The INTRUDE trial aims at assessing the efficacy of an fMRI-based neurofeedback procedure on drug-resistant auditory hallucinations. Hallucinations are complex and transient mental states associated with subtle and brain-wide patterns of activity for which we were recently able to validate an fMRI multivariate decoder. Based on this progress, we can track patients' hallucinatory status using real-time fMRI. We will test whether schizophrenia patients with drug-resistant hallucinations can be trained to maintain the brain state associated with a no-hallucination condition using appropriate strategies and thus reduce overall severity. We will refer to a double-blind randomized placebo-controlled design. A total of 86 patients will be enrolled and equally split in an active neurofeedback group (n=43) and a sham group (n=43), matched for sex, age and PANSS scores. Each patient will benefit from 4 runs of either active or sham neurofeedback. The primary outcome measure will be the mean decrease of AHRS scores relative to baseline, and at 1 month post-treatment. We expect significant clinical benefits from fMRI-based neurofeedback on drug-resistant hallucinations compared with the sham group.

Inclusion Criteria:

• Schizophrenia (according to the DSM-5 classification)
• Frequent auditory hallucinations (SAPS item #1 ≥ 4)
• Stable medication for at least 30 days
• Absence of chronic neurological disorder (including seizure)
• Able to provide free written consent to participate in the research

Exclusion Criteria:

• Pregnancy
• Contraindication to MRI scan
• Claustrophobia
• No social insurance