Transactional analysis identifies "parent, child, adult" and Freud identified "id ego superego" and then there's the developmental stuff around self-states or sensory states, etc.
Basically they're less developed than "people". We also consider sometimes "state-dependent" states. Sounds redundant.
States in this case of state-dependent is kinda the "mode" or environment one is in and what "parts" of a person are attuned to or attached to that environment. So when you are in the kitchen and think "Oh I should put this in the bathroom" and you are then in the hall like "Why did I just leave the kitchen?" then you walk back into the kitchen and are like "OH! I wanted to put this thing in my hands in the bathroom!" — that can be one type of "State-dependent memory" i.e. place-related.
Another type may be "who you're with" or "the mood you're in" or "the time of year" — and so memories of being with Bob float to the top when you're with Bob but recede when you're not. Feels & memories when you wear a favorite item, smell a familiar scent, etc. — associated states that come along with various environments or sensory experiences.
See also environment-dependent states.
Memory has been noted to be an intricate web where memories are grouped by associations and linked to one another very much like webpage links.
One does not have access to all of one's memories at any given moment. Some memories are closer to the surface than others. Some are immediately accessible, some are readily accessible given a moment, and some will take a lot longer to call forth if one tries.
There's a few mechanisms that determine which knowledge and memories are closest to the surface at any given moment. One is called "priming" in psychology -- something in the environment has called forth a bunch of associated knowledge and memories, and now that information is closer to the surface than it might otherwise have been. Priming can last for minutes, hours, days, etc.
Thus whenever one changes environments and thus states (see that there page, environment-dependent states), one also changes the repertoire of available knowledge and memories. This is normal for singletons.
To take the concept one step further for multiples, who one is at any given moment is, in some ways, an exceptionally complicated state. Each system has associated memories and memory groupings of things that happened while in that "state" i.e. while that person was front. It is thus easier to recall what happened when Joe of Joe& was front last week if Joe of Joe& is front now. If Sue of Joe& is front, she will have to poke at Joe's memories (if there is shared access) or ask him internally to recall what happened when he was fronting.
(I will eventually tie this all together better, but I'm not up to writing a thesis paper right now. honestly, there are places to go with this and the other section -- Crisses)