This repository contains code for the paper Spaak, Peelen, & De Lange (2021): "Scene context impairs perception of semantically congruent objects", Psychological Science.

Available from Donders Repository: data/, mcmc-traces/.
Available from GitHub: requirements.txt, requirements_human.txt, analysis/, results/.

requirements.txt:
Result of conda list --export >requirements.txt. Lists the specific version of all packages that were installed through conda/pip in my environment when the analyses were run. Note that not necessarily all those packages were used. Key dependencies I know are used are the typical ones: Numpy, Scipy, Matplotlib, Pandas. Also note that packages that were installed in "develop mode", e.g. ones I cloned directly from Git, are not included. Extra dependencies are:
- Pingouin (commit eaec24c8dadcd1c5f233db04ae2df97f03cfd68f on my own fork and branch https://github.com/Spaak/pingouin/tree/simple-formatting-tweaks)
- PyMC3 (v3.9.2)
- Theano (commit 93e8180bf08b6fbe587b6f0ecc877ec90e6e1681 on default master)
- Arviz (commit 0ff892c73ce504313927ec8609dcff6683078707 on default master)
- Bambi (commit f77807e2bf310e522899d2df8c8800c44d0b9f73 on default master)
- Seaborn (commit 8fe4ba37c3de9ca3171f7fb718269e859d78c457 on my own fork and branch https://github.com/Spaak/seaborn/tree/parametric_ci)

requirements-human.txt:
Human-readable version of the above (conda list >requirements-human.txt).

analysis/:
Analysis scripts. Tested on Python 3.6.6 and with the above dependencies.
- run_simple_behavioural_analyses.py: run all analyses except for the hierarchical Bayesian models, save figures and stats results.
- run_modelling_analyses.py: run hierarchical Bayesian models (see code to switch between sampling the models again or using existing traces).
- beh_analysis.py: helper functions.
- modelling.py: helper functions for hierarchical modelling.
- plots.py: helper functions for plotting.

data/:
Data files, both in CSV format and pickled dataframes.

mcmc-traces/:
MCMC traces, output of hierarchical Bayesian modelling. These are generated by, and used by, stuff in run_modelling_analyses.py.

results/:
Figure and statistics output.