Welcome to EEG_Oddball_BIDS by Mark Nelson, a project in the BrainHackSchool 2020. For this project I will be working with an existing EEG dataset collected from 1000 participants while performing 2 tasks: (1) a 3-stimulus oddball detection task in which the oddball/novel stimulus category is represented by 100+ unique oddballs, and (2) a recognition test featuring 100 stimuli including 50 oddballs from the first task and 50 new.

Participants were instructed in the first task to respond to targets only and in the second task to indicate whether each novel was OLD (recognized) or new (unrecognized). In this way, the overall paradigm was essentially designed to compare EEG activity related to implicit memory encoding, as some novels are later recognized, while others are not, but all are irrelevant to the task structure from the perspective of the participant. The goals of this project are:

OBJECTIVES:

(1) Organize an existing EEG dataset according to the BIDS standard and publish in a git hub repository.

(2) Write a python based machine learning algorithm to predict behavioral measures (object recognition) from single trial EEG amplitude recorded at stimulus encoding.

(3) Analyze data for indidividual differences across participants and plot results using custom python based plotting scripts.

(4) Organize all code into Jupyter notebooks.

TOOLS:

• Jupyter Notebook

• Python

• Anaconda (virtual environment)

• nilearn (ML python package)

• matplotlib, pandas, plotly, seaborn, bokeh (data visualization python packages)

RESULTS: