BRAID: Input-Driven Nonlinear Dynamical Modeling of Neural-Behavioral Data

2025-09-23

AI Summary

Traditional neural population modeling typically assumes autonomous system evolution, neglecting the influence of external drivers—such as sensory stimuli, upstream inputs, and neuromodulation—on neural dynamics and behavior. To address this, we propose an input-driven recurrent neural network framework that integrates multimodal external signals (e.g., perceptual inputs) and employs multi-stage optimization guided by predictive objectives. Crucially, it explicitly disentangles intrinsic reverberatory dynamics from externally driven effects, focusing on behaviorally relevant low-dimensional subspaces. By unifying deep learning with joint neural–behavioral modeling, our approach is validated on both simulated and real motor cortical data. Results demonstrate a significant improvement in neural activity prediction accuracy—averaging 12.7% higher than multiple baselines—and uncover interpretable, causal coupling pathways linking inputs, intrinsic dynamics, and behavioral outputs. This establishes a novel paradigm for causal neural decoding.

Technology Category

• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Machine Learning: Multimodal Learning
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

Neural populations exhibit complex recurrent structures that drive behavior, while continuously receiving and integrating external inputs from sensory stimuli, upstream regions, and neurostimulation. However, neural populations are often modeled as autonomous dynamical systems, with little consideration given to the influence of external inputs that shape the population activity and behavioral outcomes. Here, we introduce BRAID, a deep learning framework that models nonlinear neural dynamics underlying behavior while explicitly incorporating any measured external inputs. Our method disentangles intrinsic recurrent neural population dynamics from the effects of inputs by including a forecasting objective within input-driven recurrent neural networks. BRAID further prioritizes the learning of intrinsic dynamics that are related to a behavior of interest by using a multi-stage optimization scheme. We validate BRAID with nonlinear simulations, showing that it can accurately learn the intrinsic dynamics shared between neural and behavioral modalities. We then apply BRAID to motor cortical activity recorded during a motor task and demonstrate that our method more accurately fits the neural-behavioral data by incorporating measured sensory stimuli into the model and improves the forecasting of neural-behavioral data compared with various baseline methods, whether input-driven or not.

Problem

• Modeling neural dynamics with external inputs influence
• Disentangling intrinsic recurrent dynamics from input effects
• Improving neural-behavioral data forecasting accuracy

Innovation

• Models neural dynamics with explicit external input incorporation
• Disentangles intrinsic dynamics from input effects via forecasting objective
• Prioritizes behavior-relevant dynamics using multi-stage optimization scheme