominantly White British, although BAME British and White European ethnicities were also represented. Participants’ body mass index (BMI) ranged from 14.4 kg/m² to 19.2 kg/m² (mean = 17.3 kg/m²) and had a self-reported illness duration from 2 to 20 years (mean = 12.6 years). All participants were currently receiving ED treatment, such as physical health monitoring, dietetic consultation, and psychological therapy. Experiences of inpatient admission and cognitive behavioral therapy (CBT) were common in the sample.

Data collection

Qualitative data were collected via semi-structured dialogues with participants’ AVs, which was informed by the Dialogical Interview Schedule for Eating Disorders (DIS-ED) (Pugh, 2019). The DIS-ED was initially developed with reference to key approaches to dialoguing with voices, principally voice dialogue (Stone & Stone, 1998), as well psychodrama (Moreno, 1987), ego state therapy (Emmerson, 2003), and the “talking with voices” approach (Corstens et al., 2012). The DIS-ED was drafted by the fifth author and subsequently revised following consultations with two service-users with lived experience of AN.

The DIS-ED was divided into four phases. First, the participant was asked general questions about their experience of the AV (phase one). Next, the participant was asked to move to a second chair, placed at a comfortable distance, and adopted the perspective of their AV (phase two). The AV was then asked a series of exploratory questions regarding its functions (e.g., “What is your role in this individual’s life?”), content (e.g., “What do you tend to say to this individual?”), origins (e.g., “When did you come into this individuals life?”), intentions (e.g., “What do you want for this individual?”), relating (e.g., “How do you feel towards this individual?”), and underlying concerns (e.g., “What might happen if you didn’t perform this role?”). This phase was the focus for the qualitative analysis. In the third phase, the content of the interview was summarized by the researcher to support participants’ awareness and separation from the EDV. In the fourth and final stage, participants reflected on the interview in their original chair (the “central” position) and responded to the AV if they wished. Demographic data were collected from participants at the outset of the interview.

Qualitative data used for this study was that collected in phase two of the DIS-ED (embodying the voice). However, it is important to note that each phase has value in enabling dialogue, separation, and awareness between participants and their EDV.

Analysis

Thematic analysis, following Braun and Clarke (2006) guidelines, was chosen for its flexibility and established use in research regarding AN (Rance et al., 2017), the AV (Tierney & Fox, 2010), and chairwork (Pugh et al., 2021). A critical realist (Willig, 1999) position was taken throughout, with the AV viewed as a “real” phenomenon to be understood and situated in interaction with social discourse. Correspondingly, analysis was largely inductive, though participants’, researchers’, and readers’ contexts inevitably influence the meanings drawn. The analysis followed the six-phase approach outlined in Braun and Clarke (2006) method. This consists of familiarization with the data, generating initial codes, searching for themes, reviewing themes and generating a thematic map, defining and naming the themes, and organizing them into a coherent narrative, and finally producing the report. The first author completed the initial coding and developed the framework of descriptive themes, discussing the phases of the process with the second and last authors. See supporting information for a worked example of the analytic procedure.

Trustworthiness and ethics

Quality control guidelines (e.g., Elliott et al., 1999) were adhered to throughout the research process. “Bracketing” was applied by the first author throughout, through regular discussion with the last author and a reflective log. Details of the first author’s context are disclosed for transparency. At the time of data collection and analysis, the first author was a clinical psychology trainee in her mid-twenties. She identifies as white British, heterosexual and female. She has experience using thematic analysis in mixed-methods research into stigma, and does not consider herself wedded to an epistemological position. This was her first time using chairwork, supported by the last author who is an experienced chairwork practitioner. She has personal experience of AN and identifies with the AV concept. Broad participant demographics are provided to “situate” the sample, while protecting anonymity. The results of the analysis are grounded in examples so that readers can evaluate their fit with the raw data. Readers are also encouraged to acknowledge aspects of their own context that contribute to the meanings developed as they peruse this article. This study underwent HRA ethical approval. All participants provided informed consent.

