That's the question put forward in the article itself, just not framed like yours. The question is if the neurons are just like storage spots then there has to be some mechanism that we don't know yet that does that transfer without losing the information (in the case of RAM, we give the instruction to transfer). That not-yet-known thing is the main mystery here (that is if the neurons are just like storage spots. They can be entirely different as well).

I'm working on this at the moment; Reconsolidation should work, but requires that representations/memories be periodically re-activated so that the neurons can re-learn how to encode them. Still very preliminary: https://www.biorxiv.org/content/10.1101/2021.03.08.433413v1....