For a paralyzed person trying to use Neuralink to control an external device, how does the neural signal decoding distinguish between just thinking of a movemen

For a paralyzed person trying to use Neuralink to control an external device, how does the neural signal decoding distinguish between just thinking of a movement and the actual volition of that movement? You would not want to actuate the external device just because the user is thinking of the movement (e.g. from reading a book, watching a video, or planning for the future).