Data and material for the experiments and the modeling reported in Rothe, Deverett, Mayrhofer, & Kemp (in press). Successful structure learning from observational data. [preprint]
data_by_subject.csv contains the causal structures that participants inferred in Experiments 1 to 5.
exp_paperExperiment number, same ordering as reported in the paper
subjectParticipant ID (note, subject
1in Experiment 1 is not the same participant as subject
1in Experiment 2, etc.)
blockBlock number, same ordering as reported in the paper
2(only relevant for Experiment 1, where each block was shown two times)
graphID of the causal structure drawn by the participant (for all 64 causal structures and their IDs, see
graphs_overview.pdfshown below or the identical Figure C1 in the Appendix)
data_model.csv contains the aggregated responses from participants and the model predictions
graphIn this csv file, for each block in each experiment,
subject_relThe percentage of participants that drew this graph in this block (i.e., sums to 1 within each block)
LSLModel predictions = model posteriors over graphs (i.e., each posterior sums to 1 within each block)
instructions_outline.pdf outlines the instructions for Experiment 1 (the footnote "Original" was not shown in the experiment). The instructions for Experiments 2 to 5 were built upon these core instructions (see Table A.1 in the Appendix for an overview and descriptions in the main text for details).