The potential for computers to interface with the human mind has long been known. Systems with sophisticated algorithms can recognize eye movements or minor muscle twitches, allowing individuals who have been paralyzed to operate a computer. For some of these individuals, it is the only way for them to interact with the outside world, allowing them to speak via the written word.

Researchers at UC Berkeley have taken a monumental leap in the ability of computers to interpret what a person sees and even dreams, mapping fMRI (Functional Magnetic Resonance Imaging) data from a test subject’s visual cortex. This data was then reconstructed by a program culling images from a massive database of images (in this case YouTube). The test subject’s reactions to the newly presented material was gauged until the program had a “match” for each image. While the resulting composite images are a distorted and fairly crude reconstruction of the source images, the fact that this is possible at all is staggering.

The good folks at Gizmodo have provided a more in-depth discussion of the experiment, as well a video of the reconstruction for three of the subjects.

It’s still early to try and quantify the effects of a discovery like this, but the implications for the medical field, law enforcement, and academics are far and wide. Imagine a world where you never needed a camera—you could simply go home and upload your memories. Once that pipe dream was the stuff of science fiction. Now it’s a pipe dream that a computer program can interpret and play back to you.