(modified) Attention Network Test

This repository contains a modified version of the Attention Network Test (Fan et al., 2002) (mANT), implemented in Python.

---

The repository contains three subfolders:

• task-only contains code to run the mANT in a behaviour-only setting. See its own README for details
• task-and-eeg contains code to run the mANT during electroencephalography (EEG). See its own README for details (in short, the code is the same as in task-only, plus some lines to send 8-bit triggers to the EEG recording system)
• task-and-fmri contains code to run the mANT during functional magnetic resonance imaging (fMRI). See its own README for details (in short, the code is the same as in task-only, but the task timeline is different and there are commands to receive triggers from the fMRI scanner in form of simulated keypresses)
• mant-data-analysis contains code to analyse and plot mANT data

All subfolders are independent of each other.

In principle, subfolders would not be necessary: the same code could be used for different experiments if it included a method to select experiment-specific settings (for example, conditional logic). However, separate implementations seemed to be the cleanest solution.

---

Dependencies:

The code contained in the task subfolders (i.e., task-only, task-and-eeg, and task-and-fmri) relies on the following software:

Language/Package	Versions tested on	Suggested installation
Python	3.9.16, 3.9.18	Miniconda
PsychoPy	2022.2.5, 2023.2.3, 2024.2.2	Manual installation
pandas	1.5.3, 2.2.1	conda install pandas

While the code contained in mant-data-analysis relies on:

Language/Package	Versions tested on	Suggested installation
Python	3.9.16, 3.9.18, 3.11.8	Miniconda
NumPy	1.24.3	conda install numpy
pandas	1.5.3, 2.2.1	conda install pandas
Matplotlib	3.8.0	conda install -c conda-forge matplotlib
seaborn	0.13.2	conda install seaborn -c conda-forge
scipy	1.12.0	conda install scipy
statsmodels	0.14.4	conda install -c conda-forge statsmodels

Installation notes:

• A good guide to Python installations (not specific to the mANT nor PsychoPy) is available at this HTTPS URL
• We had problems installing PsychoPy via conda, despite different machines with different operating systems and different levels of user expertise. We found it easier to install via pip
• If you will only use Python to run the mANT, the standalone PsychoPy version (SPV) is your best installation option (simply because it's the easiest). Visit PsychoPy's website for details. However, if you use (or are planning on using) Python beyond the mANT, we strongly discourage using the SPV because:
  • It comes with its own Python installation, even if you already have one on your machine. This can get messy
  • Its main advantage over other PsychoPy versions is a GUI, whose use we discourage (code is more flexible, transparent, and reproducible)
  • It comes with its own Python code editor, which has a lot less features and provides a worse user experience than common, equally free alternatives like VSCode

---

Planned improvements:

• Add functions to compile mant-conditions.csv automatically
• Refactor for higher elegance and efficiency (if/as needed)
• Add requirements.txt files

---

Contacts:

For questions or improvement suggestions, you can:

• Open an issue at - or send a pull request to - this repository
• Write matteo [dot] dematola [at] unitn [dot] it

Matteo De Matola (UniTN | GitHub)

Last updated November 2024