From doing this I have a few theories about memory (All from my mind. We could record memory different.):
- Memory is read once write many. That is, every time we retrieve a memory we rewrite it back, modifying it slowly.
- The emotional state of the recorded memory combined with the last few recalls determines how easy the memory will automatically pop up, without digging for it. This is based on the emotional state while retrieving the memory. (eg, studying in a class room increases test scores do the environment being the same. Depression, manic depressive, DID, and many other emotional states determine what memories or thoughts will arise in the mind next in relation to the current moment.)
- Memories of just the individual moment (usually slices in time or elements of spacial awareness) are in a sort of tree structure. Details of an event are recursively retrieved while the most important parts to that moment to the individual usually sit higher in the tree. (eg, near photographic memory shows a person who constructs a deep tree, and usually recurses it quickly.)
- The key aspects of a "concept" (a unconscious part of a knowledge instance, not the whole piece) are retained through the distance of neighboring similar concepts (metaphor, or literal), and by how many neighboring concepts there are. This makes a graph. I originally suspected you'd want 3 neighboring concepts, but 2 seems to work perfectly fine for long term memory retainment (if not 1 neighboring node), as long as the neighboring concepts are not too distant. The distance between concepts strongly determines the "pressure" on the mind during learning, the resolution of the knowledge learned, and ultimately if it is retained or not. Oddly, this seems to be somewhat black and white for long term memory: you either retain the concept or you do not, even if that concept is not used, as long as the mind has some level of importance tied to it enough to record it to begin with.
- The finer details of what the patterns the mind processes is a difference between two or more patterns in space and/or time. This functions similar to a video codec where the entire detail is not recorded, but just the relational change. Without this difference between two or more patterns, we may not be able to record anything. (eg, the static on an analog tv. The difference to 'not-static' is recorded, but it is a higher level concept. The individual squares on the screen are not recorded into memory, but instead only the details that can be pattern matched to a neighboring pattern.)
- Neighboring aspects within concepts can and do bridge multiple domains of pattern matching sensory experience, but usually do not unless necessary. I've labeled these axoms and they take longer to learn / more pressure on the mind than normal concepts. (eg, learning addition is more difficult than learning multiplication, because addition crosses from the abstract into the concrete, but multiplication only requires the knowledge of addition to be learned.)
- The name of word that represents a concept can be learned through neighboring word concepts, as it is somewhat like a concept on top of a concept, but I've found the easiest way to remember the name of a thing or the vocabulary word coupled with the concept is to build a story for it. So far I've found two kinds of stories that work best: 1) Humorous metaphors, as humor for some sort of reason creates a strong memory. and 2) An interesting story about it. This could be a mythology, usually a fiction with metaphors to tie it together (see Memory Palace) but for me, my personal interest is etymology: _Why_ is that thing called what it is called? If I have a story for the thought process of why it is called that, the name is recorded. This can be fiction or nonfiction, but I find nonfiction to have the added benefit of understanding the thought process of others or a moment in time, which is a cool added bonus.
- When written down or said out loud during a memory record, it solidifies far better. I have a few theories as to why that is. One of them is akin to 'rubber duck debugging', where writing it out unchunks assumptions and forces one to walk over the entire structure.
- If within a single domain over 220-250 concepts are recorded, either headaches can happen, or old concepts will fall out, much like an LRU. This most likely ties into the size our brain, as that determines how many people we remember (roughly ~250) and the numbers seem to line up here too. By creating categories for subjects this 250 element restriction no longer applies.
This, as mentioned above, is a tree structure being recursed upon and written out in list format, but because there are subtle details beyond these points, I'm probably forgetting a lot that is there in my memory, but I don't have a reason to dive into more detail. (eg, how examples can concrete abstract thoughts solidifying them, ...most etymology for vocabulary is visual, if you go far enough back, ...)
My recall of 300 concepts learned in 3 months is easily above that of 95%. Two years later, with no use of the concepts and retainment is still clearly above 90%, but the etymology of the vocabulary in certain domains (namely logic) seems to fade for me.