Buzzing instead of beeping. There are two small vibration motors in the band, with a documented set of up to seven effects you can trigger. No shipping product uses them. Every competitor gives feedback through sound or a screen, and both pull you out of the exact state you’re trying to hold. A gentle buzz doesn’t. Could the calm score drive a buzz gentle enough to run a whole meditation with your eyes shut and your phone face down?
If someone builds that loop, there is a published pattern for how it should behave. The iso-principle comes from Ira Altshuler’s music-therapy work: begin with something that matches the person’s current state, then move gradually toward the state you want. Starcke and von Georgi tested it by inducing sadness with a film clip and comparing sad-music-then-happy-music against happy music straight away; the matched-then-shifted group ended with significantly lower negative affect. Applied here, a buzz would start at the rhythm the session is actually running at and slow from there, rather than opening at the target. Two cautions, both mine: the transfer from music to haptics is untested, and on this hardware the current state is a model score against a drifting personal baseline, so match to the trend rather than to the absolute number.
The friendliest data feed is the one nobody uses. There’s a mode (OSC Perform) that hands you smoothed numbers four times a second, already averaged so they don’t jitter. You can wire it straight to something visual with no extra work. Right now only developers and artists touch it. If your product needs to react faster than about 16 seconds, though, you’re not really using focus anymore, you’re using the raw signal, and that’s a different project. Worth knowing which one you’re in.
Making a score mean the same thing twice. Focus and calm drift with how the headset is sitting, hydration, caffeine, and time of day. So comparing your score to mine is close to meaningless, and nobody has built the layer that would fix that. Whoever figures out how to make my focus score comparable to yours owns the piece every other product on this platform needs.
Sources
My own data: three recorded console sessions (248,320 samples), an audit of seventeen training experiments, and the Notion console and sensor references.
