Researchers at USC have modeled the transforms associated with production of memory. It's a black-box: they aren't interested in the subjective experience of the transform, just how inputs (short-term memories) are turned into outputs (long-term memories). The article speaks of prosthesis for memory disorders, which is certainly the first market, but once you've figured out the IO model for memory you're in major-league Philip K. Dick territory. What happens when you record my inputs and play them into your hippocampus? Do you end up with a personalized version of my experiences?