A friend of mine got one, so I might get to try it out
do keep in mind devices like these are just like OCR: they recognize patterns based on training.
So no, imagining an object will not suffice to bring it onscreen, unless you had the object created and the system trained to the thought and that training associated to the creation of the object beforehand. Just like handwriting OCR, it won't work out of the box.
Also, positioning isn't simple. A single thought pattern will be detected, which means "move here" is a different thought than "move there" and thus cannot be abstracted as a directed motion command.
SO.... you'll have "move right", "move left" commands instead of "go over there" commands. I saw this in a test video where they attached mouse motion to thoughts. It was veeeery slow, since thoughts are gradual you cannot attach speed to recognition percentage and this forces a on-off only switch.
So yeah. It's still cool though, and recognition seems MUCH improved over what I last saw (single shot 8-second training? Wow.). Also, no gel
I would have liked a demo with a non-bald guy though.