It appears they have multiple drivers in the ear cup. Odd.
I'd have thought it would be better to process audio in a virtual sound-stage to create a binaural down-mix. But that's probably still bad at front-back differentiation without incorporating a model of the wearer's pinna.
It's all bollocks though, isn't it? Ambisonics was developed in the 1970s and thoroughly forgotten about by the time the hardware to do surround sound properly became affordable, so we're repeating all the quadraphonic-era mistakes with a few more channels.