Results

Overarching concept: an interdependent relationship

The analysis describes a symbiotic yet destructive relationship between the AV and participant – that the individual “needed me [the AV], and I latched myself onto her, and we formed a bond.” The first theme, “Pragmatism”, outlines the AV’s view that it solved participants’ problems. The second theme describes the complex dynamics of the AV-participant “Relationship”, which seemed important to the individual yet fraught, unbalanced, and, to some degree, unwanted. Finally, the third theme, “Self-preservation”, explores the AV’s determination to maintain a presence in the person’s life, driven by the imperative that “I need her”.

Theme 1: pragmatism

Part of the allure of the AV was its ability to provide tailored solutions to the person’s problems, whether past, present, or anticipated difficulties in various arenas of life (school, work, interpersonal relationships, and so on). This fostered a dependence from the person on the voice in terms of coping with daily life, creating a positive feedback loop of seemingly effective pragmatism.

Subtheme 1.1: problems

In all cases, the AV described addressing perceived problems relating to the individual’s inadequacy, vulnerability, intolerable emotions, or potential to lose control. “Solutions” offered by the AV generally involved improving the person (e.g., ridding them of their inadequacies) and punishing them when they inevitably presented signs of insufficiency. The AV viewed the person as intrinsically flawed, undesirable, lazy, immoral, worthless, at risk of rejection, or worthless. Moreover, it stressed a link between the risk of rejection and the person’s weight and shape. In some cases, the voice went as far as labeling the person as undeserving, for example, of “an easy life because she’s disgusting and a waste of space”.

The AV pushed the person away from perceived shortcomings by promoting extremely high expectations “in everything”, creating an unachievable standard for improvement. The voice promised a solution to rejection, unworthiness, and immorality, stating:

“She needs to look thin to be accepted … If they see how ugly she is from the outside, they will discover how ugly she is on the inside.”

According to the AV, failure to meet these standards further justified dependence on it to remedy perceived deficiencies, with the person’s “suffering […] mak[ing] her a better person”. It viewed the alternative as becoming “overweight”, failing (e.g., academically), or suffering total rejection. In short, “no one would want her if she didn’t have me”.

The AV provided negative reinforcement by blocking unwanted emotions, as well as positively reinforcing desirable emotions. Much of this emotional distress appeared to be a “trauma” response (e.g., to bullying, childhood abuse, family conflict or breakdown, or the challenges of adolescence) that “buil[ds] up over the years”. In some cases, the AV traced its development back to these events and the individual’s associated emptiness, loneliness, guilt, fear, sadness, anxiety, or “emotional turmoil”. It presented itself as a tool to avoid “dealing with” or expressing emotions that were considered “too much to bear”, with death considered preferable to experiencing these feelings:

“I could kill her one day but at least she wouldn’t be feeling.”

The AV sought to circumvent these emotional experiences, “blocking out” potential triggers (e.g., distressing thoughts and memories) by numbing and distracting the person:

“I can numb things for her … if she could focus on me, then she wouldn’t be focusing on the dark things that happened.”

At the same time, the AV provided positive emotions by satisfying the individual’s urge to please:

“When she does what I say… I’m pleased, then she feels good.”

Finally, the AV feared that the person “wouldn’t cope” without it. In some cases, it believed that this would likely result in suicide.

A common aversive experience for those who suffer from AN is a loss of control, on which the AV universally capitalized. Transitions such as moving school, entering adolescence, starting university, or marriage were linked to the voice’s development – departures from what was known and controllable for participants. Additionally, the voice saw the person as lacking in self-control, predicting “she’s going to lose control around food”. In this way, the AV reinforced the person’s fear of losing control and attempted to fill this void of uncertainty and discomfort with its narrative. It did so by becoming a “safety blanket” for the individual and establishing a link between low weight and a sense of control:

“When your weight is low … it shows that you’re in control because it takes control to go against what your body is asking for.”

