---

---

---

Modelling Reaction Time and Confidence Distributions in Decision Making

This repository contains data as well as code used in the paper Simultaneous modeling of choice, confidence and response time in visual perception (Hellmann, S., Zehetleitner, M., & Rausch, M., 2023, Psychological Review, doi: 10.1037/rev0000411). Preregistration on OSF was developed and published with another data set from a previous study (Experiment 2 in Rausch, Hellmann, & Zehetleitner (2018). The other experiments are available in this repository. The most recent author manuscript of the article is available here: https://osf.io/mzfkr.

Structure:

• dynWEV-source package file (.tar.gz)
• dynConfiR-source package file (.tar.gz) (for the additional analyses for review). This is the 0.0.1 version of the package with likelihood and fitting functions, which is now available on GitHub.
• folders for the experiments analyzed in the study. The two motion discrimination datasets from experiment 2 are again included in sub-folders (and have individual files for the first three bullets). Following files are in the experiment folders:
  • a experiment folder containing the files for running the experiments in Psychopy
  • a .csv file ('dataExperiment.csv') containing the raw data
  • a script R-file for the actual analyses ('Script_FitNPredict_SeqSampConfModels_Experiment.R'), including:
    • Reading, preprocessing, and Aggregating Data
    • Fitting model parameters
    • Prediction of confidence and RT distributions and aggregation of predictions
  • files to generate reported results, figures and tables in the paper (gen_descr_plots.R, gen_model_plots_and_BICAnalysis.R, and gen_table_fittedparameters.R)
  • a script R-file for model identification analysis ('Script_ModelMimikryAnalysis.R')
  • a autosave_mimikry folder with saved results from the model identification analysis
  • a saved_fits folder with two files containing the fitted parameters from the experiment for diffusion based models ('fits_2DSD_WEV.RData') and race models ('fits_RacingModels.R'), respectively
• The folder Additional_Analyses with further non-preregistered analyses conducted for the review process
  • a script R-file additional_analyses_for_review.R with code to fit and predict two models (DDMConf and dynVis) and saved model fits for both models and both experiments
  • two .RData files, collected_fitsNpredicts_Experiment_review.RData with all model fits and predictions for visualization
  • a script R-file and folder for a small parameter recovery study for the dynWEV model
  • gen_model_weights.R with code to transfer BIC values to model weights
  • simulation_dynWEV.R with code to produce a figure with simulations for the dynWEV model with differen weight parameters
  • AUC_Tau_plot.R with code to produce Supplementary Figure 1 for the relationship between metacognitive sensitivity and postdecisional accumulation time

Usage:

Setup the environment:

• Start with a completely new R 4.0.5 installation
• On windows, install rtools40 (https://cran.r-project.org/bin/windows/Rtools/rtools40.html)
• Install the renv package using install.packages("renv")
• Change the working directory to the project directory setwd('~/Material_for_SeqSamplingModelsOfChoiceConfRT')
• Use renv::restore('renv.lock') to install all packages with their respective version. Note, that this will install the packages in your default library of your R-4.0.5 installation!
• Install the local packages:

    install.packages('dynWEV_0.0.tar.gz', repos = NULL, type = 'source')
    install.packages('dynConfiR_0.0.1.tar.gz', repos=NULL, type = 'source')

Redo whole analysis:

• To redo the whole analyses with modelling fitting:
  • remove the files with the results 'collected_fitsNpredicts.RData' in the respective experiment folder
  • run 'Script_FitNPredict_SeqSampConfModels_Experiment.R' from within the respective experiment folder
• To reuse the quantitative model comparison and produce the figures, using the already computed model fits, simply use the saved results in 'collected_fitsNpredicts.RData' in the respective experiment folders for all other analyses
• Note that the scripts 'Script_ModelMimikryAnalysis.R' always run the recovery analysis, irrespective of whether saved results are present or not and that this may take considerable time!

References

Hellmann, S., Zehetleitner, M., & Rausch, M. (2023). Simultaneous modeling of choice, confidence, and response time in visual perception. Psychological Review. Advance online publication. https://doi.org/10.1037/rev0000411

Contact

For comments, remarks, and questions please contact me: sebastian.hellmann@tum.de{.email}