Sensing things that are not there, whether it is sounds, smells, or visions can be a disorienting and scary experience. These hallucinations can occur as part of a complex condition, such as schizophrenia, or as one-time events. But hallucinations are notoriously difficult to study, partly because of how hard it is to recreate and study them in lab animals. To address that, a group of researchers from Cold Spring Harbor Laboratory and the Washington University School of Medicine developed a way of invoking hallucinations in mice, and correlating their observations to humans. The researchers published their results earlier this month in the journal Science.

The team first trained mice to poke a button. If the mice could hear a certain sound played after pressing the button, they got a reward. The researchers found that about 16 percent of the mice would poke the button and confidently "hear" the sound, even no sound played. If that confidence matched the confidence of a real poke, they defined it as a hallucination. Those hallucinations increased with exposure to drugs such as ketamine. In humans, the researchers employed a similar task, asking online participants to press a key if they heard a sound, and report their confidence that they had heard it. 

Based on that, the team was then able to investigate more deeply into how exactly these hallucinations were happening with a computational model that simulated aspects of the trials, and found that hallucinations happened when the expectations were greater than the sensory inputs. They then confirmed this in mice, showing that both the expectation of a reward and the expectation of hearing something changed levels of dopamine in key areas of the brain.

This new approach to cause mice to have hallucinations opens doors to better understand hallucinations in different diseases. It may also inspire new treatments for silencing auditory hallucinations.