Genome-wide association studies (GWASs) have identified thousands of variants associated with neuropsychiatric disorders (NPDs), including autism spectrum disorder (ASD), schizophrenia (SCZ), and Alzheimer's disease (AD). However, deciphering the "causal" biological mechanisms and pathways through which these variants act remains a major obstacle that hinders translational understanding of NPD pathogenesis. NPDs are highly polygenic with contributions from pleiotropic variants across the allelic spectrum, most of which reside within large haplotype blocks in non-coding regions of the genome. Successful mechanistic insight requires identifying disease-relevant cell types and states, mapping variant-to-gene effects, and integrating findings across loci, at scale, to pinpoint pathways of polygenic convergence. Here, we discuss functional genomic, machine learning, and experimental approaches to address each step of this daunting challenge. Ultimately, the convergence of results-across methodologies and within key underlying disease pathways-will be essential to realizing the promise of clinical translation for common, complex brain disorders.