3 ms·
I mean that here-and-now is an aggregate of discrete attention spans. Without integration these spans are separate "now" without "here". Attention directed to f
by ganzuul 2y ago
I mean that here-and-now is an aggregate of discrete attention spans. Without integration these spans are separate "now" without "here". Attention directed to future events is not truly separate from those events (coupling constant unknown), doing away with some conventions about mind-over-matter/matter-over-mind. - When we make plans part of us goes to a future "now" which is not "here" yet. The future already exists in some uncollapsed quantum world which our attention can go to but we don't have a means of navigating our conscious here-and-now to that future unless we construct the path there, which we do in the conventional sense. This means that things which we are able to imagine are not dualistic illusions but real to a different degree than what we have been assuming 'real' is.
To me personally this philosophy makes for a richer and more rewarding inner world.
Computational irreducibility does away with any practical significance the simulation hypothesis could have, and it provides structure to plans-turned-paths-to-the-future. Your attention into the future is one degree of irreducibility, your consciousness is another, and the zeitgeist of our age is a third, resulting in an ego with three shells.
- rokkamokka 2y agoI feel if this was true, it would be provable by experimentation. If it's true but not provable, it also doesn't seem to have any impact.
- jazzyjackson 2y agoa framework need not be provable to be useful
- nuc1e0n 2y agoQuite. Provability can often come long after utility is established.