OK, I think this is a little misleading at a glance. From my read they didn't show a video to the subject then create images from the brainwaves. Instead they showed videos to subjects, recorded the response. This allowed them to greate a mapping of response to video. You then show videos later, read the response then look up the video.

The article states; "The reconstructed videos are blurry because they layer all the YouTube clips that matched the subject's brain activity pattern."

They could have just shown the best match, this would not have been a cool blurry image that is very easy to misinterpret as being generated from the brainwaves directly. More than a little slippery.