This repository contains the data and code for running the experiments and reproducing all results reported in our paper Validation and Comparison of Non-Stationary Cognitive Models: A Diffusion Model Application. Superstatistics are emerging as a flexible framework for incorporating non-stationary dynamics into existing cognitive model classes. In this work, we provide the first experimental validation of superstatistics and formal comparison of four non-stationary diffusion decision models in a specifically designed perceptual decision-making task.
The code depends on the BayesFlow library, which implements the neural network architectures and training utilities.
@article{schumacher2023,
title = Validation and Comparison of Non-Stationary Cognitive Models: A Diffusion Model Application,
author = {Schumacher, Lukas and Schnuerch, Martin and Voss, Andreas and Radev, Stefan T.},
year = {2023},
number = {arXiv:2401.08626},
eprint = {2401.08626},
primaryclass = {q-bio, stat},
publisher = {{arXiv}},
}All applications are structured as runable python scripts or jupyter notebooks, which are detailed below.
- Model evaluation: Visualization of inferred parameter trajectory and aggregated posterior re-simulation results.
- Response time series: Posterior re-simulation and prediction of response time series.
- Network validation: In silico model comparison and sensitivity results.
- Network application: Empirical model comparison results.
This work was supported by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation; grant number GRK 2277 ”Statistical Modeling in Psychology”)
MIT