> I dont remember the thread but the context I think was regarding reading memories digitally somehow.
Funnily enough, apparently the language in the different parts of memory process (i.e. encoding+storage+retrieval) stems exactly from "digital computer memory" :), psychologists in the 40s/50s looked at the development of computers and started making analogies with the brain.
> I wonder if revisiting memories often enough will reencode them before they are lost?
Well, you have Ebbinghaus' forgetting curve, which shows how (in this case, semantic) memory decays with time but, with rehearsal, the memory trace (i.e. the neural substrate) gets "stronger" ("neurons that wire together fire together") and lasts longer in memory.
There are multiple types of "forgetting" (Schacter's "The Seven Sins of Memory"), what we usually mean by it is the "transience sin", i.e. forgetting because of elapsed time. There are several views why it happens, mainly storage failure, i.e. that memory trace getting weaker, or retrieval failure, being unable to get the right "memory cue" to retrieve that memory.
So, probably, if you start associating one particular [0] cue with the memory, and training that over time, yes, it should be possible to not-lose the memory (I don't think it will help with Alzheimer's, though).
The funny thing is that every time you retrieve a memory from long-term storage, it gets re-encoded before it goes back to storage so, basically, every time you remember you are modifying the memory a bit with your current biases (another sin!), thus also forgetting :).
[0] And choosing the best cue is already a hard topic, as it should be different enough as to not trigger many different memories ("cue overload principle" from Surprenant and Neath's "principles of memory"), and the memory should be not so different enough that you may get the risk of other singular memories interfering with it.
> every time you retrieve a memory from long-term storage, it gets re-encoded before it goes back to storage
I don't know how literal you're being (e.g. read once, fully overwrite previous), but it's a good model for "reinterpreting past events, with a new perspective".
This may be a tool/strategy for therapists -- I've spent zero hours in clinical psychology classes.
But anecdotally, very few people are open to reflecting on past events with greater charity for the remembered villain of the story. :)
Well, another one of the "sins of memory" (categorised along the sins of <i>comission</i>, instead of <i>omission</i>) is <i>bias</i>, which means modifying the actual transpired event with our beliefs and previous knowledge, either at encoding or during retrieval.
So, if at retrieval time, your beliefs (e.g. now you support legalisation of marijuana) and knowledge are different than what they were during the memory encoding (e.g. you didn't support legalisation of marijuana), because you see yourself as "consistent" you may actually remember the memory tinted with your actual beliefs (e.g. you were a supporter all along).
Furthermore, as we lose the complete experience details from our episodic memories, we start filling the gaps with our current knowledge and beliefs, too, to achieve some consistency of the event...
Quite interesting, but obviously, lots of variables and different things come into play in this topic.