A Young and Promising Therapy for Hallucinations

John Watson


April 18, 2018

Quieting an Often-Painful Inner Voice

For the estimated 25% of patients taking pharmacologic therapy for psychotic conditions who still experience auditory verbal hallucinations,[1] these internal voices can feel like real-world antagonists whose words are stinging and ever-present. The researchers behind AVATAR therapy are hoping that by giving these voices digital form and allowing therapists to embody them, they'll give patients greater control over their hallucinations.

Results from a recent trial[2] indicate that this strategy may indeed hold promise. Comparing patients with a previous diagnosis of schizophrenia spectrum or affective disorder (mean length of illness, 10.8 years) randomly assigned to receive either supportive counseling (n = 75) or AVATAR therapy (n = 75), researchers found that the latter experienced significantly greater reductions in auditory verbal hallucinations at 12 weeks. These effects were sustained for up to 24 weeks.

Medscape spoke with principal investigator Thomas K.J. Craig, PhD, FRCPsych, emeritus professor of social psychiatry at the Institute of Psychiatry, Psychology & Neuroscience at King's College, London, to discuss AVATAR's digital innovations and the surprising insights his study had to offer regarding the role of traditional counseling.

Bringing the Empty Chair to Life

Medscape: What was the original hypothesis that led to AVATAR therapy?

Craig: For some time, therapists have been experimenting with voices dialogue. This idea is sometimes referred to as the "empty chair" approach, where the therapist is in one chair and the patient in another, with the person's voice that they hear represented by an empty chair. The therapist attempts to set up a kind of conversation between the patient and the voices he or she hears as coming from the empty chair, asking questions like, "What do you say? Why do you say it? Why are you tormenting them?" Then the patient will attempt to answer. That's been around for a number of years.

My colleague, Professor Julian Leff, who is now retired, was interested in this therapy and had the idea that instead of an empty chair, you could have the person have a conversation with a simulacrum of the voice—an avatar. He approached Mark Huckvale, PhD, a professor of speech and language, and together they devised and developed AVATAR therapy, publishing positive results from the pilot study in 2013.[3] I then got intrigued, and we took it on from there.

Inner Voices Get a Face

Medscape: How do you go about creating the avatar patients interact with?

Craig: The therapy takes place over seven sessions. The first session is all about creating the avatar, which you do by taking as detailed an account as possible from the patient of what they hear, such as whether they hear one voice or many voices. If it's many voices, it's then about selecting the voice that is the most important or that they would most like to get rid of.

You then ask things like: If the voice is part of an organization or a gang, is it someone they know from their past? All of that information helps you get an idea of what this voice means in their life.

You and the patient then sit in front of a computer to create the avatar. On the computer are a set of transforms that can turn the therapist's prerecorded voice into a variety of alternatives. For example, if I'm seeing a patient who says that he hears a woman's voice calling him a fool, my voice can be turned into that of a woman. We can change the voice even further to make it older, younger, higher pitch, lower pitch, more or less breathy until eventually, after fiddling about with this system, the patient says that it's a very good match to what they hear.

Then we do a similar process with the face. In this example, we'd have a basic female face and can change the hairstyle and color, the eyes, the expression of the mouth, the cheeks, the color of the skin, and so on. We go through iterations to come up with something the person feels reflects what they imagine the thing talking to them looks like. It doesn't have to be human, though in fact most are; it can be angels, devils, whatever the person thinks is talking to them.

Clues to Recovery

Medscape: I'm picturing this almost like the work of a police sketch artist.

Craig: Absolutely. The face creation looks and, I think, feels a lot like that. Just like in the police sketch, you might start and say, "No, I don't think you quite got the mouth right." You get that kind of response creating the avatar as well.

Just like a police sketch might not be an absolute likeness of the criminal, but it's enough to give the police something to go on, it's much the same thing in the therapy.

Medscape: Does the process of building the avatars offer practitioners information of therapeutic value that they may not have otherwise had?

Craig: Yes, definitely. At the very least, some of the patients said to us, "You're the first person that's heard what I've heard and sort of shared the experience," which is just not possible in an ordinary therapy situation.

The process helps somehow. The people are edging towards something that, when it crystallizes, has a lot of meaning for them.

At a more fundamental level, for some patients—though certainly not the majority—the creation of the avatar changed their perception of the voice. Having a face that you could talk to and relate to in a way that you can't when it's disembodied was, I think, therapeutic.

There were also a few people who told us at the beginning that they really didn't know who the entity was that was talking to them. Some created a face that looked surprisingly like themselves, which itself is quite an insight. You can then say to them, "This thing that you say is talking to you looks a bit like you, doesn't it?" Others created someone who helped them dredge up the memory of some event that they believe was very important in the origin of their voices. Those are the sort of things you wouldn't normally get in a therapy situation.

AVATAR in Practice

Medscape: What happens after you've put together the face and voice?

Craig: The therapy begins a week later. In that week's gap, we ask the participant to keep note of anything their voices usually say, and we then use those statements verbatim as a guide to where to begin the therapy.

