Skip to main content

First thoughts on Notion DK1



  • Suz Hinton

    Hi! Super early thoughts having unboxed this 90 minutes ago (it's charging so I'm spending the time on this feedback while I wait!) -

    1. It was very straightforward to get started

    2. For a Dev Kit, everything including the online resources and the Github repo felt more polished than I would have expected

    3. I had zero issues running the Hello World calm subscription example

    4. The ethics section of the documentation is brief but an extremely welcome addition. This is a large reason why I support Neurosity. Keep up the great work.

    I have some questions that kept popping up when I was playing with the device tonight. I'll detail them below.

    What I'd love to know is more about where the boundaries of data processing are, and exactly what the device uses internet connectivity for. For example, is the internet only used to login and send device status (battery, os version, etc), or is there more to it? Theoretically, how feasible is it to use this device on a wifi network that is not internet connected, and is just a local network?

    As far as the calm, focus and kinesis thought training features go, is this all processed on the device locally? Does the device run the training that you recorded on the device itself (does it download / store the resulting model) or is it processing elsewhere? Right now I am assuming that the device just reports calm events instead of raw data to the SDK runtime if you subscribe to them, and not if you aren't connected to it. Is this right?

    Sorry for all of the nosey questions, but knowing more details around the boundaries helps me understand what is and isn't possible with the device. I'm excited to dig in further this weekend when I have more free time!




  • Suz Hinton

    More thoughts that I forgot to include about the hardware:

    1. It was way lighter and smaller than I was expecting. Nice work!

    2. I can see the potential of what the finished production device will look like and it's cool

    3. It's comfortable to wear

    4. I'm not sure what the cryptic led flashing is from time to time

    5. The case is really well designed


  • Alex Castillo

    Hi Suz,

    Thank you for the feedback! We are glad you had an overall positive initial experience. To answer your questions:

    1) Internet vs. local connection

    Currently, all communication between the device and the API happens via the internet.

    We plan to add a local wifi transport to the API in the future. I would love to learn more about potential use cases and how this feature could help users.

    Another option is to access EEG data and Markers via Lab Streaming Layer (LSL). LSL uses your local network to stream data but it doesn't include any metrics. This option is usually helpful in a research setup. LSL can be enabled in the Developer Console via device settings.

    2) Processing metrics

    All metrics including calm, focus, and kinesis, are completely processed locally on the device and only when you subscribe to them (using NotionJS or via Developer Console). The device then reports the resulting metric event via the API. 

    Lastly, the LED flashes once the device is charging and therefore in sleep mode. 

    Is this helpful?

  • Matthieu Riou

    Hi all,

    I'll follow Suz's example (charging too) and post a few thoughts and remarks, in somewhat random order:

    1. I can second Suz's compliments and had very similar questions.
    2. Regarding local API communication, on my end the principal concern would be privacy. Sending data about my thoughts, even if it's just my left foot to start with, can become concerning for the future. And it's not that I wouldn't trust you guys, but then it raises more questions about transport security (SSL? version? etc.) and eavesdroppers.
    3. Putting on the device feels a little "spiky" at first but you actually get used to it faster than I imagined.
    4. I was surprised about having to change electrodes. Totally made sense once I read the doc, just not something I was expecting. I found how to remove them, it's simple, but perhaps a guide would help others.
    5. After just 5 minutes of testing with the dashboard, I wasn't able to get the signal quality to green on all electrodes, just a couple flashing green occasionally while everything else is orange. Should I worry? Perhaps a little more guidance here would be helpful.
    6. Speaking of the dashboard, it seems to use quite a bit of CPU (40-50% on my machine). Not a big deal, especially at this stage, but Chrome already killed the page a couple times due to freezes.
    7. Can the device be used at all while charging? It seems to get in sleep mode automatically. I'm not sure what the autonomy is yet, but I could see that affect productivity during development.

    That's all for now, I'll be doing some more poking around during the weekend. Thanks!

