I believe it's an anchoring function, like Ni. Ni anchors to a symbolic vision rooted in the future, as a talisman for transformation (of themselves or the world or whatever). Si anchors to specific, sense-oriented memories rooted in the past, as something concrete to hold onto/memory, to preserve, or to bring the what is treasured in the past into the present. Ni transforms and pushes into the future (because its anchor is in the future), Si preserves and stays rooted while circumstances fluctuate; they both seem to be used to anchor people, just with different goals and reference points.
This is what you 'use' them for, I think. I think passively they act as storehouses too, 'worldviews' (from podlair). Ni stores information about the world in patterns, shifting patterns underneath the surface grid of sensory reality, so that each event can be seen as simply a variant of a timeless pattern (and therefore outcomes can be predicted). So there's Se on the surface, and Ni underlying it (bad way to put it, but roughly what I mean). With Ne-Si types, our worldview is concrete, not pattern-based. Thousands of little data points that we connect bit by bit. Our 'underlying reality' is a bunch of discrete events, and we connect them on the surface as they occur with Ne. Architect linked to an article about our learning (about the world - broadening and deepening/stabilising understanding) being maze-like/trial-and-error. Going down multiple little offshoot paths to check if they're the right ones. (Can't remember if the article was about Ne, but it applies.) Ne can accelerate the process, maybe, but can also stall it when it casts its net too wide.
So in navigating the world, both Ni and Si tell us what's going on and what's likely to happen, based on different storage systems; they also tell us what *should* happen.
This is what I think so far, anyway. There's more but it's even less clear than this. I hope some of it made sense at least.
I find it a pity that Si is dismissed as simply a love of details, or a good fact-oriented memory. I think the latter trait especially has nothing to do with type. I don't think it makes sense for types with a naturally weak grasp of a function to dismiss it based on their own limited understanding and control, though it *is* typical of humanity. We should be getting our understanding of functions we don't understand or use well from people who live and breathe it naturally. Fe is quite fascinating too, I think. We demonise it because we suck at it and because the levels we use these tools at are way beneath their potential, so whether we're dealing with our output or others' input, we interpret the process and data transfer at a very low level. I think at the top, each function is amazing.
I think types with lower Si often express it as a glum adherence and acceptance of reality, but with a stiff upper lip. Sort of british in nature. They also tend to be skeptical - when you talk to them there tends to be a movement down 'back to earth' in their thinking style. Common-sense. And I *think* as they grow older they seem to produce anecdotal evidence more to support their views. I see this with INTPs. Si-Fe, as it develops through life, starts to connect and solidify its little underlying data points on people and relationships with less of the younger INTP's focus on exploration and withholding certainty till enough data has been amassed (Ti-Ne fear of incomplete ie inaccurate models). Their data congeals into understanding that comes from mental exploration and solid experience. Their minds stabilise a little and became weighter, more grounded.
Types with high Si like to use Ne to play, I think. Ne seems to be naturally playful, but high Ne is used for change and growth as well. Low Ne seems to be used for fun, just surfing off little patterns in the present that they see, for banter, occasional spontaneity etc. Might have more to say on this later.
High functions in general seem to be used to output something transformational. Active use, rather than passive use.
