Mentally well voice-hearers have a heightened ability to detect real speech

Hallucinating voices isn't always distressing. While the experience is commonly associated with schizophrenia, some people – an estimated 5 to 15 per cent of the general population – hear voices that aren't real without finding it upsetting or debilitating (they may even welcome it) and in the absence of any of the other symptoms of psychosis, such as delusions or confusion.

Now new open-access research published in Brain has revealed a perceptual advantage for this group of people: they can detect hard-to-comprehend speech sounds more quickly and easily than people who have never hallucinated a voice.

The study, led by Ben Alderson-Day at the University of Durham, involved twelve "non-clinical voice hearers" and 17 controls (matched for age, sex, handedness, education, and National Adult Reading Test scores) who do not hear voices. While their brains were scanned using fMRI, the volunteers listened to a series of so-called "sine-wave" speech recordings.

These recordings of spoken sentences – such as "The clown had a funny face" – were acoustically altered so that they were either totally unintelligible, or potentially intelligible. Initially, many people report that potentially intelligible sine-wave speech sounds like "aliens" or birdsong. But with training, it's possible to recognise and understand the distorted words.

During the first part of the study, the volunteers had no idea that it was about speech perception. They were asked simply to press a button every time they heard a distinctive target sound. It was only after 20 minutes that they were asked if they'd noticed any words or sentences among the recordings they'd just listened to, and if so, to indicate when they had first noticed them and whether they had been able to understand any of them.

Nine out of 12 voice-hearers (75 per cent) said they'd realised words were contained in the sounds, compared with eight (47 per cent) of the non voice-hearers, and the voice-hearers noticed the words earlier, on average. Seven of the voice-hearers and five of the controls said they could understand the words, and five in each group could accurately recall some of them.

After this first 20-minute run, the volunteers were trained in how to listen for words in sine-wave speech. Then the trial was repeated, and, afterwards they were again queried about what they'd heard. For both groups, the training helped, but this time, the voice-hearers' performance was no better than the controls.

Other work on people in the early stages of psychosis has found that they are better at identifying objects in an ambiguous visual scene than healthy controls, if they are primed about what to look for. Prior expectation seems to give them a superior ability – compared with controls – to spot genuine signals in noise.

In this study, being explicitly primed to listen for speech didn't benefit the voice-hearers any more than the controls, so something different seems to be going on compared with the detection of visual patterns in psychosis.

The fMRI scans showed that, like the controls, the voice-hearers' brains responded differently to potentially intelligible versus unintelligible sine-wave-speech. This suggests that voice-hearers aren't biased to hear speech in any sound, but only when there is the possibility of a meaningful signal being present.

The scans also revealed a key difference between the brain activity of the voice-hearers and the controls. In both, the potentially intelligible recordings engaged the brain network typically active in normal voice processing, but the voice-hearers showed stronger responses to this kind of speech in two regions: the dorsal anterior cingulate cortex and the superior frontal gyrus (responses that were not seen when the recording was unintelligible).

"This suggests an enhanced involvement of attention and sensorimotor processes, selectively when speech was potentially intelligible," the research team writes. "This suggests that the fundamental mechanisms underlying hallucination involve – and may develop from – ordinary perceptual processes, illustrating the continuity of mundane and unusual experience."

This has implications for understanding the significant number of people who hallucinate voices but do not need psychiatric care. It also feeds in to the idea that, within the general population, there's a continuum of experiences normally associated – in their extremes – with psychosis.