This is wild. These folks at UC Berkeley and ATR Computational Neuroscience Laboratories in Kyoto, Japan, have discovered a way to analyze fancy MRI images of your brain and extract the image that you saw.
The article, The Brain Records, Scientists Retrieve What the Eye Sees , is short on details, but it claims that the image on the left is what the test subject was looking at, while the image on the right is derived from the subject's brain activity.
A couple of thoughts... First, wow! This is really amazing. This is a close as I've seen to reading someone's mind.
Second, if you look at the example picture, it seems that the picture in the mind is very different from the source image. Is this a mistake in the publication? Is it that this person really sees something different than the image he is presented? Just look at the background. All the detail is gone. What about the hat? Its missing and there is clearly someone else's hair there. Is this a memory being mixed with the image?
Third, what would happen if you mixed the images they are getting from the fMRI with the de-blurring algorithm from my last post? The answer is that you wouldn't get much. The de-blurring algorithm makes many assumptions about how the image got blurred (like that it was a result of camera motion) in order to do its magic. There is no similar basis for this blurring, so the algorithm will likely not be able to do much of anything.... still its an interesting concept for a science fiction story, dont you think?
UPDATE: There is more information in this article.