Conversely, the AV believed that without it, the person would inevitably “get overweight”, assuming they lacked control over their eating habits. The AV did, however, acknowledge its paradoxical nature as both a consistent and uncontrollable force in the individual’s life:

“I’m a certain in her life … everything else is uncertain. And I am predictable and uncontrollable (laughs)”.

Another aversive experience that the AV aimed to remedy was the individual’s negative self-concept and lack of identity. This was apparent in the participants’ self-hatred and fear of “truly” becoming themselves:

“[She] realized that she was just genetically programmed to be disgusting.”

The AV responded to this by allowing the person to “be something” and gain comfort in the identity of the “anorexic”. The behaviors it encouraged served to bolster participants’ self-esteem, mastery, and sense of accomplishment (e.g., restricting food intake to feel better about oneself):

“She can be the thinnest … this is something she can be good at – she can take pride in.”

The AV believed that the alternative would be an inability to cope with self-loathing and a sense of nothingness. Therefore, the AV positioned itself as invaluable by promising foundations for identity and self-esteem.

The AV also underlined its vital role in the person’s life by presenting the world, and the person, as inherently unsafe and in need of protection. It described the individual as “vulnerable”, the world as “a scary place”, and others as “unpredictable” or unlikely to meet the person’s needs. Similar to its role as a “security blanket” that provided control, the AV often took on a protective and almost motherly stance toward the individual:

“I’m like a mum to her, I can keep her safe … like a guardian angel”.

The voice sought to defend the person against both the internal threats mentioned previously – inadequacy, unwanted emotions, and a lack of self-concept (“an escape from feeling lonely and feeling unwanted”) – and external ones. By limiting the person to “safe foods”, it implied that anything else was somehow dangerous. Moreover, the AV described protecting participants “from herself” and others, “so she can’t get hurt by anyone”. However, it acknowledged that this protection was inauthentic, giving the person a “false sense of security” in exchange for obedience:

“I’m nurturing until she does anything against me and then I turn into a monster.”

These threats to the person’s psychic and physical stability were furthered by the voice’s insistence on its irreplaceability. Without it, the individual “would become reliant on people and people aren’t reliable”. Indeed, in its absence, the voice wondered, “who would she turn to?”

Subtheme 1.2: solutions

The AV claimed to address many of the person’s aversive experiences, including painful emotions, a lack of control or identity, and attacks within their internal and external world. The solutions it offered were multifaceted and highly constraining, limiting the person’s freedom and furthering a sense of reliance and absolutism. For instance, whatever the perceived problem, the AV’s solution invariably involved intense criticism:

“I want what’s best for her … If that means I have to be hard on her, then so be it.”

The AV was also competitive, endeavoring to “make sure that she’s the best”. Its criticisms tended to focus on the areas of perceived inadequacy mentioned earlier – unattractiveness, failure, laziness, immorality and worthlessness – as well as deviation from the identity of “the anorexic”, loss of control, or vulnerability. Furthermore, the AV was highly demanding, establishing and enforcing strict rules and instructions that entangled weight and appearance with solving problems. Its language often included “should”, “shouldn’t”, and “can’t” statements:

“She can’t go up a clothing size …. her clothes shouldn’t fit her. Her clothes should hang off her.”

The AV’s rules typically related to food and exercise and included extensive instructions and specific conditions under which food was allowed, including amounts (through “weigh[ing] out” the food), types (only “certain types”), timings (“not allowed to eat before the evenings”), and compensatory behaviors. Part of the AV’s relentlessness was its invention of new (and often unsatisfiable) standards, which propagated a cycle of disordered eating and intense exercise:

“If you then eat additional amounts … [or] types of food, I need to invent a new amount of exercise to compensate … Sometimes I’m not satiated so I will just continue.”

Another part of the pragmatism that characterized the AV was its position that it was the only solution for problems in participants’ lives. The voice argued that “she needs me” and would limit the person’s agency, thereby not giving the individual a “choice to do anything else” and muddying boundaries between itself and the participant:

“I’m so intertwined in everything she does, she almost doesn’t know … and even I don’t know … where she starts and I end”.

