Defective Muse?


I am a Windows developer and I am having fun making some very pretty displays of my Muse data as received from Muse-IO.
I have tons of ideas for applications but I am especially interested in the field of sleep.

I had my Muse for 2 weeks now and I believe I may have defective hardware on the left-forehead channel.

I have the following reasons to suspect a hardware problem:

  1. raw_fft1 is returning a value of “minus infinity” intermittently, once or twice a minute.
    This value was corrupting my arrays and downstream calculations.
    I am now trapping the minus infinity, but my left forehead display appears out of sync with the other 3 channels.

  2. delta_relative is ALWAYS returning minus infinity for the left forehead. Same for all 5 wave types.

  3. The CSV file created my Muse-Player is showing things like…
    1419357445.111000, /muse/elements/horseshoe, 1.0, 4.0, 1.0, 1.0
    1419357445.111000, /muse/elements/is_good, 1, 0, 1, 1
    1420489471.824000, /muse/elements/alpha_relative, 0.281212806702, nan, 0.0927164703608, 0.304776340723
    1420489471.824000, /muse/elements/beta_relative, 0.29111045599, nan, 0.571482121944, 0.367583781481
    1420489471.824000, /muse/elements/delta_relative, 0.212421387434, nan, 0.0208893101662, 0.0661793574691
    1420489471.824000, /muse/elements/gamma_relative, 0.097597964108, nan, 0.275857150555, 0.183689847589
    1420489471.824000, /muse/elements/theta_relative, 0.117657385767, nan, 0.0390549302101, 0.0777706727386

I wonder if my Muse needs to be sent in for repair?


Hi Douglas,

Try using the “horseshoe” indicator (a.k.a the headset status indicator, the round white icon with the different electrode indicators on it) in either MuseLab or the Calm app to see if you can adjust the headband into a position where it can get good signal on that channel.

Are you wearing the headband in any specific context - you mention you’re interested in sleep - like lying down or moving around? Or are you simply sitting still?

The left forehead channel is the leftmost silvery contact on the inside of the headband. Ask a friend to look down on the headband from above you to make sure that it is contacting your forehead stably.

What does the time domain EEG signal look like for your headband when you are wearing it and sitting still for a few minutes?

If you are still having issues, please contact to initiate the return process.


Hi Tom,

Thank you for the suggestions. Today I spent a few hours further investing this problem. I tested with MuseLab, muse-player, and my own software. I am seeing the same problem in all three programs. Interestingly, after more than ten minutes of continuous use, the bad channel (horseshoe) will sometimes come alive and start to work. Perhaps the problem is temperature related. I do not own any cell phone, so I did not test Calm. I modifed my program so it will ignore data from any channel whose horeshoe is >1.

The problem did not just happen at some point, but its been there since out-of-the-box. I will make arrangements for return.

It would be nice if Muse-IO would trap the minus infinity numbers and not send them. That way, destination software would not be required to detect and handle the corruption. CSV files written by muse-player should never contain “NAN” cells.

I have not taken my muse to bed yet:) However, I believe the sleep/insomnia market may be HUGE, much larger than the relaxation/meditation market. The potential of the sleep market may well justify a more sleep-friendly form factor. I will start another topic to elaborate on my opinion.