con_pit

Task data, analysis code and computational models for Adolescents flexibly adapt action selection based on controllability inferences (Raab, Goldway, Foord & Hartley, Learning & Memory, 2024; DOI, OSF project).

Overview

90 participants aged 8-27 performed a probabilistic Go/No-Go learning task in controllable and uncontrollable environments. The analyses test how action selection adapts to inferred controllability across development and fit reinforcement-learning models to the choices.

Repository contents

Path	Contents
Con_pit_analysis_main.Rmd	R Markdown analysis: regressions, model comparison and figures
Con_pit_analysis_main.html	Rendered output of the analysis
computational_modeling_code/	MATLAB reinforcement-learning models. Entry point: Main_fitting_code.m. Likelihood functions in lik_functions/, parameter recovery in param_recovery/. Uses the mfit toolbox (mfit-master/, with its own license).
data/task_data/	Per-participant task files (MATLAB .mat files and logs)
data/age_cov_n90.csv	Age, age group and gender per participant
data/subDemographicStats.csv	Sample composition
data/w_trial_by_trial.csv	Trial-by-trial data used in the R analysis

Requirements

• MATLAB R2023a with the Optimization Toolbox (models are fitted with fmincon).
• R with: R.matlab, afex, cowplot, dplyr, emmeans, ggplot2, grid, gridExtra, lme4, modelbased, nlme, pander, psych, rempsyc, report, see, segmented, tidyverse.

Running

1. Modeling (MATLAB): run computational_modeling_code/Main_fitting_code.m.
2. Analysis (R): knit Con_pit_analysis_main.Rmd. The file paths at the top point to the data files, so adjust them to your local copy.

Citation

Raab, H. A., Goldway, N., Foord, C., & Hartley, C. A. (2024). Adolescents flexibly adapt action selection based on controllability inferences. Learning & Memory, 31(3), a053901. https://doi.org/10.1101/lm.053901.123