Automatically Controlled: Task-Irrelevance Fully Cancels Otherwise Automatic Imitation

Hemed E, Mark-Tavger I., Hertz U., Bakbani-Elkayam S., & Eitam B. (2021, Journal of Experimental Psychology: General)

Corresponding author: Eitan Hemed

---

About

This repository contains the accepted manuscript, a text document of the supplementary materials and all code and data required to reproduce the analyses and plots included in the paper.

For the published version of the paper, please click here.

If you find any errors or inaccuracies in the code or documentation, please submit an issue or email Eitan Hemed directly.

---

How-To

AuimPy consists of classes and functions used to pre-process, analyze and plot the study data. The data is saved in the individual experiment directories (e.g., ./exp_2/input). The code generates plots and print-ready formatted textural reports that will be saved in the respective experiment directories (e.g., ./exp_4/output/texts).

1. For exploratory data analysis or reproducing the statistical analysis and plots for a single experiment:

• Open the Jupyter Notebook of the required experiment (e.g., ./exp1/exp1.ipynb') and run the notebook's cell.
• The first cell contains imports and general parameters for the current session (e.g., whether to save the figures you plot).
• The 1-2 following cells contain data wrangling, individual to each experiment. In the end of the process the raw data with all relevant variables are available in the data pd.DataFrame.
• The following few cells include usage of the auimpy.prepair.PrepAIR, used for pipelining the wrangling, pre-processing and aggregating the raw data. It is created with prpr = prepair.PrepAIR(data=data, **session_params).
  • The PrepAIR object can generate a textual report of the pre-processing results (prpr.report_invalid_trials()), or plot the aggregated behavioral measures (RT or error rate; prpr.plot_descriptives('rt', by_condition=True, pre_filteration=True))
  • The call to prpr.get_finalized_data() returns the data used with the air.AIR class, see below.
• The last few cells use mainly the air.AIR class. AIR runs and stores the statistical analyses included in the paper, and produces inferential plots. The analysis are run when air.AIR is initialized (rep = air.AIR(aggd_data, contrasts, processed_data, **session_params)).
  • The AIR object is further used to generate the textual report of the statistical analysis and produce the inferential plots for the specified variable (rep.report_results('rt')).
  • We used the EZ-DM (See Wagenmakers, Van Der Maas & Grasman, 2007 ) to fit a drift diffusion model to the behavioral data from the experiment. Implementation and validation are found in the auimpy/ezdm.py module. AIR can also be used to plot the model:
    rep.plot_model_output()

2. In order to produce the meta-analysis plot (Figure 5 in the main text of the paper), go into the meta_plots directory and run the meta_plot notebook. The called code for producing the plot can be found under auimpy/mmp.py.

---

Dependencies

The code might work with other versions, but this has not been tested.

• Python 3.7.4
• Jupyter Notebook
• Pandas 0.23.4
• Matplotlib 3.1.1
• NumPy 1.15.4
• Seaborn 0.9.0
• rpy2 2.9.4 (and an R version 3.6.1, the required R packages should be installed automatically by auimpy.pyrio.PyRIO)