Think of the mind’s eye like an incredibly complex and totally unknown biological video codec, with the MRI we can see the raw data of the resulting files but don’t know how the information it contains corresponds to the input. So we show a camera the uses that codec a known image and model the observed output, creating a crude map (or as the researchers call it “dictionary”) of how the codec functions. What is being modeled is vastly more complex than this analogy suggests because the brain reconstructs the viewed image in different areas and looks for symbolic and gestaltic relationships far beyond anything we’re used to encoding in a video file, and the MRI images far slower than the videos shown or the brain processes that are interpreting said video.

It gets even more complex because what we see isn’t necessarily what we get, the brain can hallucinate, block out or hide elements in plain sight, and dream. This is just the very, most elementary beginning of being able to, in essence, read a mind with technology.