Experiment

EEG signals were collected to test if user watching videos with green LED lights in the background would have higher valence in their EEG recordings than those that watched the videos without the lights. Signals were recorded for participants watching videos with the LED lights and participants watching video without the lights. Baselines were collected from each participants before stimuli were presented.

Results and Things I’ve Learned

Data Recording

Hardware
  • OpenBCI board was used to collected EEG signals from electrodes.
  • Arduino was used to control the LED lights.
Software
  • OpenViBE was used to record the EEG signal.
  • EEGLab and MATLAB was used for signal processing.
Electrode Sites & Bands

Alpha and Beta bands were used.

Data Processing

I spearheaded the signal processing phase by writing the prototype MATLAB scripts to extract the PSD of the signal and feature normalization for one single file. It was then passed onto another team member to batch process all the signal files for all participants.

Features

Alpha-Beta ratio was taken for the right channels and left channels. Three features that past research have shown to correlate with positive valence were contructed and used from the six channels we collected data from, Fp1, Fp2, F3, F4, F7, and F8.

αFp2/ βFp2 - αFp1/ βFp1
αF4/ βF4 - αF3/ βF3
αF8/ βF8 - αF7/ βF7
Classification

I developed the classification script using logistic regression. I chose logistic regression because our data set was small and it had clear class boundaries.

Results & Shortcomings

We showed that those watched the videos with green LED background had higher valance in their EEG signals then those watched the videos without it. Although we received positive results, we had our shortcomings.

  1. We didn’t have enough data and so our result was statistically insignificant.
  2. Due to half of the team members getting very sick in the middle of the semester, we realized that we didn’t have enough resources to implement the original plan by our deadline, which was to use the change of emotions detected from openBCI to control LED light colors to demonstrate Brain-Computer Interface capabilities. Thus, we pivoted our goals to meet our deadline. A team member and I wrote up everything that we did in a guide-oriented white paper.

Full Implementation White Paper

Things I’ve Learned

  1. How to process EEG signals.
  2. How to use the tools such as openBCI, openViBE, and EEGLab.
  3. The importance of having very clear goals and questions to be answered at the beginning and designing all procedures around answering those questions.
  4. The basic requirement for conducting solid research - having enough data to prove any results is conclusive.
  5. Of course, failing but still making the best out of the situation.

Overall, I learned an incredible amount from this project and hope to implement the actual project plan in graduate school.