Almost any motor imagery will cause a small drop of amplitude in the scalp recorded EEG on the middle opposite side of the head. So, the number of such may be pretty arbitrary.
More important is the question of how many different movement types can be reliably discriminated between. That in turn depends on how reliable the discrimination between imagery types needs to be. So it depends.
With a good subject and lots of training, IMO one might expect to tell apart each hand and each foot movement visualizing about 3/4 of the time. This is not just a limitation of the MIV headset versus another headset.
Yes, you are correct. What I should have asked was how many different movement types can be reliably discriminated between.
In the example of someone using a BCI headset and flying a small drone/helicopter, would you say the number of different movements types discriminated are at least six?
* up / down (z axis)
* left / right (x axis)
* left / right (y axis)
* slow / fast (speed?)
If yes, are these six movements able to be reliably reproduced?
One obvious caution with a drone, is that when the classifier makes the inevitable mistake, drones flying into humans or other objects would be extremely unpleasant. ;-)
I would say that six parameters (maybe visualizing moving the face, arm and leg?) would be very, very hard to do with sufficient reliability to control a drone. Why not try it with two (left, right) and add more axes if the left versus right works out?
Comments
the example of someone using a BCI headset and flying a small
drone/helicopter, would you say the number of different movements types
discriminated are at least six?