Moreover, the dependence between AV and the person was bidirectional and compulsive (“I can’t stop”), reinforcing the belief that there was no alternative.

Subtheme 1.3: a faulty solution

Paradoxically, the AV often recognized that its solutions were faulty – ineffectual, damaging, unnecessary, or legitimizing of harm. Accordingly, it tended to rationalize its approach by suggesting its intentions are good.

“I’m … trying to make her happy … acceptable … good enough. But then also, I know that I’m not”.

In addition, the AV consistently acknowledged “damaging” the individual, noting that it did so physically (“lack of energy … brain fog … feeling cold … hair falling out … feeling really ill”) and emotionally, making the individual “anxious and sad”. The AV was aware of what it was doing to the person: it took responsibility for experiences of depression, anhedonia, suicidality, guilt, shame and obsessiveness, and labeled itself a “handicap”. In most cases, it went as far as acknowledging that the person may well “be happy and healthy” without it and could “live a fulfilled life”.

At the same time, the AV argued that it had “done a lot” for the individual and that it was the latter’s disobedience and noncompliance that undermined its efficacy or made its efforts “useless”. It often tried to absolve itself of blame by claiming “good intentions” and asserting that it is “not all evil”. Yet, in all cases, the AV also acknowledged that its problem-solving rationalized wrongdoing; that it was at times an “evil…spirit…enveloping and suffocating her”, with sadistic intentions:

“It will be so nice again to destroy her life”.

The AV’s defensiveness and justification stood at odds with the imagery it used to describe itself, which echoed its destructiveness. Metaphor reinforced its empty, parasitic qualities: it pictured itself as a “faceless” or “hollow-eyed” shadow. The AV’s threatening disposition was reinforced by descriptions such as digging its “claws” in, or “latching” onto the person, underlining its dependence on the participant. Some AVs likened themselves to fictional monsters:

“Like a dementor in Harry Potter … sucking the happiness and soul out”.

Theme 2: relationship

The relationship between the AV and the individual was emphasized as extremely important. This connection displayed elements of co-dependence but was also fraught, unbalanced and, in most cases, unwanted by the individual. The AV tended to adopt a powerful, superior role, describing how the person “looks up to me” and “wants to please me”. In all cases, the AV “keep[s] her controlled… and under [its] powers” while the individual was comparatively powerless:

“She’s laying on the ground and I’m standing on her back.”

The AV described how it was “intertwined like in everything she does” and felt that its relationship with the individual was akin to a parent-child or romantic relationship. Noted previously, the voice also likened itself to a “motherly entity”, but often took on the role of an abusive parent or partner.

This entanglement was dichotomous as there was clear co-dependence (“we’ve grown together… we both need each other”) and yet also a “love-hate relationship”. This resulted in instability and conflict, which would often become “a big fight” where there was “winning” and “losing”. Opposition, rebellion, or desire to rid themselves of the voice could lead individuals to disregard it at times:

“Every time I spoke to her, she blocked me out. I wasn’t allowed to her birthday party”.

Being rejected was painful for the AV, especially given its dependence on the individual. Furthermore, it described separation as “very difficult” (further emphasizing the seeming impossibility of separation) and it often felt unwanted:

“We’re in a relationship where I want to stay and she doesn’t want me around”.

While the AV-individual relationship was important and valued, the analysis points to a complex, abusive, and precarious connection.

Theme 3: self-preservation

There were many ways in which the AV sought to keep itself alive and functioning. This included increasing its volume and demands, as well as punishing the person and obstructing engagement in treatment. Through this, the AV ensured that its needs were met, which appeared to be part of what made AN so pernicious.

There was a progression in the AV’s behavior and influence, with it starting out “innocuously” but growing “louder”, “more demanding and more controlling”, and potentially lethal over time:

“I don’t know how much stronger I’ll get – I could kill her one day.”

Ever more “fixated” on the individual’s eating, weight and shape, the AV used “pretty much anything” to influence them and strengthen their relationship. This included colluding with toxic “social norms”, “family pressures”, and “bullies” – whichever was most convenient. It also latched onto other physical experiences to strengthen its power and influence:

“She got some weird stomach bug … I used that to get her to stop eating”.