The patient and the therapist then sit in separate rooms linked by computer. There's a small video feed on the patient's computer, so the therapist can see them at all times. The therapist is able to switch between speaking to the patient either in his or her own voice or in the voice of the avatar. When talking as the avatar, what the person sees is the face on the screen that they created with lip sync, its eyes blinking, those sort of basic facial expressions.

By using verbatim the things that the person usually hears, only now with the face they've created, we try to provide as realistic and immersive an experience as possible. It's pretty good. Most patients say they're 80%-90% convinced that they're really talking to their voices. Because of the way it's set up, they know the therapist is behind the voice. But like any good computer game, within a very short period of time the person is immersed in the conversation and feels as though they are talking to their own voices.

As therapy proceeds, the avatar might say something, and in the early stages the therapist might come in and give advice on how to confront and challenge the voice. The idea is that as the person becomes more receptive, challenging what the voice says, the avatar's demeanor and responses change to be more conciliatory. This creates the impression that the person is beginning to take more control over the conversation, taking some of the power back.

From that experience and learning more about the voice's purpose in the individual's life, it's possible to set up a dialogue that actually goes a bit deeper than just rebutting what the avatar has been saying but actually goes back to maybe some of the experiences and psychological sources of where the voices came from in the person's life.

For example, quite often we had people who had been very traumatized earlier in their life, and they would be confronting, in the form of the avatar, the person or the entity that's traumatized them. You'd be able to do something in that scenario that got the avatar to apologize for being horrible in the first place or explaining why it does the things it does. A very simple example of that would be the avatar saying to the person, "I only say those bad things because you think them of yourself." That opens up a dialogue about why they are thinking these things about themselves. It becomes a bit like any therapy of acceptance, but it is done by the therapist through the avatar.

In the trial, this went on for six sessions in order to replicate the pilot study, but it was quite clear that some people would have benefited from more sessions.

Digital Innovation Reveals Traditional Therapy's Benefits

Medscape: At 6 months, AVATAR and the control arm seemed to produce similar results. Does AVATAR's impact diminish?

Craig: What is happening is much more mysterious than that. The AVATAR group improves early and then maintains it, whereas the control group continues to improve, so that at the 6-month point, the difference between the interventions is no longer statistically significant.

However, it's not because the AVATAR impact wears off. It's because something odd is happening in the control group, and we really are struggling to understand what is driving that change. We really needed and didn't have a third arm of no treatment to be able to unpack what was going on.

Everybody who reads the paper naturally assumes it's like other trials of this kind where you get a kind of erosion of the benefit from the target therapy. There isn't an erosion; in fact, it's retained. We just don't have a good enough trial design to tease out what's going on.

That's what we certainly hope we will get in future studies and move to the next phase. We're currently trying to persuade funders to pick up another study, a three-arm study this time.

Medscape: AVATAR isn't approved for clinical use yet, but are there lessons here for practitioners seeing these patients now?

Craig: When I trained in psychiatry, and I think this was true until quite recently, we were taught not to bother too much with the content of hallucinations but instead look at the form. The quantity and type of delusions, whether hallucinations were in the third person—all of these things came together to make a diagnosis, but content wasn't that important.

It now turns out that the content is critically important in a way that we didn't appreciate before. It's what the person is experiencing that produces the functional impairments. This is not news to people who practice psychological therapies for psychosis, but maybe for the jobbing clinician it is. Even if at the end of the day you still reach for that prescription pad, I think the understanding and the sharing of the experience are very powerful.

If you want to write an extraordinary story, it's of our control group. They had young therapists providing a friendly ear with simple advice on how to get by from day to day but with no particular high expertise. Doing that for 40 minutes once a week produced a result at 6 months equivalent to AVATAR. Both AVATAR and the control group had that in common: therapists who were interested in the people and their stories and listened to them. Maybe that doesn't sound like news, but I can tell you: The way we practice psychiatry, it's a lesson for us.

Medscape: It's fascinating to see that effect can take hold no matter whether this is a flesh-and-blood interaction or a digital one.

Craig: The thing that was really staggeringly impressive is that you must remember that our study was done with people with decades of psychosis, an average duration of 10 years. Everybody was on regular medication, with more than a third on clozapine. All had voices continuously for 12 months before they entered the trial.

In that population, we have 12 people who stopped hearing voices completely within six sessions. That itself is staggering. You only get that kind of effect in a young first-episode psychotic who is medicine naive, and you're treating them with pharmacology for the first time. That just does not happen with pharmacologic manipulations in the population we're working with.

Whatever AVATAR is doing and whether it's unique to AVATAR or it's something that's in common with just listening and talking, it's more potent than any other intervention, and I can say this with confidence. If you take the meta-analyses of cognitive therapies for voices, AVATAR is about twice as powerful in terms of its effect size compared with a year's worth of cognitive behavioral therapy.

That's really something. I think this therapy has really got legs.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: