Open ExG Headphones for BCI & Earables Research

tl;dr

Long story short: We present an open-source hardware design for BCI & Earables research by integrating OpenBCI components with 3D printable headphones. These headphones allow you to study brain, heart, eye, and facial muscle activity in an aesthetically cherished form. The system offers 21 electrode positions that you can select and use with any of the OpenBCI signal amplifiers (Ganglion, Cyton, or Cyton+Daisy). You can easily replicate this design if you have an OpenBCI Ultracortex Mark IV, 3D-print the remaining parts, and follow our instructions on GitHub. If you want to know why and how we did it – read on!

Are BCIs just 30 years out of reach?

Continuous brain data collection could enable so many cool applications. From the prevention of physical and mental illnesses, to the invocation of tools that enable us to learn new things with incredible speed and ease, to the elicitation of powerful experiences of mastery and satisfaction. And we are not even talking about what it would mean to control computers, robots and other things with the mind…

This future feels especially possible with the recent breakthroughs in AI technology. After all, if you believe the Twittersphere, at least 10 groundbreaking new AI tools have been released as I write this sentence. There should be some BCI tools in there soon, right?! No, but seriously, have you seen this recent Nature Neuroscience publication where they combine large language models (like ChatGPT) with stationary brain recordings (fMRI) to predict the gist of what people are actually thinking?!

With news like this, it seems like we are getting closer to making awesome BCIs a reality. Now you might ask: “Well, Michael – where are they then?!”. And I might have to reply: “Mhm, feels to me like we are still not quite there, and – honestly – it feels like we are a bit stuck”. The EEG has been around for 100 years – shouldn’t that have been enough time to get there? Like nuclear fusion, it seems like real-world BCIs are always just a few more years out of reach…

Why is that? Sure, we still have some groundwork to do. Fortunately, right now there seem to be more players working on advanced solutions than ever before (seriously, how cool is the work on wearable brain monitoring solutions by companies like Kernel, Neurable, Enophones, and more, many more!). Still, these developments take time. Developing hardware, algorithms, and applications that can be reliably used in the general population (you know, for widespread technology adoption) takes – what’s the technical term – forever! That is why many of the devices mentioned above are not yet on the market. This means that while we are waiting for others to bring something to market, we mere mortals can only do little of the cutting-edge research. But now is exactly the right time for that kind of work!

Second, it also feels like we are still focusing a lot on getting the right kind of signals (for good reasons) and less on whether people actually want to wear the tinfoil hats we are developing. This limits immensely the likelihood that people will put on our devices for data collection. So, all in all, would it not be very helpful for the aforementioned BCI groundwork if we had much broader access to practical and fun-to-wear recording tools with which we could generate large amounts of data? Well, we think so. And we believe in taking the community with us towards even more innovative ExG research!

Are you sending or receiving signals with that?

Well, admittedly, that is easier said than done. With the goal of creating a more ubiquitous and easily wearable ExG system, it immediately became clear that we needed to figure out what greater wearability would look like. You know, get to a point where people would feel comfortable putting on a system in their everyday lives – but also something that would provide the flexibility to study a wider range of phenomena. There were really only three options that we could think of that would meet our goals: glasses, helmets, or headphones. If you can’t imagine that, AI can:

Glasses are a great direction. According to the Vision Council, 64% of Americans wear them every day (!). Furthermore, 85% of all Americans wear non-prescription sunglasses (!). That means almost everyone owns a pair of glasses these days. And there are already some very cool projects on smart glasses with ExG sensors to follow (e.g. AttentivU & EEGlass). However, glasses also have a big disadvantage: They offer very few usable electrode positions – and apart from the face, very few in regions of interest for current BCI paradigms. And placing the OpenBCI on glasses is – er – tricky…

Okay, so not glasses. What about helmets? Helmets are also a cool invention (head-wound-free since ’93!). While a little less on the “always wearable” side, they are used on many average days by many people and are the closest thing to covering the whole head with electrodes as in a lab-EEG system. This allows collecting potentially all possible brain activity information that is available – so it is a very information-rich design. There are already cool projects like this project that integrates EEG sensors into bicycle helmets. But of course, we have the same problem with helmets that we want to solve anyway: How would you feel about wearing a helmet during a Zoom conversation? Not the most stylish option, is it?

So I guess if we really want to get a ubiquitous design, we have to do headphones… Mhm, come to think of it, that could actually be really cool… Headphones are very common, people use them to work, to listen to music, and to look cool while doing it.

Also, headphones provide an interesting coverage for biosignal recordings, as there are many possibilities for recordings in the ear region (e.g. auditory and visual neural processes, heart rate, and muscle activity) and on the top of the head (e.g. motor neural processes and mental workload). Done deal, we do headphones. So let’s take a closer look at what such a system should look like…

Open ExG Headphones – Versatile & Standardized

To develop a quickly usable headphone sensing system, we decided to build on available open source projects. By using such projects’ designs and components, we thought we could jump right into a working prototype without having to do extensive manufacturing inquiries and validation studies (although we did some – more on that below). In short, we developed our wearable sensing system by combining an Ultracortex headset with the headphones project from Print.Plus.