The AV’s need for “the dependence she has on me” often led to it instigate cycles of dietary restriction which reinforced and perpetuated its “job” or “purpose”:

“It’s almost like her not eating gives me fuel”.

Another way the AV strengthened its relationship with the individual was through threat and punishment, with the AN becoming “furious” if their bond was weakened:

“When she tries to loosen the rules, I come back with a vengeance”.

Universally, the AV described efforts to impede recovery from AN. It often dismissed or minimized the condition and its symptoms, arguing that “there’s nothing wrong with her” despite evidence to the contrary:

“She woke up with cardiologists around her trying to bring her back to life … but she was fine the day before, so I think they were all exaggerating”.

It made promises of a better life if the individual picked its side (“everything will be okay if she starts restricting her food again”) and often mocked their attempts to recover:

“She’s pretending she’s okay … trying ridiculous methods to recover”.

Furthermore, the AV often sought to disrupt treatment, for example, recounting how it “convinced her to stop going to therapy” and had encouraged another individual to “attach weights to her legs … to keep her out of hospital”. It also challenged competing relationships by painting other people as threatening or ill-intentioned (they want “to make her fat”), fearing that “she’ll find someone else” if it did not. This led to difficulties in participants’ interpersonal lives.

“[Her boyfriend] was a big threat to me… so I pushed him away”.

Consequently, the AV would capitalize on “difficulties in life” (e.g., with studying, conflicts with others, or isolation) to help it “get back in control”:

“Everyone is full of support for the first two or three weeks [post-discharge] … Gradually, they all get back to what they are doing … She’s lonely … She comes back, listening to me”.

Overall, the AV represented its power as seemingly never-ending. Moreover, it anticipated returning to the person’s life whenever such an opportunity arose.

“I’ll always stick around and wait for her”.

Discussion

This study aimed to provide insights into the nature of the AV using a novel method of data collection informed by the voice dialogue approach: dialogical enquiry. The analysis revealed symbiotic yet destructive relationship between the individual and the AV that appeared to play a central role in the experience of disordered eating.

The AV and the individual seemed to be locked in an interdependent relationship, whereby the AV offered solutions to the person’s problems yet perpetuated them by encouraging disordered eating. It promised to solve problems related to inadequacy, intolerable emotions, loss of control, negative self-concept, and vulnerability, and provided a way to cope with trauma and life transitions. Moreover, its solutions were tailored to the problem in question, but consistently featured criticism, inflexible rules, and few alternatives. Nonetheless, the voice acknowledged that it was often ineffectual, damaging, and impeded life. The relationship between the AV and the individual was important to both parties, yet fraught and partly unwanted by the person, who appeared relatively powerless. However, the AV was reliant on the individual for its continued existence, escalating and adapting to preserve its influence over them. It displayed confidence in its purpose, pleasure in disordered eating, and frustration when it was ignored. If it was thwarted in obstructing the individual’s recovery, the AV intended to regain power.

While the concept of the pragmatic AV (proposing tailored solutions for perceived problems) may be novel, the problems and solutions identified correspond with existing research. In regard to “problems”, the AV’s concerns that the person was fundamentally inadequate mirror the pronounced levels of self-criticism accompanying EDs (Zelkowitz & Cole, 2020). The voice sometimes traced its origins to traumatic and/or transitional experiences, echoing previous EDV research (Pugh et al., 2018) and the “Making Sense of Voices” framework, which conceptualizes voices as a meaningful reaction to these kinds of events (Romme & Escher, 2000). Similarly, the difficulty managing “intolerable emotions” corresponds with research identifying reduced emotional awareness, increased emotional dysregulation, and the use of maladaptive coping behaviors among those with AN (Oldershaw et al., 2015). Loss of control has also been identified as a contributor to the AV’s development (Tierney & Fox, 2010), while the AV’s reference to identity dissatisfaction and ambiguity is consistent with the role of the “lost sense of emotional self” in AN (Oldershaw et al., 2019). Finally, qualitative studies indicate that fears of vulnerability are central to the maintenance of AN (Fox & Diab, 2015).