  • Jim

    Great feedback so far from Suz & Matthieu. Mine may get pedantic so bear with me:

    1. When I first get started, the Signal Quality was never "all green", which subconsciously made me feel I didn't wear it right. What was the official recommendation? Should I have tolerated a few yellows? The best I ever got was 5 greens.

    2. Was there a benchmark/standard about the value for awareness/calm metric timeline? what does the value mean?

    3. I realized I could change between "awareness/calm" & "awareness/focus" for "Metric Timeline" but not for "Metric Predictions", though I could go with both calm & focus for Predictions.

    4. for the "Getting Started: Hello World" tutorial, did the screenshot below "Then run the program with npm start in the CLI. If all worked, then you should see:" supposed to show the "Logged in" successful message? if so, it's either missing the right screenshot or cropped.

    5. I click the two links "learn how to train" and "API section" in the Getting Started: Hello World tutorial and both links are broken.

    Just getting started, happy to help reproduce issues or clarify my feedback!

  • Alex Castillo

    Hi Matthieu,

    Thanks for the feedback!

    2) Local API Communication & Transport Security

    Local API communication is something we've had in the roadmap for some time and look forward to adding to the API. We've designed the API with this in mind and have the codebase in a great place to add this feature. You can follow the progress of Local WiFi Transport here:

    Once we launch it, you'll be able to enable this from the Developer Console or via the API with one line of code.

    At Neurosity, privacy and security are top priority.

    That aside, some context on our transport security approach: In a nutshell, we use authentication, authorization (ACL), SSL, and encryption. What type of security would make you the most comfortable?

    3) Device Fit

    We also had a similar experience where the device seems to loosen up and adapt over time. Also, you can always extend the sides in order to increase flexibility. My preferred setting is fully collapsed. I know AJ's setting is extended to level 2 or 3. I would suggest trying the different settings until you find the ideal one for you.

    4) Replaceable Electrodes

    I'm glad you figured out how to replace the electrodes. This will allow us to iterate on electrode design faster and help ensure you get the best possible signal quality. I've added a ticket for us to add a section on "How to replace electrodes" to the knowledge base. Thanks for the feedback!

    5) Signal Quality

    We'll be adding more information on ensuring great signal quality to the knowledge base soon. In the meantime: Green signal quality is ideal. Orange means there is room for improvement. Gray means the electrode is not making proper contact. Some things that help go from orange to green are:

    - Ensuring enough pressure is applied to the head. This can be accomplished by making sure the headset is sitting at the right angle: roughly 45 degrees. Additionally, extend the headset to your ideal length.

    - Ensuring the electrodes are going through the hair. This can be accomplished by slightly combing the headset in a circular motion until you can feel the electrode prongs.

    The headset was designed to feel "right" once all electrodes are making contact and applying an equal amount of force to the head. I'll post here once we add a full guide on this.

    6) Developer Console (Dashboard)

    We'd love to reproduce this issue and get it fixed. I've gone ahead and opened a ticket. Can you please add steps on how to reproduce?

    7) Can the device be used at all while charging?

    No, the device is automatically set to sleep mode while charging (or upgrading the OS). We designed it that way for safety reasons. We are working on extending battery life for the next versions. Would this help?

    This is great feedback, Matthieu. We've created action items and we'll be updating everyone soon on our progress.

  • Alex Castillo

    Hey Jim,

    Thanks for the feedback!

    1) For signal quality questions, please see my comments here.

    2) Calm score

    The calm score is derived from your passive cognitive state. We based this metric on the alpha wave. The calm probability goes from 0.0 to 1.0 and the higher the score the higher the probability. To get the calm score over 0.3 is significant. Things that can help increase the calm score are:

    • Closing your eyes for 30 seconds or more
    • Seating or standing still
    • Breathing exercises
    • Meditating

    Is this helpful?

    3) This is correct, metric predictions allow you to preview multiple metrics simultaneously. You can always remove a selected metric by clicking the "X" next to it.

    4) You are correct. The log is not visible in the screenshot since the terminal is not open. We'll get that fixed.

    5) It seems AJ already fixed the broken links. Can you try again? 


Please sign in to leave a comment.