At his psychiatric clinic in the Connecticut Mental Health Center, Albert Powers sees people every day who experience hallucinations. The condition is often a hallmark of psychosis, occurring in an estimated 60 to 70 percent of people with schizophrenia, and in a subset of those diagnosed with bipolar disorder, dementia and major depression. Auditory hallucinations are the most common type experienced. Some patients report hearing voices; others hear phantom melodies. But increasing evidence over the past two decades suggests hearing imaginary sounds is not always a sign of mental illness.

Healthy people also experience hallucinations. Drugs, sleep deprivation and migraines can often trigger the illusion of sounds or sights that are not there. Even in the absence of these predisposing factors, approximately one in 20 people hear voices or see visual hallucinations at least once in their lifetimes, according to mental health surveys conducted by the World Health Organization. Whereas most researchers have focused on the brain abnormalities that occur in people suffering at an extreme end of this spectrum, Powers and his colleagues have turned their attention to milder cases in a new study. “We wanted to understand what’s common and what’s protecting people who hallucinate but who don’t require psychological intervention,” he says.

Normally when the brain receives sensory information, such as sound, it actively works to fill in information to make sense of what it hears—its location, volume and other details. “The brain is a predictive machine,” explains Anissa Abi-Dargham, a psychiatrist at Stony Brook University School of Medicine, who was not involved in the new work. “It is constantly scanning the environment and relying on previous knowledge to fill in the gaps [in] what we perceive.” Because our expectations are usually accurate, the system generally works well. For example, we are able to hear the sound of running water or the murmur of a friend talking across the room and then react in an instant, Abi-Dargham says.

One theory posits hallucinations arise when the brain relies too strongly on these expectations, filling in details even when an actual auditory input does not exist. Culture and religion may also play a role in interpreting what individuals perceive, and whether the voices they hear are helpful or disruptive. To test the idea that hallucinations are the result of an over-expectant brain, Powers and fellow Yale University psychologist Philip Corlett decided to study a diverse group of people who reported hearing voices on a regular basis—including those who had been diagnosed with psychosis, along with self-identified psychics who had not been diagnosed with any psychiatric illness.

The team visited a local Connecticut organization for psychics and began interviewing people. They vetted individuals using forensic psychiatry techniques to ensure that people were not simply pretending to experience auditory hallucinations. Almost immediately the two noticed that the psychics’ descriptions of hearing voices were remarkably similar to the experiences of their patients diagnosed with psychosis. “They sounded the same in terms of how loud the voices they heard were, the frequency of occurrence, where they were hearing them in space—within or outside their heads—and length and complexity of what the voices uttered,” Powers says.

Next the researchers designed a series of experiments to introduce new beliefs about sensory information. The team introduced this new information—in the form of a Pavlovian learning task—to the psychics, patients diagnosed with psychosis, and others in a control group who had not heard voices before. The latter group included both people who had been diagnosed with psychosis and healthy adults. They paired a visual stimulus of a checkerboard on a computer screen with a brief 1-kilohertz tone, presenting the light and sound repeatedly until participants learned to associate the two. Then they measured how much people relied on this prior sensory knowledge when shown the visual stimulus without the sound.

At first, at least some members of all of the groups heard the sound even when it wasn’t there. But the researchers found that both the psychics and people who were prone to psychosis were more likely to hear the tone when none was presented than were those who did not hear voices. The two voice-hearing groups were also much more confident in their assertion that the sound had occurred. Powers and Corlett took these reports to mean these groups had developed extremely strong beliefs that the visual cues were associated with tones. Their prior belief that a tone was always accompanied by a sound was driving the auditory hallucination.

When the researchers performed additional no-tone trials, however, the psychics and the group of healthy adults who did not hear voices were able to revise their beliefs about the association, or lack thereof, between the checkerboard and the tone. But those in the study who had been diagnosed with a psychotic illness—both voice-hearers and nonhearers—were unable to detect that the tone was no longer present. “The results fit quite nicely with what we observe clinically on a daily basis here in the [Connecticut Mental Health] Center,” Corlett says. “People with a psychotic illness find their perceptions really difficult to abandon even when everybody around [them] agrees that what [they] are hearing is not actually happening.” Functional magnetic resonance imaging revealed that those who had trouble updating their cognitive beliefs had less neural activity in their parahippocampal gyri and cerebellums, regions associated with memory formation and making predictions about one's own body.

The findings, published today in Science, provide insight into a common neural mechanism that may drive auditory hallucinations as well as what may make these experiences more debilitating in some people. “This study lends support to the idea that there is some sort of continuum from mental illness to health,” Abi-Dargham says. Researchers may be able to use these insights to guide the development of new therapies—whether drugs or brain stimulation (such as transcranial magnetic stimulation) that targets regions most affected in patients with schizophrenia and other disorders, she says. 

Although it may take awhile before such therapies are ready for clinical use, Powers and Corlett remain cautiously hopeful that they can still learn a lot about how the brain works by looking at the biggest difference between the patients with psychosis and the psychics: specifically, how a change in beliefs can affect perceptions. They liken the phenomenon to the placebo effect whereby people who believe a pill will work see an automatic alleviation of their symptoms. “The power of the mind over itself is amazing,” Powers says. “We’re only just beginning to understand the biology behind that.”