neural-dynamics-decision-making

What problem does it solve?

Unified framework to model neural decision dynamics by connecting decision theory with neural activity, enabling integrated analysis and interpretation of neural decision processes.

Core Features & Use Cases

• Support for classic drift-diffusion (DDM), multi-dimensional accumulators, variable/collapsing boundaries, and discrete jumps in decision dynamics.
• Parameterized RNN-based state-space models (RNN-SSM) with variational Laplace EM inference for fitting spike data and inferring latent decision states.
• Applications include neural decoding, model comparison across decision architectures, and classification of accumulator vs sensory neurons.

Quick Start

Analyze spike-train data by fitting a multi-dimensional accumulator model with collapsing boundaries and infer hidden decision states using variational Laplace EM.