Can the data integrity be relied on?


Although I generally trust the output data of the Muse, I wonder if sometimes my less successful sessions are the result of faulty data output by the Muse itself. For example, I’m pretty confident that I can now begin a session in a calm state (thanks to Muse), but sometimes the winds pick up to gale force immediately, even though I am relaxed and feel I’m focussed on my breath. This leads to more brain activity on my part "is it me or is it the Muse?, "what could be causing the high winds? “should I even trust the data in this session” etc. so that in some sessions I feel that my poor performance is the RESULT of the Muse itself.

I understand that there must be a certain amount of “wiggle room” in the software in that it doesn’t automatically end the session as soon as one of the sensor readings is lost for an instant, but I wonder if we as users should also be able to see the tracking efficiency of every session to give us an idea of the integrity of the data?

In a nutshell, if I’m feeling calm and focussed at a given moment, but the Muse is telling me otherwise, which is correct?


How do you determine what is correct? To track efficiency, I would think that the data in question would have to be compared to baseline data.



Thanks for your response. I’m not a trained technically, so this is probably an oversimplification, but if there are seven sensors utilized on the Muse over a given session, would it not be valuable to know how many sensors were actively polled for what percentage of the session? As I mentioned in my original post, the session doesn’t automatically end as soon as a single sensor isn’t polling for an instant, so a percentage value (all sensors polling for the entire session being 100%) may give the user a frame of reference as to its accuracy.

Tonight for example, I was receiving high winds indicating very strong brain activity, before the session ended with the message that the Muse had lost communication with its sensors. If it had maintained only slightly better receptivity (not enough for the software to pause the session), I would have completed a session showing high brain activity (low score) but without any indication that the software had tracking issues at the time.


I believe that the middle 3 sensors are used to set a baseline and are not specifically channels to be read.

I’m not sure if it is possible to operate a Calm session and capture the raw data at the same time.

I have been working with data from the sensors using muse-io and Muse Lab…in muse-io ,when there is no data coming from a sensor, the value “NaN” is sent. If you were to record a session with Muse Player, these NaN values would likely show up in the data stream.

I would do that, but the install process for the Mac SDK is still broken and MusePlayer requires that the SDK be installed.