Gaze Bias Differences Capture Individual Choice Behavior

This repository contains data and python code to reproduce all analyses performed in

Gaze Bias Differences Capture Individual Choice Behavior
Thomas, A. W.* & Molter, F.*, Krajbich, I., Heekeren, H. R. H. & Mohr, P. N. C.
Nature Human Behaviour, 2019, X(X), p. XXX.
doi: http://dx.doi.org/10.1038/s41562-019-0584-8

*shared first authorship with equal contribution

The main analyses can be followed within multiple Jupyter notebooks (files with .ipynb extension):

• 0_data_preprocessing.ipynb: Preprocessing of the four included datasets
• 1_individual_differences.ipynb: Descriptive results and basic behavioural analyses
• 2_relative_model_fit.ipynb: Evaluation of results from within-subject model comparison
  • Model fitting and comparison performed using the GLAM_insample_comparison.py script
• 3_absolute_model_fit.ipynb: Evaluation of absolute model fit (out of sample prediction)
  • Model fitting and prediction performed using the GLAM_oos_prediction.py script
• 4_glam_parameters_predict_behaviour.ipynb: Analysis of relationships between model parameters and behavioural measures

Additional supplementary analyses are contained in the following Jupyter notebooks:

• SI_0_convergence_check.ipynb: Convergence checks for MCMC traces
• SI_1_parameter_estimates.ipynb: Visualization of parameter estimates (Supplementary Figure 1)
• SI_2_multiplicative_vs_additive.ipynb: Individual comparison between multiplicative and additive GLAM variants (Supplementary Figure 2)
• SI_3_additive-vs-multiplicative_group-averaged.ipynb: Group comparison between multiplicative and additive GLAM variants (Supplementary Figure 3)
• SI_4_OOS_predicted_behavioural_metrics.ipynb: Visualization of out-of-sample predicted individual differences and relations on behavioural metrics (Supplementary Figure 4)
• SI_5_6_Individual_RT_distributions.ipynb: Visualization of group and individual response time distributions (Supplementary Figures 5 and 6)
• SI_7_parameter_recovery.ipynb: Visualization of parameter recovery analysis
  • Recovery performed using GLAM_parameter_recovery.py script

The files analysis_functions.py and plotting_functions.py contain shared functions that are loaded by each notebook separately.

The data from Folke et al. (2016) are licensed under a CC BY 4.0 license and can originally be obtained from figshare.