Emotion detection using OpenBCI

fmafma Luxembourg
in General Discussion

Hello guys,
First of all, accept my apology if the discussion has redundancy with other previous topics in the forum.
We are a couple of engineers with different backgrounds who came together to form a team in order to create a "Neurofeedback based emotion detection" tool (mainly to be used by game developers for playtesting purposes).
We started by going through the body of literature on the topic which seems very promising. Then we created a prototype using Muse 2 based on Russel's 2-dimensional emotion detection model. The first tests with the tool looked promising although we could feel the lack of enough fidelity in the reports. So according to the fact that InteraXon stopped serving the Muse SDK and on the other hand to be able to involve more parts of the brain in the signal capturing process, we decided to immigrate to OpenBCI. We got a Cyton and a Ganglion and changed the application.
Now everything seems to work fine, but we still need to improve the accuracy of the measurement to prevent the results to look like random noises.
I should mention that our general approach is to transform the signals into brainwaves and measure the valence based on the alpha wave asymmetry in the front lobe and the arousal based on the beta to alpha activity ratio which seems to be a valid idea based on many papers (e.g. https://www.sciencedirect.com/science/article/pii/S2351978915003777)
Now my questions are:
1. Are there any real-world applications/tools using brain signals to detect a cognitive feature? If yes, are there any of them which use an OpenBCI device to do the job? (please don't refer to papers)
2. Do you guys, in general, think that what we are trying to do, can be done with a good level of accuracy? (and by that I mean the final measurement results just not to be random and without any specific meanings)
3. What are the possible ways to improve the accuracy of our solution?

Any answers to my questions or any hints would be more than appreciated and definitely would help us a lot.
Cheers,

Comments

  • wjcroftwjcroft Mount Shasta, CA

    Fma, hi.

    Thanks for your post. I just wanted to point out, that using the "Google Advanced Search" button in the upper right column, brings up a large range of previous threads on Affective BCI.

    https://www.google.com/search?as_q=affective+bci&as_sitesearch=openbci.com

    I'm not clear why you are mentioning "please don't refer to papers", since the OpenBCI citations list is a rich resource.

    https://docs.openbci.com/docs/08FAQ/FAQLanding

    I count over 10 papers (using OpenBCI) referring to the term 'affective'. Similar numbers for 'emotion' term.

    Best regards, William

  • retiututretiutut Louisiana, USA
    edited April 2021

    You probably need a multi-modal sensor network to increase the accuracy of emotion recognition. Heart rate is the easiest metric to add using Cyton pins and modified hardware settings for a channel. You can also use the pulse sensor plugged into a Cyton Aux pin (D11).

    I would think this is an area of active research and development for many companies and any ready-made solutions are not being shared publicly. I highly encourage anyone working with OpenBCI hardware to share progress on Affective Computing with the Community and on GitHub under an MIT or other open-source license.

    For such reasons, OpenBCI is developing Galea.

    https://www.galea.co
    https://openbci.com/community/introducing-galea-bci-hmd-biosensing/
    https://openbci.com/community/affective-computing-and-mixed-reality-guillermo-bernal/

    Also, Emotibit will be launching a kickstarter soon! A Cyton + Emotibit will be a powerful combo.

    https://www.emotibit.com

    Hope this helps!
    -RW

  • fmafma Luxembourg

    Thank you William for the quick reply.
    The reason I asked about the papers, is because I already checked most of the related stuff and specifically the google search button in the forum for many combinations of Cognition, Affection, Emotion, and a few other words. I share the same idea that there is lots of great material in the papers. But the reason I'm looking for applications is to find these theories in action, with obvious repeatability and not in a controlled (in-lab) setting.
    Cheers,

  • wjcroftwjcroft Mount Shasta, CA

    Fma, thanks.

    But the reason I'm looking for applications is to find these theories in action, with obvious repeatability and not in a controlled (in-lab) setting.

    My hunch is that each paper written about Affective BCI applications, will also contain 'validity' test results. For example, over a series of trials, what percentage of the trials, was the Affective emotional valence prediction actually matching what the subject felt. Or was 'directed' to feel by exposure to external stimuli. This would give a good metric to compare algorithms over a range of possibilities.

    On the commercial application side, if such devices do not come with research proving their efficacy, then one might have to just trust in the company's 'marketing' savvy. ;-)

    As Richard mentioned, the accuracy of this field is currently in flux, so that might explain why there are not large number of commercial companies already claiming they have mastered this capability. The class of research papers called 'review' papers, might offer a step towards what you are looking for. This 2017 review paper covered 285 publications on affective bci:

    https://www.researchgate.net/publication/321398821_Review_and_Classification_of_Emotion_Recognition_Based_on_EEG_Brain-Computer_Interface_System_Research_A_Systematic_Review

    Regards, William

  • wjcroftwjcroft Mount Shasta, CA

    In alignment with Richard's comments about multi-modal emotion recognition, here is a 2020 paper,

    https://www.mdpi.com/2076-3425/10/10/687/htm
    "Advances in Multimodal Emotion Recognition Based on Brain–Computer Interfaces"

    Abstract With the continuous development of portable noninvasive human sensor technologies such as brain–computer interfaces (BCI), multimodal emotion recognition has attracted increasing attention in the area of affective computing. This paper primarily discusses the progress of research into multimodal emotion recognition based on BCI and reviews three types of multimodal affective BCI (aBCI): aBCI based on a combination of behavior and brain signals, aBCI based on various hybrid neurophysiology modalities and aBCI based on heterogeneous sensory stimuli. For each type of aBCI, we further review several representative multimodal aBCI systems, including their design principles, paradigms, algorithms, experimental results and corresponding advantages. Finally, we identify several important issues and research directions for multimodal emotion recognition based on BCI.

  • fmafma Luxembourg

    Thank you Retiutut for the effective response.
    You are right about multi-modal approach, We did a prototype with GSR and some research to integrate the PPG or HRV signals in our solution as well which seems to be crucial.
    But regarding the current status, our mission is to bring as much fidelity as possible out of EEG signals, and create something solid out of it and then use other helper methods for fine-tuning, calibration and stabilization.
    I looked at Emotibit, and it looks very promising. The only thing I didn't find out was the type of output it will provide (is it bio-signals to extract the emotion in the user side or the emotion itself).

  • fmafma Luxembourg

    Thank you William for the papers you mentioned. I'll definitely read them.
    And also I agree with you on the accuracy of the field to be in flux and I believe that this is a job that needs to be done as the number of the usages are so wast and the final value which it can bring to industries would be huge.
    This is a mission that requires every party's collaboration and we are so eager to accomplish it.

Sign In or Register to comment.

© Vanilla Keystone Theme 2026