Interesting concept with some odd gaps in logic, and a very ambitious goal that technology will struggle to support. At this point at least, it would be extremely difficult to get the experience right – a lot of assumptions need to be made about mood and mood-music mapping, even if you could train and personalize your device to “know” your music choices based on your mood. Further, just how is mood being assessed? EEG patterns are constantly changing – I’d love to know how they’ve chosen to interpret a given mood (that lasts longer than a few seconds). It could be interesting to combine the headphones/app with users’ Pandora accounts and other personal data. In any case, given the high expectations of a product that is supposed to read your mind, it’s likely many users won’t have a lot of patience if it isn’t “working” from the get go, though the novelty factor could carry early adopters through a training phase. Cool concept, tough road ahead.