Neurosynth
Overview
Neurosynth is a platform for large-scale, automated meta-analysis of functional MRI (fMRI) data. Developed by Tal Yarkoni and colleagues and first published in 2011, it combines text-mining of published fMRI articles with coordinate-based activation meta-analysis to produce brain maps linking cognitive and behavioural terms to spatial patterns of activation. Its database contains activation coordinates extracted from over 14,000 published fMRI studies, and provides a publicly accessible interface for performing automated reverse inference — estimating which cognitive processes are associated with a given brain region, and vice versa.
Neurosynth draws on Cognitive Atlas terms as its controlled vocabulary for cognitive and behavioural concepts, linking text-mined term frequencies in abstracts to peak activation coordinates. This makes it the primary operational link between the Cognitive Atlas taxonomy and empirical neuroimaging findings. A newer companion tool, Neurosynth Compose (2024), extends the original platform with a PRISMA-compliant interface for researcher-curated systematic meta-analyses, integrated with NeuroStore (a database of over 30,000 studies with pre-extracted coordinates) and the NiMARE Python library for reproducible meta-analytic workflows.
Use in Neuroscience
Neurosynth is widely used for topic-based brain decoding, functional parcellation, region-of-interest characterisation, and seed-region identification in connectivity analyses. Its database and tools are freely accessible, and all underlying data and code are open-source.
Connections
- Uses vocabulary: Cognitive Atlas (term-to-activation mapping)
Resources
- https://neurosynth.org (original platform)
- https://compose.neurosynth.org (Neurosynth Compose)
- https://github.com/neurosynth/neurosynth
- https://doi.org/10.1038/nmeth.1635 (Yarkoni et al. 2011, Nature Methods — original paper)
- https://doi.org/10.1162/IMAG.a.1114 (Kent et al. 2026, Imaging Neuroscience — Neurosynth Compose)