Likewise, the AV’s “solutions” echo findings from prior research. Descriptions of AN as a form of self-improvement are prevalent in pro-anorexia forums (Bates, 2015; Knapton, 2013) – one argument by which the AV justified its influence. The AV’s offer of emotional numbing concurs with evidence that starvation numbs emotions in AN (Brockmeyer et al., 2012; Serpell et al., 1999), while the illusion of control and sense of accomplishment it provides are well documented in AN (e.g., Nordbø et al., 2006). Similarly, the AV’s protective functions (Tierney & Fox, 2010) and role in fulfilling core needs (Sands, 1991; Williams & Reid, 2012) have been previously identified. Hence, many well-established maintenance factors for AN are reiterated through discussion with the AV, but this time appear as a series of problems and solutions offered by the voice.

The bond between AV and individual also concurs with existing literature. Affiliation toward the ED and attachment insecurity are both associated with ED symptomatology, supporting conceptualizations of EDs as an attachment relationship (Mantilla et al., 2018, 2019). Individuals have emphasized the AV’s dominance of the self and described valuing their relationship with it (Williams & Reid, 2012), while conflicts between valuing and feeling abused by the AV reflect the mixed feelings many individuals have about recovery (Dawson et al., 2015). Consistent with the results of this study, individuals have likened living with the AV to an abusive relationship, characterized by punitiveness, coercion, and entrapment. Moreover, the AV was content to “lurk in the background” and jeopardize future attempts at healthier living (Tierney & Fox, 2011). This provides one explanation as to why relapse is common in AN (Berends et al., 2018).

The development of AN may be understood in relation to the AV’s self-preservation instinct. Here, the AV described a gradual escalation to build its influence over the person, corresponding with experiences of the AV as initially supportive but increasingly hostile over time (Tierney & Fox, 2010). Drawing from the voice dialogue literature, the evolution of the AV from benevolent to malevolent is reminiscent of the concept of the “Killer Critic” (Stone & Stone, 1993), whereby there is a discernable shift in the “inner critic” toward dominance, contempt for the individual, and a desire to inflict harm. Indeed, the voice’s intention to retain or regain exclusive power over the person may contribute to low or mixed motivation to recover for some individuals with AN (Blake et al., 1997; DeJong et al., 2012; Knowles et al., 2013; Watson & Bulik, 2013), as well as interpersonal mistrust, avoidance, and social isolation (Bruch, 1982; Oldershaw et al., 2015; Robinson et al., 2015).

Limitations

This research has several limitations. Recruitment was curtailed by the COVID-19 pandemic which resulted in a relatively small but acceptable sample size compared to other reflexive thematic analyses (e.g., Landrum & Davis, 2023). Arguably, saturation was still reached as all subthemes were represented in the first two transcripts alone, with subsequent interviews elaborating on rather than introducing new themes. However, future work should aim to include larger and more diverse samples in order to capture a more complete view of AV experiences. In particular, greater diversity in terms of participant gender, ethnicity, cultural background, age, AN duration, and comorbidity is needed. It is also important to acknowledge the range in BMI and treatment experiences in our sample. Whether the AV is impacted by different forms of care in which inner voices and their externalization are given more or less attention is unclear. Future work should aim to explore the AV with an eye to potential interactions with the types and length of treatment individuals have received. Further credibility checks, such as participant validation or independent coding would also improve the trustworthiness of this research. Finally, data for this study centered on information conveyed by the voice (phase two of the DIS-ED). While this is a novel research direction, incorporating participants’ reflections on the interview (phase four of the DIS-ED) might have provided a more comprehensive understanding of these experiences.

Future research

This is the first study to investigate EDVs through direct conversation rather than indirect methods (e.g., symptom measurement). Further studies are needed to confirm the feasibility and acceptability of dialogical enquiry as a research tool and method for data collection.

