Skip to content

When you choose to publish with PLOS, your research makes an impact. Make your work accessible to all, without restrictions, and accelerate scientific discovery with options like preprints and published peer review that make your work more Open.

PLOS BLOGS The Official PLOS Blog

A Social Visit with Hallucinated Voices by Vaughan Bell

By Vaughan Bell

I’ve met a lot of people who hear hallucinated voices and I have always been struck by the number of people who feel accompanied by them, as if they were distinct and distinguishable personalities. Some experience their hallucinated entourage as hecklers or domineering bullies, some as curious and opaque narrators, others as helpful guardians, but most of the time, the voice hearer feels they share a relationship with a series of internal vocal individuals. Not everyone who hears voices experiences them as social entities but this type of social hallucinated voice is not rare or exotic.

talking trees

Studies show that the majority of voice hearers experience their voices as individuals who can be distinguished by the characteristics of their speech and even their personalities. For some, this means the voice is experienced as a specific person that others are aware of. It may be someone they know (perhaps a family member), a famous person, a religious figure, or even a fictional character from popular culture.

For others, the voices have an identity named solely by that person (“Jeremy, a little boy”), and sometimes, the voice is unnamed and just recognised by its personal characteristics like the “unknown old woman” or “a man with a deep voice”, as reported in a 1997 study by Leudar et al. Surprisingly, scientific theories of how auditory verbal hallucinations are formed have rarely attempted to explain this social, even personal, aspect of the experience.

Currently, the main scientific theories of voice hearing suggest that hallucinated voices occur through a combination of atypical activity in the language and memory systems of the brain and a tendency to attribute internally generated mental phenomena as coming from an external source. The idea is that voice hearers think phrases to themselves as part of their internal monologue but hear them as coming from outside due to problems with adequately distinguishing internally from externally generated experiences. There is lots of evidence from experimental studies to back up these theories, but they don’t really address the issue of how voices are perceived as social identities.

 So how do internal thoughts become experienced as other people?

It seems not many people are interested. A recent paper, published in 2012, aimed to synthesise all the existing research to give an up-to-date cognitive model of how hallucinated voices occur, and had very little to say on how voices come to be perceived as social identities. This is most of it: “The content of AH [auditory hallucinations] may be determined by factors such as perceptual expectations, mental imagery, and prior experience/knowledge (e.g. memories) that shape a perception of reality that is idiosyncratic and highly personalized.” Considering that the personal nature of hallucinated voices is a central and defining feature for most voice hearers, this is a striking omission for a causal theory.

The paper I wrote for PLOS Biology, “A Community of One: Social Cognition and Auditory Verbal Hallucinations” aimed to encourage a better understanding of this issue for hallucinated voices by taking the aspect of social experience seriously and looking at the current evidence that would support a social cognitive and social neurocognitive theory for voice hearing.

As I point out in the paper, clinical psychologists in particular have looked a great deal at people’s social experience of voices.

Numerous studies have now found that voice hearers 
understand their connection with the voices in terms 
of relationships and interact with their voices in 
ways that “share many properties with 
interpersonal relationships 
within the social world” [6]. 
Most obvious in this regard is the fact that 
over 80% of people who experience auditory 
verbal hallucinations have reported that they are 
able to engage in interactive conversations 
with their voices [7],[8]. 

Judgements about the identity of hallucinated voices 
rely on perceptual features similar to those required 
to judge identity when listening to the voices of 
other speakers, with perceived 
identity being an important mediator of 
distress [11].

But these findings haven’t been well integrated into research on auditory hallucination formation completed by experimental psychologists and neuroscientists who have tended to focus on individualistic, information processing theories. However, if you look at which neural circuits turn up in brain imaging studies, and how people perceive their voices in psychology studies, which is the evidence I review in detail in the paper, there is a good case for making social processing in the mind and brain a central part of understanding voices. I also suggest that one place to start making sense of this data could be in how we generate and use internal models of other ‘social actors’ when we’re thinking and reasoning about social situations.

Essentially, we spend a lot of time thinking about how certain people might react in certain situations, what they might say, and what they might do, even when they’re not present.

Here, again, from my PLOS Biology article:

It would be most parsimonious to assume that these 
phenomena stem from our normal ability to internalise
models of people we know and their voices, rather than 
auditory hallucinations involving a de novo 
generation of persistent and internally vocal 
social identities. Accounts including internalised
models of social actors suggest that we internalise 
others' voices and personalities so that we can 
predict what someone would say or do in any given 
situation [30]. These internal models can be for 
specific people, so I can imagine how my spouse 
might respond in a hypothetical conversation, or for 
generic stereotypes, so I can imagine how a 
policeman or shopkeeper
might respond.

A Different Approach

I argue it’s more likely that the content of hallucinated voices comes from a normal ability to internalise models of people we know and their voices, rather than involving a separate step when we re-interpret our own thoughts as being from a range of individual characters based on some vague notion of ‘perceptual expectations’.

In the (PLOS Biology) article, I also make a series of predictions as to what this approach would entail. If you’re not familiar with science, making a series of predictions is research talk for ‘come and have a go if you think you’re hard enough’ based on the collective belief that reality ends up kicking everyone’s ass.

As a result, I’ve begun working with people who want to knock the ideas into shape or who are interested in stress testing similar concepts. I’ve been working on the role of agent representation – essentially how we make sense of other autonomous beings in the world and how we understand them in terms of their choices, actions and mental states – and how it applies to hallucinations, with philosopher of mind Sam Wilkinson. We’re just finishing a paper so hopefully we can continue the debate and spark some new approaches to hallucinations. We’re also working on the scientific implications with two cognitive neuroscientists who are much cleverer (and harder) than me, but we’re still getting our first draft of the paper down, so I’ll wait to we’ve got the details hammered out to say a bit more.

Finally, there is something important to note in how causal models of hallucinated voices have tended to ignore content and personal significance as irrelevant, while clinical models have tended to ignore neurocognition as inconsequential.

This is a trend present to varying extents throughout psychopathology research and it tends to distance lived experience from an integrated scientific approach to an experience that needs to be better addressed when distressing or disabling, and better understood as part of human nature.

We need to develop an understanding of difficult experiences that span mind and brain and ensure that clinicians, psychologists and neuroscientists talk to each other more than they presently do.


Vaughan Bell is a neuropsychologist at Kings College London and a clinical psychologist at the Maudsley Hospital. He blogs at MindHacks and serves as a PLOS ONE Academic Editor. On Twitter @VaughanBell



Click here to listen to “Hallucinations and Designer Drugs,” an August 2013 Mind the Brain Podcast interview with Vaughan Bell, conducted by former PLOS Biology Senior Editor Ruchir Shah.

  1. How do you see these findings and your proposal that we take the social-personological aspects of voices into account meshing with trauma models of voice hearing, such as those championed by people involved with Intervoice? If you see them doing so at all.

    In relation to that, I wonder what if there is any kind of feedback effect. That is, if its the case that voices are similar to our ability to model other social agents then to what extent are our models of those social agents anything but delusions themselves, albeit ones that are pragmatically validated within the sphere of coordinated social action?

  2. There are potential avenues to explore in light of trauma models of voice hearing – not least how our tendency to represent agents might be affected by intensely negative experiences of other agents (in this case, people) perhaps during the period when our ability to represent and incorporate agents into our social cognitive structures is still developing. However, as we really don’t have enough strong evidence to fully test or even elaborate these models as yet, this is an open hypothesis.

    In the original PLOS Biology paper I noted research suggesting a link between early relationship trauma and auditory hallucinations. Although this finding has been replicated several times a recent large population study by van Nierop et al. (doi: 10.1093/schbul/sbt150) did not find a link to specific symptoms but did find a strong link between hallucinations and delusions in general and trauma involving the ‘intention to harm’ (deliberate abuse). The fact that there is a very well replicated link between hallucinations and delusions (mostly which involve atypical experiences of agency) and experience of trauma from agents – rather than non-agents, suggests that there is a good basis for further research in this area.

    As to the question of whether “our models of those social agents anything but delusions themselves” I’d probably reframe the question. As models, they are mental tools and so designed to be widely applicable and reality is not a criterion for their validation. For example, we can represent fictional agents (reading a novel requires this) so the question would be not whether they are ‘delusions’ but how they work to allow us to apply them across a range of contexts.

  3. Thanks for this! I work with survivors of ALL forms of sexual trauma, although a high percentage of our clients also have early childhood sexual abuse, in addition to physical abuse, and neglect.

    From what I see, and some reading I have done, neglect and severe neglect are most closely associated with auditory hallucinations.

    In the clients I see, I encourage them to interact with their voices, to be curious about them rather than afraid of them. Over time, they feel less buffeted by the voices, and some voices actually become allies. The clients experience the voices as other people (often introjects of people they have known, particularly the abuser), so we work with them that way (unless the client is D.I.D., in which case we try to limit their conceptualization of the voices/parts as separate people and see them, instead, as split off parts of the client’s self).

    Anyway, wanted to say thanks for this.

Leave a Reply

Your email address will not be published. Required fields are marked *

Add your ORCID here. (e.g. 0000-0002-7299-680X)

Back to top