Resource
Computational Psychiatry in the Era of Open Science
A manifesto for data, computation, and natural behavior
The Opportunity
Psychiatry is entering a new era.
For the first time, large-scale brain, behavioral, and genetic datasets are becoming openly available. At the same time, advances in machine learning and computational modeling allow us to analyze these datasets at an unprecedented scale.
Computational psychiatry emerges at the intersection of two revolutions:
Its goal is not only to predict symptoms, but to understand the mechanisms linking brain, cognition, and mental disorders.
The First Resource: Data
Large open datasets are transforming the study of the human brain and mental health. These datasets provide the empirical foundation for computational psychiatry.
Major Resources
OpenNeuro
Large repository of openly shared neuroimaging datasets
Human Connectome Project
High-resolution brain data with rich behavioral phenotyping
UK Biobank
Population-scale imaging, genetics, and health records
ABCD Study
Longitudinal brain and mental health development data
IMAGEN Consortium
Imaging genetics data on adolescent psychiatric risk
The Second Resource: Computation
Data alone is not enough. To understand the brain, we must compute it.
Computational psychiatry draws on several methodological foundations:
Statistical & Probabilistic Modeling
- Bayesian inference
- Hierarchical models
- Normative modeling
Machine Learning & AI
- Representation learning
- Deep neural networks
- Generative models
Computational Cognitive Models
- Reinforcement learning
- Decision-making models
- Cognitive control models
The Third Resource: Naturalistic Behavior
Traditional laboratory tasks capture only a small part of human cognition. A new direction in neuroscience is the study of naturalistic behavior.
Open Resources
Natural Scenes Dataset
Brain responses to thousands of natural images
Narratives Dataset
Neural responses during real-world story listening
Natural Movie Datasets
Large-scale behavioral recordings in natural settings
Naturalistic paradigms allow researchers to study cognition as it unfolds in complex, real-world environments.
The Closed Loop of Scientific Discovery
Modern brain science increasingly operates through a closed loop:
Understanding Data
Analyze existing datasets to reveal patterns in brain organization and behavior
Computing Data
Use computational models to uncover hidden structure and causal mechanisms
Generating Data
Design new experiments and datasets informed by these models
This creates a scientific cycle where open datasets are not only research material— they are training grounds for the next generation of experiments and theories.
Toward a New Science of Mind and Mental Health
Computational psychiatry aims to bridge multiple levels of explanation:
By combining open data, computational models, and naturalistic behavior, we can move toward a more mechanistic and quantitative understanding of mental health.
The long-term vision is clear:
A science where psychiatric disorders are understood not only through symptoms, but through computational models of the brain in the real world.
Open Science & Code
We are committed to open science. Code and data from our publications are made publicly available where possible. Visit our GitHub for analysis pipelines, computational models, and datasets.
GitHub @xiec199