In any case it must be possible to retrieve a memory with little effect on displacing old memories. A chemical concentration shift is what one would expect of evolution to signal ‘remember this’. Highly distributed chemical dispensers signaled by nerves might be fast enough. I bet the modern brain has superseded that stage.
Hawkins introduced me to the ‘frame’ which is some situation with parameters. The frame makes predictions and when some frame has been activated and the predictions are not met we may have a situation where we record short term memories as longer term memories, probably in a different technology. Certainly high ‘valence’ surprises (‘really good’ or ‘really bad’) warrant long memories so as to influence future behavior to increase or decrease repetitions of the frame. The frame notion suggests that some part of the head is always running ahead a few frames looking for dangers or opportunities, or ‘landmarks’. Kanerva’s ideas greatly extend this sort of idea; frames are linked together with pointer structures that a programmer would recognize.