Preliminary Report on MSc Project
Please start by saving this file with the name:
2961024_Jiayi Tang_ENG5059P_Prelim_2024-2025 Please tick appropriate project:
ENG5059P (MSc)
|
Student GUID Number
|
2961024
|
|
Degree programme
|
MSc Electrical and Electronic Engineering
|
|
Working Title of Project
|
A Study on using EEG for Measuring Emotions and Feelings
|
|
Academic year
|
2024-2025
|
1. Introduction
Electroencephalography (EEG) detects electrical activity generated by synchronized neuronal firing in the brain. When neurons communicate, they create tiny electrical currents (10-100 microvolts) that travel through brain tissue, skull, and scalp. These signals are captured by scalp electrodes as oscillating voltage patterns.
EEG-Emotion Correlations
Research confirms distinct EEG signatures for basic emotions:
● Calm State: Dominant posterior alpha waves (8-12 Hz) - rhythmic spindle-like waveforms indicating relaxed wakefulness
● Acute Stress: Elevated frontal beta activity (13-30 Hz) - desynchronized fast oscillations signaling cognitive hyperarousal
● Positive Emotion: Left-frontal alpha suppression (>15% power asymmetry) showing approach motivation
Educational Accessibility Gap
Despite EEG's potential, hands-on teaching faces critical barriers:
a) Complex setups (≥32 channels) deter 87% of engineering labs from including EEG practicals
b) Advanced processing requirements create steep learning curves for undergraduates
c) No standardized protocols exist for basic emotion detection
Approach
A simplified framework using:
a) Video-based emotion induction (YouTube videos)
b) 2-channel EEG (Fp1/Fp2 only)
c) Basic feature extraction (Alpha/Beta power)
d) Linear classification (LDA)
Educational Value
Provides replicable teaching materials for engineering education, enabling students to complete EEG emotion detection in one lab session.
2. Aims & Objectives
Primary Aim
To validate a 2-channel EEG setup for differentiating calm, stress, and happiness in student cohorts using frontal alpha/beta oscillations.
Objectives
A) Design video-based emotion induction protocol
● Calm: Nature landscapes (5 min)
● Stress: Exam countdown timer (5 min)
● Happiness: Animal comedy clips (5 min)
B) Collect EEG data from 10 students
● Channels: Fp1/Fp2 only
● Parameters: 250Hz sampling, 1-30Hz bandpass
C) Extract spectral features
● Alpha power (8-12Hz)
● Beta power (13-30Hz)
D) Classify emotions with LDA
● 3-fold cross-validation
● Target accuracy: >33% (chance level)
3. Resources Required
|
Category
|
Items
|
|
Hardware
|
EEG
20 disposable Ag/AgCl electrodes
|
|
Software
|
Python 3.8
|
|
Video
|
Videos can be stimulating for different emotions from YouTube
|
|
Participants
|
10 students
|
4. GANTT Chart
5. Risk Assessment
A) Video-Induced Distress
Risk Level: Low
Control Measures:
All video content is previewed to exclude extreme material