Now, despite our best efforts, this process was also neither fast nor pain-free. Here is an example of an early head-puncturing design that we decided not to move forward with:

In fact, our design process took many iterations after all, and we learned a lot about headphone characteristics, such as finding the right clamping force, rotating the earcups, or balancing material properties and cable routing designs to create a durable yet comfortable unit.

And that was only what we learned about headphones. Adding mechanisms for locking, adjusting, and buffering the electrodes for proper usability took another bunch of time. At first, we tried a purely screw/thread-based electrode adjustment logic – which we had to discard as we learned about the need for spring-buffering our electrodes to keep them safely contacted with the skin. Next, we tried printing springs directly onto the electrodes. And while that looked cool and allowed for a minimal footprint, we later decided to use metal springs because they offer much less compression force and greater durability. But who knows, maybe with a material other than PLA such an integrated spring design might even be possible for a next iteration.

After these rather mechanical challenges were sorted out, we also made sure that the electrode positions adhered to established standards (the 10-20 system for EEG and the around-the-ear positions used with cEEGrid electrodes).  This way, we can do research that is much more comparable to existing work and between studies using this system. Cool right?! We even share these schematics as PowerPoint files if you want to use them for reporting your study layout.

Now here is the kicker: As promised, you can use this system yourself! Like, right now! Even better, you can use it with all three OpenBCI amplifiers (Ganglion, Cyton, and Cyton+Daisy). All you need is one of these amplifiers and an Ultracortex kit, which you can easily convert into these headphones by following our instructions on GitHub. 

To put our money where our mouth is, we evaluated the system thoroughly and just presented our results from two studies at the prestigious CHI conference.

In these studies, we highlight how you can use the system to observe eye, muscle, heart, and brain activity by placing the electrodes in adequate positions:

And recently, we have even started a study comparing the wearability of different wearable EEG systems. Our preliminary results show, that participants were quite excited about trying out the headphones in comparison with other, more traditional EEG systems. Of course, there is also room for improvement, and we are working on those areas. You just stay tuned 😉

Now, you might still ask: But what exactly can you do with this system? Well, there’s a lot of potential in wearable EEG and earables research (see this literature review by Röddiger et al. 2022 for an overview of current earables research). But to make it more tangible, let’s focus on something we find particularly promising – changing the way we conduct meetings.

Meeting Me, Meeting You, Meeting BCI – Wohooo!

Do you also hate meetings? Or perhaps, too many, too long, or unproductive meetings? 

We think there is a lot to improve about meetings. And that using a headphones ExG system could be a very powerful tool to do so. Why is that? First of all, did you know that during the peak of the first Covid-19 lockdowns, Microsoft conducted an EEG study on how people become fatigued from prolonged participation in virtual meetings? Specifically, they learned that their EEG recordings showed increasing levels of fatigue, especially when there were too few or too short breaks between meetings. When these breaks were taken, the EEG levels showed a clear recovery effect. So what if we use ExG headphones to accompany meetings and end them somewhat automatically when everyone gets too tired?

Second, did you know that a lot of research from MIT and others has looked at what makes a good team meeting? Interestingly, a very strong predictor is how evenly the speaking time is distributed in the team. If everyone gets to speak, it’s usually a good team meeting; if only one person speaks, it’s often a one-man show (oh how we love them!). Now, if you don’t need to know exactly what is being said, then we could use ExG headphones to classify when a person is speaking (in fact, we ourselves published research on around-the-ear cEEGrid electrodes that show such classification is possible) and provide feedback on the distribution of speech in a particular meeting. Pretty neat, right?

Finally, if we collect this information over time (and not just in a single session in the lab), we might reasonably learn how to better structure our days and social interactions. To build healthier routines and stronger bonds with our colleagues. Ideally, all that would take is putting on a pair of headphones.

Takeaway

Ok everyone, we are not saying that these applications are now a piece of cake to build. Rather, we highlight them because we believe they are achievable based on known findings and recent evidence. However, in order for something like a Meeting BCI to become a reality, we will need to collect a lot of data. As mentioned earlier, ExG applications are still difficult to develop. That’s why there aren’t that many you can buy – yet! If we want to get there, we should get more people involved and make it much easier to collect more data in a flexible but standardized way. That’s why we realized this project. Because it gives us a chance to do our studies on these issues right now and spread the technology. So yeah, we hope to bring you along, to enable growth and to challenge the status quo.

How are we doing? Be sure to let us know or contact us if you have any feedback, questions, or ideas for moving this project forward!

One last thing: We make these tools freely available to everyone, along with instructions on how to build and use them. At the same time, we realize that some people may prefer to buy a complete kit to speed up the process. Therefore, we are considering offering such a kit – if we get enough requests. If you are interested, please send us a short message at [email protected]

Links

• Materials and Assembly Instructions: https://github.com/MKnierim/openbci-headphones
• Evaluation Study: Michael Thomas Knierim, Daniel Puhl, Gabriel Ivucic, and Tobias Röddiger. 2023. OpenBCI + 3D-Printed Headphones = Open ExG Headphones – An Open-Source Research Platform for Biopotential Earable Applications. In Extended Abstracts of the 2023 CHI Conference on Human Factors in Computing Systems (CHI EA ’23), April 23–28, 2023, Hamburg, Germany. ACM, New York, NY, USA, 7 pages. https://doi.org/10.1145/3544549.3585875
• More of our Earables Projects: https://earables.teco.edu/