EEG-Fringe-P3-analysis

This repository contains code for my MSc dissertation and an ongoing research project that I am working on as lead author, alongside Professor Howard Bowman, Dr Alberto Aviles and other colleagues at the University of Birmingham, with the intention to seek its publication upon completion.

Background

This project constitutes an analysis of electroencephalography (EEG) data collected from an experiment conducted by Dr Aviles, in which subjects viewed streams of faces presented via Rapid Serial Visual Presentation (RSVP), using a system known as the Fringe-P3 Brain-Computer Interface (BCI). In analysing the electrophysiological responses to faces of different forms and strengths of personal significance to the subjects (referred to as 'salience'), we aim to detect a neural signal known as the P300 (P3) - an electrophysiological response recognised as a reliable neural signature of perceived salience - and examine how differences in the form or strength of perceived salience can affect this pattern of neural responding. In doing so, this research aims to expand our current scientific understanding of the P3 component and determine the utility of the Fringe-P3 BCI in its application to more realistic real-world contexts, where the salience of stimuli is likely to naturally vary in form and strength. In establishing the conditions of salience under which the Fringe-P3 BCI can still effectively identify the P3 component, our study offers insight into the possible contexts in which this tool could be utilised to detect the recognition of items, and thereby discern withheld knowledge, without the requisite of explicit disclosure. This research therefore has potentially profound implications for this techology's value in forensic deception-detection, non-physical communication, objective visual acuity tests, and more.

Investigation of this topic at the group-level (across subjects) constituted the focus of my MSc dissertation. Investigation of this topic at the individual-level (within individual subjects' data) is currently being studied.

Please see my research proposal (located in the Proposal/ folder of this repository) for a full description of the research aims and methods. The manuscript for the full project is currently being prepared with the intention of seeking its publication in 2024.

Disclaimer

This project is an ongoing work-in-progress. Repository folders, subfolders and files are subject to change.

Contents

README.md

-This file.

Proposal/

-This folder contains the written research proposal for the project (in .docx and .md format). This proposal outlines the background, aims, methodology and planned analysis of the study.

Code/

-This folder contains the required code for this project, contained in the following subfolders:

• Preprocessing/

  This contains all scripts and functions used in the data preproccessing of this project:

  • preprocess.m

    A function that applies all of our initial preprocessing steps to input raw data - including segmentation, baseline-correction, band-pass filtering, band-stop filtering, re-referencing and artifact exclusion thresholding.

  • preprocessing.m

    A script that loads the raw data, calls the custom preprocess() function to apply the pipeline, then facilitates Independent Component Analysis and channel/trial rejection for further data cleaning.

• ERP/

  This contains all scripts and functions used to generate the Event-Related Potentials of this project:

  • generateSubjectERPs.m

    A function that generates the subject-level ERPs by averaging across pairs of trials that involve adjacent probe/irrelevant morph presentations.

  • erpGeneration.m

    A script that calls the generateSubjectERPs() function to generate the subject-level ERPs, then tidies/prepares the array containing these averaged signals for the subsequent statistical analysis by removing empty cells and any ERPs that do not have a probe/irrelevant equivalent. The script then includes code to optionally adjust the latency of the subject-level ERPs, code to average across the subject-level ERPs to generate the group-level ERPs, as well as code to visualise the resulting group-level signals.

• Analysis/

  This contains all scripts and functions used in the data analysis of this project:

  • groupLevelAnalysis.m

    A script that runs the cluster-based paired-samples permutation test to compare the subject-level ERPs of the probe and irrelevant conditions of each pair level, in each block. The script first defines the neighbours for each channel in the data (to be later used in clusters), then runs the permutation test and plots the significant clusters on topographic maps over time.

Poster/

-This contains a PDF poster that summarises the research project, used for my MSc dissertation conference presentation.

Acknowledgements

This project is being conducted in collaboration with Professor Howard Bowman at the University of Birmingham.

This project uses data collected by Dr Alberto Aviles, as a part of his PhD under the supervision of Professor Howard Bowman.

Invaluable guidance on this project has been provided by PhD student Cihan Dogan, from the University of Kent.

This project utilises the FieldTrip toolbox for data preprocessing and analysis, developed by Oostenveld, R., Fries, P., Maris, E. & Schoffelen, J. M. (2011), at the Donders Institute for Brain, Cognition and Behaviour.