For decades, the dominant approach to distressing voices in schizophrenia has been a combination of antipsychotic medication and, where available, CBT for psychosis. Both help many people. Both leave a substantial group whose voices remain intrusive, frequent, or commanding even at adequate doses. The frustration of that gap is what motivated AVATAR therapy.
AVATAR — short for "Audio Visual Assisted Therapy Aid for Refractory auditory hallucinations" — is a structured talking therapy in which the patient creates a digital representation of the entity they hear, and then has guided conversations with it, with a therapist co-creating the dialogue. It sounds strange. The trial evidence is more promising than that initial reaction would suggest.
AVATAR therapy uses a computer-generated face and voice to give the auditory hallucination a presence the patient can speak with directly, with the goal of reducing distress and shifting the power balance in the relationship with the voice.
Where it came from
The therapy was developed by Julian Leff and colleagues at University College London in the early 2010s. The first published trial appeared in The Lancet Psychiatry in 2018, conducted by Tom Craig and the AVATAR team — a randomised controlled study comparing AVATAR therapy to supportive counselling in 150 people with persistent auditory hallucinations despite ongoing treatment. AVATAR produced a significantly larger reduction in voice frequency and distress at 12 weeks, with effect sizes that compared favourably to the best CBTp results.
The follow-on AVATAR2 trial, results of which were published in 2024, replicated the effect in a larger, more diverse population and tested a longer course. It found that an extended dose strengthened benefits and that the gains held at six months.
How a session actually works
The therapy unfolds over six to twelve weekly sessions, each typically 50 minutes. The first session or two are spent helping the patient describe the voice — its tone, accent, gender, perceived age, the entity's name if it has one — and then collaboratively building a digital avatar that matches it. The therapist uses software that allows real-time facial animation and voice modulation.
In subsequent sessions, the patient sits in front of a screen showing the avatar. The therapist sits in another room (or behind a screen) and voices the avatar, while a second therapist coaches the patient. Crucially, the dialogue is structured so that the avatar starts out hostile (matching what the voice usually says) and gradually concedes, retreats, or shifts as the patient gains confidence in pushing back. The patient, in real time, learns to assert themselves against the voice's content.
The therapy is very deliberate about safety. The therapist has full control of the avatar at all times, can pause instantly, and the dialogues are scripted with the patient in advance.
Why it might work
There are several plausible mechanisms, and they may all matter:
- The patient practices direct response to the voice in a controlled setting, building skills they then carry into daily life.
- Externalising the voice as a visible avatar may shift the patient's relationship with it from "intrusive part of me" to "entity I can engage with from a distance."
- The structured concession by the avatar may produce real corrective emotional experience — the voice loses authority as the patient watches it back down.
- Therapy of any kind for distressing voices, with skilled clinicians, tends to help. Some of the AVATAR effect is the active ingredients of good CBTp.
What the evidence does and doesn't show
The first AVATAR trial showed clear advantages on voice-related distress at 12 weeks compared to active control, with reduction in measured voice frequency. Some of the difference narrowed (without disappearing) by 24 weeks. AVATAR2, the larger study, broadly confirmed the early results and showed that more sessions extend benefit.
What AVATAR does not do, on the current evidence:
- It does not eliminate voices in most patients. The realistic goal is reduced distress and, in some cases, reduced frequency — not silence.
- It does not replace antipsychotic medication. Patients in the trials continued their medication.
- It is not the right approach for active acute psychosis, command hallucinations directing self-harm, or disorganised states. Patient selection matters.
Availability
AVATAR therapy is now available in a small but growing number of NHS Trusts in the UK and a handful of academic centres in the US, Australia, and parts of Europe. It is rarely available in routine community care. Several research groups are developing platform tools that may make it easier for clinicians to deliver elsewhere, but as of writing, access is limited and usually requires referral through specialist psychosis services.
How it compares to CBT for voices
Standard CBT for voices is widely available, has the larger evidence base by total sample size, and addresses the same underlying problem with a different toolkit. AVATAR is best understood not as a replacement for CBTp but as a complementary or alternative approach for patients who have tried CBTp without enough benefit, who find an embodied approach more accessible, or who participate in research.
What patients describe
The brief patient sketches that follow are composite illustrations drawn from published trial accounts and clinical descriptions, not real individuals.
People who have benefited from AVATAR often describe the moment of "talking back" as turning a corner. One trial participant, in an interview published alongside the AVATAR2 results, described it as the first time they had felt able to push back at all — and then realising they could keep doing so when the voice returned in everyday life. Others describe the experience as emotionally hard work, particularly the early sessions when the avatar is hostile, and useful only because the therapist made it safe to stay in the room.
It is not a panacea, and patients are clear about that. Some find no benefit. The published response rates suggest a meaningful minority of patients gain a lot, a larger group gain something, and a smaller group gain little.
If you want to ask about it
- Bring a clear summary of what voices you hear, how often, and what they say.
- Ask whether AVATAR or relational therapy for voices is available locally — many regions have nothing close, but some have research programs.
- If access is unavailable, ask about good standard CBT for voices, which shares some of the active ingredients.
- Be wary of any commercial product claiming to deliver AVATAR therapy without a trained therapist on the other end. The therapeutic relationship is the active ingredient, not the avatar itself.
This article is for educational purposes only and is not medical advice, diagnosis, or treatment. Always consult a qualified mental health professional. If you or someone you know is in crisis, call or text 988 in the US, or your local emergency number.