cmdsddfeitc - Research compendium for the report on the cognitive mechanisms of the defer-speedup and date-delay framing effects in intertemporal choice by Zandbelt

Compendium DOI

DOI: 10.5281/zenodo.3258148

The files at the URL above will generate the results as found in the preprint. The files hosted at https://github.com/bramzandbelt/cmdsddfeitc/ are the development versions and may have changed since the preprint was published.

Author of this repository

Bram Zandbelt ([email protected])

Published in:

TBA

Overview

The packagae cmdsddfeitc is a research compendium of the research project Cognitive Mechanisms of the Defer-Speedup and Date-Delay Framing Effects in Intertemporal Choice by Bram Zandbelt. This project was conducted at the Donders Institute, Radboud University / Radboucumc, Nijmegen, the Netherlands, and registered at the Donders Centre for Cognitive Neuroimaging under project number 3017051.01 (DCCN PI: Roshan Cools).

This research compendium contains all data, code, and text associated with the above-mentioned publication and is organized as follows:

.
├── R
├── analysis
│   ├── bash
│   └── notebooks
├── data
│   ├── derivatives
│   └── raw
├── documents
│   ├── content
│   └── context
├── figures
│   ├── 03_computational_modeling_analysis
│   ├── 04_sanity_check_control_trial_performance
│   ├── 05_eda_grp
│   ├── 06_model_comparison_grp
│   ├── 07_observed_vs_predicted_performance_grp
│   ├── 08_analysis_of_model_parameters
│   └── 09_sanity_check_effect_framing_on_model_predicted_auc
├── man
├── metadata
│   └── raw
├── packrat
│   ├── lib
│   ├── lib-R
│   ├── lib-ext
│   └── src
└── reports
│   ├── 01_preprocessing_idv
│   ├── 02_eda_idv
│   ├── 03_computational_modeling_analysis_idv
│   ├── 04_sanity_check_control_trial_performance_grp
│   ├── 05_eda_grp
│   ├── 06_model_comparison_grp
│   ├── 07_observed_vs_predicted_performance_grp
│   ├── 08_analysis_of_model_parameters_grp
│   └── 09_sanity_check_effect_framing_on_model_predicted_auc_grp
``

The `R/` directory contains:

- R code specific to the present project; functions are organized into files (e.g. functions for plotting are in `plot_functions.R`)

The `analysis/` directory contains:

- R Markdown notebooks implementing the analyses (`notebooks/` directory), numbered in the order in which they should be run;
- shell scripts running the R Markdown notebooks with appropriate parameters, if any (`bash/` directory).

The `data/` directory contains:

- the raw performance data (`raw/` directory);
- the data derived from the raw data (`derivatives/` directory), organized by notebook name.

The `documents/` directory contains:

- documents describing the content of the experimental data (`content/` directory), such as codebooks;
- documents describing the context of the data (`context/` directory), such as ethics documents, data management plan, and preregistration;
- documents related to the report of this research project (`manuscript/` directory).

The `figures/` directory contains:

- visualizations of descriptive and inferential statistics, organized by notebook name.

The `man/` directory contains:

- documentation of objects inside the package, generated by `roxygen2`.

The `packrat/` directory contains:

- R packages the research compendium depends on; for more info see [https://rstudio.github.io/packrat/](https://rstudio.github.io/packrat/).

The `reports/` directory contains:

- static HTML versions of the knitted R Markdown notebooks, organized by notebook name.

Finally, this research compendium is associated with a number of online objects, including:

| object | archived version                 | development version              |
| ------ | -------------------------------- | -------------------------------- |
| preregistration | [https://osf.io/rzqh9/](https://osf.io/rzqh9/) | NA |
| data management plan | [https://doi.org/10.6084/m9.figshare.4720978](https://doi.org/10.6084/m9.figshare.4720978) | NA |
| stimulus presentation code | [https://doi.org/10.5281/zenodo.3243777](https://doi.org/10.5281/zenodo.3243777) | [github.com/bramzandbelt/itch_time_framing_task](github.com/bramzandbelt/itch_time_framing_task) |
| cognitive modeling code | [https://doi.org/10.5281/zenodo.3243806](https://doi.org/10.5281/zenodo.3243806) | [https://github.com/bramzandbelt/itchmodel](https://github.com/bramzandbelt/itchmodel) |

## How to use

This repository is organized as an R package, called `cmdsddfeitc`. The R package structure was used to help manage dependencies, to take advantage of continuous integration for automated code testing and documentation, and to be able to follow a standard format for file organization. The package `cmdsddfeitc` depends on other R packages and non-R programs, which are listed below under [Dependencies](#Dependencies).

To download the package source as you see it on GitHub, for offline browsing, use this line at the shell prompt (assuming you have Git installed on your computer):

Install `cmdsddfeitc` package from Github:

- From R:

devtools::install_github("bramzandbelt/cmdsddfeitc")

- From the command line:

git clone https://github.com/bramzandbelt/cmdsddfeitc.git ```

Once the download is complete, open the file cmdsddfeitc.Rproj in RStudio to begin working with the package and compendium files. To reproduce all analyses, run the shell script analysis/bash/run_all_analyses.sh. This will run all RMarkdown notebooks in correct order. Note, however, that this will not reproduce the computational modeling analyses performed in the document 03_computational_modeling_analysis.Rmd), only the result of the optimizations. This is because optimization of all 708 models (59 participants (defer-speedup, N = 28; date-delay, N = 31), 6 parameterizations, 2 architectures) was done on a computer cluster and would take simply too long to run on a regular computer. In order to reproduce the computational modeling analyses, run 03_computational_modeling_analysis.Rmd as a parameterized report with argument optimize=TRUE.

