Predicting Neural Activity: Understanding Brain Structure and Dynamics through Generative Modeling and Causal Analysis

Understanding how the brain anticipates future states and transmits or reconstructs information remains a central challenge in neuroscience. This Research Topic invites studies that leverage machine learning and generative modeling to forecast neural activity and behavior from spikes, ECoG, or fMRI. We especially encourage work probing the structural networks and directional information flow that enable neural predictability, clarifying how specific anatomical and functional architectures give rise to these computations. We also welcome studies that train models on one dataset and evaluate them on others to uncover cross-dataset and cross-species relationships, revealing shared principles across individuals, species, and recording modalities.

This Research Topic aims to integrate predictive modeling and causal inference to reveal the principles by which neural circuits forecast, encode, and transmit information. Specifically, it seeks contributions that bridge cutting-edge machine learning with analyses of brain connectivity and causal pathways, and which test the robustness of predictive approaches across species and data modalities. By evaluating when and why models succeed—or fail—and how predictions relate to the underlying neural architecture, we strive to gain deeper insight into how brains compute and represent future states.

To gather further insights into the intersection of predictive modeling, structural connectivity, and causal information flow, we welcome articles addressing, but not limited to, the following themes:

- Time-series prediction of neural activity (e.g., Transformer, diffusion-based models)

- Reconstruction and validation of neural signals using generative models

- Analysis of information flow and causal structures (e.g., Transfer Entropy, Granger Causality)

- Relationship between predictability and structural connectivity

- Cross-dataset model transfer to examine inter-data relationships

- Fluctuation and prediction: handling uncertainty and evaluating predictive reliability

- Data standardization and sharing (e.g., NWB, BIDS) for reproducibility

- Connections to theoretical frameworks (e.g., Predictive Coding, Free Energy Principle, Active Inference)