I'm interested in human-like storage layers and wondering if anything has been done here. Content-associative memory is one idea but doesn't go far enough. Ne

I'm interested in human-like storage layers and wondering if anything has been done here. Content-associative memory is one idea but doesn't go far enough. Neural nets are a lot closer but at the moment they seem limited to pattern matching rather than storage and exploration, although I guess you could say supervised training is equivalent to storage. I'm thinking of a DSL on top of a pure associative memory, which will remember Things with weighted connections to other Things. Matching consists of not only showing the idea but also related ones more distant. Is there anything like this?