dynamic_noise_estimation

Public repository for the dynamic noise estimation method paper.

Citation

@article{LI2024102842,
title = {Dynamic noise estimation: A generalized method for modeling noise fluctuations in decision-making},
journal = {Journal of Mathematical Psychology},
volume = {119},
pages = {102842},
year = {2024},
issn = {0022-2496},
doi = {https://doi.org/10.1016/j.jmp.2024.102842},
url = {https://www.sciencedirect.com/science/article/pii/S0022249624000129},
author = {Jing-Jing Li and Chengchun Shi and Lexin Li and Anne G.E. Collins}
}

File structure

Simulation code (Probabilistic_Reversal/)

• code/simulate_lapses.m: analysis code for simulating data with lapses of attention and comparing the static and dynamic models (Fig 2)
• code/validate_models.m: validation analysis for both models against data simulated by the dynamic model (Fig 3)

Empirical data ([task_name]/data/)

• Dynamic_Foraging (Grossman et al., 2022)
  • Public data repository: https://datadryad.org/stash/dataset/doi:10.5061/dryad.cz8w9gj4s
• IGT (Steingroever et al., 2015)
  • Public data repository: https://osf.io/8t7rm/
  • payoff_lookup.csv, payoff_schedule_1.csv, payoff_schedule_2.csv, payoff_schedule_3.csv were created by Jing-Jing Li according to the payoff schedules described in the paper
• RLWM (Collins 2018)
  • Public data repository: https://osf.io/5gbr3/
• 2-step (Nussenbaum et al., 2020)
  • Public data repository: https://osf.io/we89v/
  • data_processing_scripts/concatenate_mats.m was created by Jing-Jing Li to reorganize the data structure

Modeling code for empirical datasets ([task_name]/code/)

• static_model_llh.m and dynamic_model_llh.m: functions to compute the negative log likelihoods of data given the static and dynamic model parameters
• static_model.m and dynamic_model.m: functions to generate data using the static and dynamic models
• fit_models.m: model fitting code for the static and dynamic models (Fig 4)
• compare_params.m: compares the same parameters between the best-fit values of the static and dynamic models (Fig 5, Fig 6, Fig A11)
• identify_models.m: model identification analysis for both the static and dynamic models (Fig A8)
• recover_params.m: generate and recover analysis for parameters of the dynamic model (Fig A10)
• recover_latent_probs.m: recovery analysis of p(Engaged) trajectory (Fig A10)
• validate_models.m: validation analysis against behavior for both models (Fig A9)

Plots ([task_name]/plots/)

• Output plots for all figures in .png and .svg