Parameterization of analysis notebooks

The analyses can be customized, by specifying a number of parameters. Below is an overview of the parameters that can be set and the notebooks in which they are used.

• participant_id: the participant identifier (ranging from 1 to 93);
• task: the task from which to process the data:
• defer_speedup: intertemporal choice task with neutral, defer, and speedup frames;
• date_delay: intertemporal choice task with delay and date frames;
• visualize: whether or not to visualize the results (e.g. set to FALSE when fitting the model to the data);
• optimize: whether or not to optimize the parameter values (i.e. set to FALSE when only visualizing the data);
• pars_from_file: whether to use parameter values stored on disk (i.e. set to TRUE when visualizing the data)
• algorithm: specification of the optimization algorithm:
• DEoptim: differential evolution algorithm with (lower and upper) bound constraints;
• DEoptimR: differential evolution algorithm with (lower and upper) bound and nonlinear constraints (i.e. ensuring that P(LL choice|SS amount = 0) > 0.75 and P(LL choice| SS amount = LL amount) < 0.25);
• model_name: name of the model class to fit to the data, can be any of the following:
• DDM: Drift-diffusion model (fits choices and response times);
• DFT_CRT: Decision field theory (fits choices and response times);
• DFT_C: Decision field theory (fits choices only);
• parameterization: name of the parameterization to fit to the data:
• time_scaling: lets \kappa parameter of the time function vary between frames;
• value_scaling: lets \mu parameter of the value function vary between frames;
• time_and_value_scaling: lets \kappa parameter of the time function and \mu parameter of the value function vary between frames;
• time_scaling_t0: lets \kappa parameter of the time function and non-decision time (t0) vary between frames;
• value_scaling_t0: lets \mu parameter of the value function and non-decision time (t0) vary between frames;
• time_and_value_scaling_t0: lets \kappa parameter of the time function, \mu parameter of the value function, and non-decision time (t0) vary between frames;
• bound_setting: controls the lower and upper bounds on the model parameters:
• standard: uses parameter bounds specified in the preregistration document;
• wide: uses wider parameter bounds, allowing parameters values to account for 'reverse' framing effects (e.g. more LL choices for delay than date frames);
• max_iter: maximum iterations performed before optimization algorithm stops;
• rel_tol: tolerance of the optimization algorithm's stopping criterion;
• n_pop_per_free_param: number of population members per free parameter (see NP in DEoptimR::JDEoptim and DEoptim::DEoptim.control).

These parameter are used in the following analysis notebooks (indicated by their number only)

parameter	01	02	03	04	05	06	07	08	09
participant_id	X	X	X
task				X	X	X	X	X	X
visualize	X	X	X
optimize			X
pars_from_file			X
algorithm			X			X	X	X	X
model_name			X
parameterization			X
bound_setting			X
max_iter			X
rel_tol			X
n_pop_per_free_param			X

Licenses

Manuscript: CC-BY-4.0 http://creativecommons.org/licenses/by/4.0/

Code: MIT http://opensource.org/licenses/MIT, year: 2019, copyright holder: Bram B. Zandbelt

Dependencies

Below is the output of sessionInfo(), showing version information about R, the OS, and attached or loaded packages:

sessionInfo()
#> R version 3.6.0 (2019-04-26)
#> Platform: x86_64-apple-darwin15.6.0 (64-bit)
#> Running under: macOS Mojave 10.14.5
#> 
#> Matrix products: default
#> BLAS:   /Library/Frameworks/R.framework/Versions/3.6/Resources/lib/libRblas.0.dylib
#> LAPACK: /Library/Frameworks/R.framework/Versions/3.6/Resources/lib/libRlapack.dylib
#> 
#> locale:
#> [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
#> 
#> attached base packages:
#> [1] stats     graphics  grDevices utils     datasets  methods   base     
#> 
#> loaded via a namespace (and not attached):
#>  [1] compiler_3.6.0  magrittr_1.5    htmltools_0.3.6 tools_3.6.0    
#>  [5] yaml_2.2.0      Rcpp_1.0.1      stringi_1.4.3   rmarkdown_1.13 
#>  [9] knitr_1.23      stringr_1.4.0   xfun_0.7        digest_0.6.19  
#> [13] packrat_0.4.9-3 evaluate_0.14

Packrat takes care of dependencies.

Model optimization (notebook) was performed in R 3.5.1. on the Donders Institute computer cluster.

Acknowledgment

This research project was funded through European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No. 703141 to Bram B. Zandbelt and through a Vici grant from the Netherlands Organization for Scientific Research (NWO; grant number 453-14-015) to Roshan Cools. I thank Roshan Cools (RC) for financial support and constructive feedback. I thank Ben Marwick for inspiration on how to create, organize, and describe research compendia.

Contributor roles

We specify the contribution of all people involved in the research (contributing non-authors included), according to the Contributor Role Taxonomy.

	BBZ	RC
Conceptualization	X	-
Methodology	X	-
Software	X	-
Validation	X	-
Formal analysis	X	-
Investigation	X	-
Resources	X	-
Data curation	X	-
Writing - original draft	X	-
Writing - review and editing	X	-
Visualization	X	-
Supervision	X	-
Project administration	X	-
Funding acquisition	X	X

Contact

Bram B. Zandbelt