My Project Diary 2: Gaze at my gaze data + some frustrations and bad choices
Ok, so here we are at the second part of this diary. As I am writing this, I am waiting for a python package to download so I can move forward with the code. This was actually what motivated this story being written now: just HOW LONG it is taking to download a simple seaborn package.
I mentioned in part 1 that my old notebook was working like a charm — except for the internet part. It has a problem with the internet connection, which I am guessing is what is behind the long wait for a seaborn download. I am intending to download seaborn to make a scatterplot with the x and z positions of the eye gaze data.
When working with commercially available devices, we might be seeing devices that do not have great precision. Although there are studies backing up the Gazepoint device use in research, the same was not true for the mindwave device. Despite that, I can still look out for meaningful correlations for the data that I am collecting and further investigate them in the future. It will, of course, affect what I can say about any possible correlations and their applications. We do not want to jump to conclusions due to bad data.
Since the project is an engineering project, I would say the focus is on building a way to collect gaze and brain data simultaneously and analyze their data. If it does not work “minimally well” for commercially available devices, I can test with more precise equipment and use the same algorithms to lookout for interesting relationships between eye gaze and brain data with a larger number of participants — and probably with well thought visual stimuli to test for working memory or other cognitive functions — in the future.
About the bad choice that I mentioned in the title? Not buying a notebook with the operational system that would definitely work with the project devices and trying out a new OS. It has got me to experience three different OS now, but it did not help much in the development of this specific project.
I did take a loong time to finish building the gaze + EEG data collector. During this break, I moved out, changed jobs (twice), had all that craziness of not knowing where things were inside the house, was keeping up with the Artificial Intelligence post-graduation… So, for now, I:
- finished furnishing the house
- Gave me a break from the second post-graduation (am going on my own pace)
- I am completing some amazing courses on working with neural data, linear algebra, MATLAB (although I may not use it here, it is a great tool to work with matrices and know what you are doing), Fourier transform, and a few others I believe will be helpful.
Finally, let's look at how the data is right now:
Here I could see that during the first seconds of data collection, the EEG presented a poor signal. Part of the preprocessing will be to remove any lines with poor signal.
Regarding the eye gaze position, I got this scatterplot for testing purposes (does not account for order/time of the gaze, only position):
So, with this, I finally complete the data collection on both EEG and Eye Gaze. Next, I want to:
- Find out how to capture pupil size from Gazepoint using python as well
- Understand more about each product limitations, frequency of update, account for delays due to computational limitations, and so on
- Collect a LOT of data and decide on a time interval that makes sense giving what I will learn about the devices’ update rate
After getting some more information, I will get back to finish writing the project. For now, I will be focusing on checking for improvements on the device setup/code, reading more articles on the subject, and searching for correlations between the data. That will be it for now.
