This is a resubmission application, originally in response to RFA-MH-16-406 and consists of R61 and R33 phases. Auditory verbal hallucinations (AH) have long been a hallmark of schizophrenia (SZ) and are one of its major diagnostic features Andreasen and Flaum 1991; DSM-IV). They are difficult to manage with existing treatment options. Here, we propose that neurofeedback aimed to regulate the superior temporal gyrus (STG) activation will not only lead to activation changes in the STG, but also to changes in the default mode network (DMN) (R61), as well as to reductions in AH (R33), and that the brain and clinical changes will be correlated (R33). The theoretical framework for the current proposal is an AH model that assumes that AH result from abnormalities in a network of regions including STG, and medial prefrontal cortex (MPFC) and posterior cingulate cortex (PCC), the two latter regions are core medial hubs of DMN that are related to self-referential processing. This model is supported by theoretical papers Northoff and Qin 2010, Alderson-Day,2016, and experimental evidence Gur 1995, Liddle 1992, Dierks 1999 as well as our preliminary data (PD). In both R61 and R33 we will study SZ patients with medication resistant AH in the rt-fMRI intervention arm and in the sham-rt-fMRI arm. In both arms, the task and the rt-fMRI session structure will be identical. The SZ-intervention group will receive feedback from the STG while SZ-sham group will receive feedback from the motor cortex. In addition, 2 functional fMRI tasks will examine the effect of rt-fMRI neurofeedback and of sham-rt-fMRI on brain response. In the R61, we will randomly assign 48 SZ patients to either SZ-intervention (n=24) or SZ-sham-rtfMRI (n=24). The STG targeted neurofeedback is predicted to bring changes in brain regions involved in AH (STG and DMN) in SZ-intervention group only. The R61 GO criterion will be BOLD signal reduction in the STG, and resting state connectivity reduction between MPFC-PCC, post rt-fMRI-feedback in SZ-intervention group. In the R33, SZ-intervention group (random n=52) will receive 5 sessions of rt-fMRI feedback targeting STG, while SZ-sham group (random n=52) will receive 5 sham-rt-fMRI sessions. Based on our PD, we predict that rt-fMRI feedback aimed at STG will reduce AH which will be, in turn, associated with reductions in the STG activation and in the DMN connectivity (i.e., brain changes achieved in R61 and replicated in R33) in SZ- intervention group only. Five sessions of rt-fMRI feedback will address the question of dose response at brain and clinical levels. The impact of rt-fMRI neurofeedback and of sham-rt-fMRI on AH (primary outcome), and on delusions, negative symptoms and working memory (WM) (exploratory outcome) will be assessed with clinical and neuropsychological measures. In an exploratory aim, based on the existing literature Garrity 2007; Whitfield-Gabrieli 2009; Rotarska-Jagiela 2010, we predict the improvement in delusions, negative symptoms and in WM score, only post-rt-fMRI neurofeedback targeting the STG and not post-sham-rt-fMRI.

This application proposes to examine whether rt-fMRI neurofeedback directed at the superior temporal gyrus (STG) can reduce auditory hallucinations (AH) by changing activation levels in brain regions implicated in AH in schizophrenia patients (STG), medial prefrontal cortex (MPFC) and posterior cingulate cortex (PCC). In R61, we propose to use rt-fMRI neurofeedback to demonstrate theoretically motivated target engagement of brain regions involved in AH (STG, MPFC, PCC) and improvement in these brain regions' function, post-fMRI neurofeedback. In R33, we will 1.) replicate the R61 results in a larger sample, 2.) show AH reduction, post- neurofeedback, 3.) relate the clinical and neuropsychological changes to brain changes, post-rt-fMRI neurofeedback and 4.) examine neurofeedback dose impact on clinical and brain level changes.