Probabilistic Prediction of Neural Dynamics via Autoregressive Flow Matching

2026-04-13

AI Summary

This work addresses the challenge of predicting dynamic neural responses under naturalistic stimulation to advance both fundamental neuroscience and neurotechnology applications. We propose a generative forecasting framework based on Autoregressive Flow Matching (AFM), which, for the first time, integrates an autoregressive architecture into flow matching to explicitly model the joint conditional distribution of current neural activity given both historical neural states and concurrent multimodal sensory inputs. Evaluated on the Algonauts 2025 fMRI dataset, our method substantially outperforms non-autoregressive flow matching and standard generalized linear models, achieving high-accuracy, short-horizon probabilistic predictions of BOLD signals across the entire cortical surface. The approach demonstrates strong generalization capabilities and comprehensive cortical coverage, highlighting its potential for modeling complex, real-world brain dynamics.

Technology Category

• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Machine Learning: Multimodal Learning
• Natural Language Processing: Generation

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

Forecasting neural activity in response to naturalistic stimuli remains a key challenge for understanding brain dynamics and enabling downstream neurotechnological applications. Here, we introduce a generative forecasting framework for modeling neural dynamics based on autoregressive flow matching (AFM). Building on recent advances in transport-based generative modeling, our approach probabilistically predicts neural responses at scale from multimodal sensory input. Specifically, we learn the conditional distribution of future neural activity given past neural dynamics and concurrent sensory input, explicitly modeling neural activity as a temporally evolving process in which future states depend on recent neural history. We evaluate our framework on the Algonauts project 2025 challenge functional magnetic resonance imaging dataset using subject-specific models. AFM significantly outperforms both a non-autoregressive flow-matching baseline and the official challenge general linear model baseline in predicting short-term parcel-wise blood oxygenation level-dependent (BOLD) activity, demonstrating improved generalization and widespread cortical prediction performance. Ablation analyses show that access to past BOLD dynamics is a dominant driver of performance, while autoregressive factorization yields consistent, modest gains under short-horizon, context-rich conditions. Together, these findings position autoregressive flow-based generative modeling as an effective approach for short-term probabilistic forecasting of neural dynamics with promising applications in closed-loop neurotechnology.

Problem

• neural dynamics
• probabilistic prediction
• naturalistic stimuli
• fMRI
• BOLD

Innovation

• autoregressive flow matching
• probabilistic forecasting
• neural dynamics
• generative modeling
• fMRI prediction