Voice dialogue and dialogical enquiry treat inner voices as psychosocially meaningful events (Longden et al., 2021), thus offering researchers and participants a new source of information and ensuring that aspects of experience that might otherwise be discounted or ignored are attended to (Hartman, 2015; Schnackenberg et al., 2018). Accordingly, dialogical enquiry may provide insights into the nature of other distressing experiences where voices feature, including depression (Moritz et al., 2019), bipolar disorder (Smith et al., 2023), obsessive compulsive disorder (Gangdev, 2002), and post-traumatic distress disorder (Anketell et al., 2010).

Research is also needed to develop a more comprehensive understanding of the role of internal voices in AN and other EDs such as bulimia nervosa (Pugh et al., 2018), and to translate new insights into improved treatments. Samples from a range of cultures, ethnicity and gender identities are needed. The current study included individuals at different stages of treatment for their AN. Longitudinal research could examine whether changes in EDV content or strength precede or follow changes in ED behaviors, identifying warning signs for deterioration. Given that AVs share similarities with more common hostile inner voices such as the “inner critic” (Stone & Stone, 1993) or “punitive mode” (van Maarschalkerweerd et al., 2021), dialogical enquiry could explore points of convergence and divergence. Crucially, researchers should investigate ways of working with EDVs to promote recovery from EDs, monitoring the costs and benefits of different approaches and comparing their efficacy against that of current treatments.

Implications for clinical practice

Exploring the content, functions, and origins of AV could assist clinicians with case conceptualization and tailored interventions for AN, helping identify key beliefs, thinking styles, or memories associated with disordered eating. For instance, if the AV offers an escape from challenging emotions, treatment might focus on improving emotion identification, regulation and tolerance (Adamson et al., 2018; Lynch et al., 2013). Alternatively, reliance on the AV for establishing a sense of identity could be reduced by building a positive identity beyond AN, such as in MANTRA, which is an evidence-based therapy for AN (Schmidt et al., 2014, 2015). Equally, this study highlights the role of the AV-individual relationship in the maintenance of AN and how this may be likened to an abusive and/or punitive interpersonal dynamic (Tierney & Fox, 2011). Helping individuals develop new ways of relating to the EDV (e.g., assertive relating) may help to improve outcomes (Pugh et al., 2023). While we present a preliminary investigation into the AV’s perspective, some of the themes identified are novel, such as the concept of a pragmatic AV. While previous work has highlighted the kinds of problems EDs claim to solve, our work, in asking the AV itself, uncovered the tailored and specific way in which the AV does so. Indeed, by understanding the AV’s motivations and listening to it in its own language, clinicians may have the opportunity to target its distressing and/or unhelpful messages more effectively. For example, pointing to the paradox in the voice’s defensiveness and destructiveness, as shown in our findings, may only be possible when in direct conversation with the voice. Those experiencing highly critical and distressing AVs may benefit from dialogical interventions involving their voice(s) (e.g., Cardi et al., 2022; Chua et al., 2022; Hayward et al., 2017; Heriot-Maitland et al., 2019; Hibbs et al., 2021). Depending on the outcomes of future research, conversations with AVs could form the basis for new understandings and treatments for EDs, as well as an improved understanding of these experiences.

Conclusion

This thematic analysis of dialogues with the AV maps out an interdependent relationship: the voice considers its pragmatism indispensable to the person, engages them in a fraught relationship, and becomes an increasingly controlling presence in their life. The findings complement existing research but draw out the interdependence of the AV-individual relationship, the specifics of its role as a valued problem solver, and the methods by which it extends its influence. Evidently, the AV can present as a developed character that an external person can converse with. Ongoing research should investigate how to utilize this knowledge to promote recovery. For now, clinicians’ awareness and curiosity regarding the nature and impact of EDVs may help to identify barriers to recovery and tailor treatments.

Disclosure statement

The authors report there are no competing interests to declare.

Data availability statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy and ethical restrictions.

Additional information

Funding

Funding for this study was provided by the UCL Clinical Psychology Doctoral Training Course.