Revision a04585ff20b1389c713709b543a1af420bd300c1 authored by Pradyumna on 03 August 2020, 10:31:23 UTC, committed by Pradyumna on 03 August 2020, 10:31:23 UTC
1 parent 2a1bb9e
README.md
Code for Perceptual and Value experiments analysis are presented here:
The folders for both experiments are structured in a similar way.
-----
1. <b>'analysis'</b>: results from the behavioural analysis.
<i>ResultsBehavioural</i>: analysis for logistic choice models (Figure 1), gaze distribution (Figure 2) and behavioural measures comparison.
<i>FixationsAnalysis</i>: last fixation to chosen item analysis (Figure 3).
<i>RegressionModels</i>: hierarchical regression models for choice and confidence (Figures 2 and 4).
<i>SupplementaryModels</i>: hierarchical regression model comparison for choice and confidence.
2. 'glamModels'</b>: estimation and results from GLAM model analysis.
<i>GLAM_individual_...</i> : model parameter estimation for like and dislike (frame-evidence* and preference-evidence) in Value Experiment. In Perceptual Experiment most and fewest(frame-evidence and default-evidence). Separate scripts for each frame estimation.
<i>Individual_Simulations_BalanceEvidence</i>: balance of evidence simulations, with and without gaze influence.
<i>Out_of_sample_comparison</i>: individual out-of-sample comparison test set vs GLAM predicted (Figure 5).
<i>BE_ExtraFigs_SumVal</i>: generation of figures for balance of evidence simulations (Figure 6).
<i>ParamsCompare</i>: comparison of GLAM parameters between frames.
<i>ModelComparison</i>: WAIC score comparison for frame-evidence and preference-evidence models.
2.1 Additional folder <b>'glamModels/results'</b> contains output files from GLAM estimations plus our balance of evidence simulations:
<i>glamModels/results/estimates</i>: contains GLAM estimation parameters.
<i>glamModels/results/predictions</i>: contains simulated trials using GLAM parameters for each participant.
<i>glamModels/results/preprocessed</i>: contains test sets for out-of-sample comparison.
<i>glamModels/results/simulations_evidence</i>: contains balance of evidence simulations datasets.
<i>glamModels/results/waic</i>: contains WAIC scores for each model and participant.
-----
*Inv: (inverted) refers to frame relevant evidence (value or number of dots) used for model estimation/results. This notation is used in some files and variable anmes in the GLAM model scripts.
** In the Perceptual Experiment scripts use More and Less to refer to labels Most and Fewest, respectively, used in the manuscript.
Computing file changes ...