PROJECT SUMMARY Schizotypal personality disorder (SPD) is similar to schizophrenia (SZ), but with fewer and attenuated abnormalities, thus representing an important yet understudied intermediate SZ-spectrum phenotype. Examination of abnormalities in SPD will provide information regarding etiology, genetics, treatment and risk factors associated with psychosis. Although individuals with SPD demonstrate marked temporal lobe abnormalities that resemble SZ, we hypothesize that relative “sparing” or “functional enhancement” in the frontal lobes (e.g., dorsolateral prefrontal cortex), may protect these individuals from frank psychosis and the severe social and cognitive deficits typically observed in SZ. Studying SPD is powerful as antipsychotic medication and hospitalization confounds observed in SZ are not present. Moreover, there is no study examining neurobiological changes in the SZ-spectrum that incorporates individuals with SPD using a longitudinal design as proposed here. This novel approach will help disentangle potential risk and protective factors for psychosis in the SZ spectrum. This is the first longitudinal study to utilize multimodal MR imaging and Research Domain Criteria (RDoC) approaches in SZ-spectrum disorders to identify aberrant neural circuitry along a continuum from healthy controls (HCs) to SPD to SZ and examine changes in these measures in relationship to impairments in symptom severity, neurocognition and functional outcome. We propose studying three groups (80 in each) of demographically matched and rigorously diagnosed individuals (age 18- 40): HCs (no Axis I or personality disorder), unmedicated individuals with SPD (and no Axis I disorder), and early-onset (first 2 years of illness) SZ patients at baseline, 9-, and 18-month follow-up. Measures assessing frontal and temporal lobe integrity include multimodal MR imaging (structural MRI, DTI, resting-state fMRI, and task-based fMRI with a nonverbal event related working-memory task; baseline and 18-months) and neuropsychological assessment (all three timepoints). We will utilize dynamic causal modeling to test competing neurobiological models involving abnormal frontotemporal connectivity in the SZ-spectrum and machine learning approaches to integrate multimodal neuroimaging, neurocognitive, and clinical assessment data. We focus on three specific aims: (1) Investigate the longitudinal course of frontal-temporal lobe/circuitry abnormalities in the SZ-spectrum using multimodal MR imaging; (2) Investigate the longitudinal course of neurocognition, clinical, and functional outcome in the SZ spectrum; (3) Determine which factor or combination of factors differentiate groups in the SZ-spectrum to identify those that are associated with risk for and protection from SZ using machine learning.