Alternative Facts and Realities: How the Brain Anticipates Perception

by Jalees Rehman

The Russian physiologist Ivan Pavlov is best known for his studies on classical conditioning showing that dogs repeatedly presented with a combination of food and a sound would subsequently salivate upon hearing the sound alone, in anticipation of the meal. The combination of the two stimuli – food and sound – over time "conditioned" the dogs' brains to link these two stimuli. A variation of this experiment was performed on human subjects by Ellson and published in the Journal of Experimental Psychology in 1941. In Ellson's study, 40 subjects were "conditioned" over time by hearing a sound and seeing a light. Ellson later on exposed the subjects to only the light, yet 32 of 40 subjects claimed to have also heard the sound. Ellson concluded that such conditioning could lead to hallucinations – the hearing of sounds which, objectively speaking, are not present.


Recently, the Yale University psychiatrist Philip Corlett and his colleagues conducted a very interesting variation on this earlier study by asking whether some people are especially vulnerable to having auditory hallucinations induced by conditioning. The researchers recruited four groups of study subjects: 1) Fifteen patients with severe mental illnesses who also regularly heard voices (an auditory hallucination), 2) fifteen patients with severe mental illnesses who did not hear voices, 3) fifteen individuals without any evidence of mental illnesses who also claimed to hear voices and 4) fourteen healthy individuals who did not hear voices. Group 3 consisted of voice-hearing psychics ("clairaudient psychics") who identified themselves as such via their own websites, at psychic meetings, or referrals from other psychics. Another important innovation in Corlett's study was the inclusion of brain imaging studies on all subjects, thus allowing the researchers to study functional brain responses when exposing them to auditory and visual stimuli. The researchers then repeatedly exposed the study subjects to a checkerboard image and 1 kHz tone while they were lying in the brain scanner. The subjects were asked to press one button to indicate that they heard the tone, and a second button if they did not. They were also instructed to press down the button longer, the more confident they were in having heard the tone.

After conditioning the subjects, the researchers then intermittently began to show them images of the checkerboard without playing the tone. As expected, many subjects indicated having heard the tone even when it had not been played. However, patients with severe mental illness and a history of hearing voices (group 1) as well as healthy psychics with a history of hearing voices (group 3) were significantly more likely to wrongly indicate that they had heard the non-existing tone. Members of these two groups were also more confident that their hallucination was actually real, since they pressed down the button for longer. Healthy subjects and patients with mental illness who did not have a history of hearing voices were comparatively more correct in identifying whether or not the tone was present. Importantly, when the researchers repeatedly showed the image without the tone, voice-hearing, mentally ill patients were unable to "update" their beliefs when compared to the other groups, whereas the psychics gradually recognized that the tone was non-existent.

Brain imaging showed that the brain regions which respond to sounds were activated by the auditory hallucinations, meaning the brain perceived the non-existent tone as being real! The researchers also found that the subjects who continued to perceive the non-existent tones and did not update their beliefs – even after repeatedly being exposed to the image without the tone – had reduced activation in the brain areas which help integrate and coordinate predictions and perceptions. This could suggest that suppressed activity in these brain regions may contribute to persistent voice-hearing or other hallucinations.

The study has some important limitations: 1) The sample sizes are quite small and the some of the finds – especially those related to the brain imaging – are often only borderline statistically significant; 2) as is the case in many behavioral brain imaging studies, it is difficult to establish cause and effect – was the reduced activity in the parts of the brain which coordinate and integrate perceptions the cause of the inability to update beliefs or merely a correlation; 3) tones are well-suited for conditioning experiments in the laboratory but in the real world, conditioning may occur via specific words or phrases and this could have greater bearing on understanding voice-hearing.

Future studies could therefore address these issues by increasing the sample size, introducing more complex stimuli such as phrases or words, and perhaps also ask other inter-connected questions. Are individuals who are prone to voice-hearing also more likely to have visual hallucinations and hallucinations of smell? Can magnetic stimulation of selected brain areas perhaps improve the ability to update false beliefs? Does conditioning by political, religious or cultural education also affect us in a way that we perceive our hallucinations? For example, would adherents of an extreme political ideology mis-perceive words uttered by political opponents because their brains are conditioned to expect a narrow spectrum of answers and they are unable to update their beliefs? The study by Corlett and colleagues provides some very interesting insights into how our brain anticipates reality due to conditioning which could have important implications for understanding the nature of hallucinations in mental illness and develop new targeted therapies. However, it raises even more fascinating questions which – if addressed in future studies – could have an even broader impact on understanding our perception.


AR Powers, C Mathys, PR Corlett. (2017). Pavlovian conditioning–induced hallucinations result from overweighting of perceptual priors Science, 357 (6351):596-600; DOI: 10.1126/science.aan3458