ColourfulBirds

Predicting working memory contents from FOOOF.

fooof >= 1.0.0 neurodsp >= 2.1.0 pandas >= 1.3.0 seaborn = 0.9.0 (only for plotting)

Data

Open Data 2018: Contralateral delay activity tracks fluctuations in working memory performance https://osf.io/8xuk3/

Data Preparation

1. Rename files with 2 digits for the number
2. Resave file unpacked with older version of mat using the conv mat file in utils.

What is this project about?

Colourfulbirds is a project using data from Adam et al. (2018). The data were collected from a visual working memory task.

Experiment 1: different set-sizes and occipital alpha power.

Experiment 2: set-size kept at 6 items. Look at the performance within subjects (good vs. poor). Interested in occipital alpha power, and midline frontal theta power.

Replicating the original results

Our first goal was to replicate the results fom Adam et al. (2018). This was to verify that we understand the data structure and were able to combine the behavioral data correctly with the EEG data.
The replicated results can be found in the "replicated_analysis.ipynb" notebook
Steps:

1. Delete all the to be rejected trials. This is based on the pre-processing from the original paper
2. Select EEG data based on set-size/performance/side
3. Per electrode and per trial, calculate the amplitude (using amp_by_time from neurodsp)
4. Baseline per trial and electrode --> Baselined hilbert transfer
5. Average the hilbert transfers per condition and per subject
6. Lateralize the data: contralateral - ipsilateral

Our approach -- Spectral parameterization

To research the independent contributions of oscillatory and aperiodic activity!
Steps:

1. Delete all the to be rejected trials. This is based on the pre-processing from the original paper
2. Select EEG data based on set-size/performance
3. Per electrode group and per trial, calculate the PSD (using neurodsp) for the retention period and the baseline period
4. specparam the PSDs and save the peak values and aperiodic offset and exponent
5. Subtract the baseline period output from the retention period output

Statistical analysis -- Spectral parameterization

Statistical tests are performed on peak height, aperiodic exponent, aperiodic offset, bandpower, and abundance (% of trials with an oscillation peak):


A 2-way RM ANOVA is used to analyze the data for the occipital electrodes (alpha).
A Pairwise t-test is used for midline frontal theta in Experiment 2, and a 1-way ANOVA in Experiment 1