m neurons each of which could be firing or not sets the upper bound of n as 2^m 😀
Conversation
Replying to
there must be redundancy then or don't have to walk the whole thing or something practically asymptotes or i dunno 😁
1
Replying to
Yeah neural macrostates are going to be a much smaller subset. A tighter upper bound would be long-term memory size as a function of age. That’s harder to estimate though. It’s the compression of life log input.
1
Replying to
concretely speaking, afaikt, older people generally have to meditate longer to get same effects as younger. but it's ofc not linear. there is indeed some sort of compression/chunking but surprisingly lossless-seeming for unique/one-off events. i think incompressibility is bound
1
1
compressibility is mind-contingent though. will be tradeoff between unique/incompressible experiences and that person's prereflective learned ability to compress or find elegance. maybe 5% genetics/*hardware* and 95% self-/nurture-(meta-)bootstrapping *software*
1
1
so like that person's mental kolmogorov complexity will be upper bound
1
1
er, rather, that's best a person can do and will likely be worse? and upper bound will be if that person never figures out better compression in course of meditating, which is unlikely.
1
1
Replying to
That’s roughly the concept behind my freyrag staircase concept
2
1
Replying to
Like, can you add more stories about the stories but in a way that somehow shrinks the staircase even as n increases.
1
Perhaps number of stories increases but each story is more elegantly built out of previous stories, which are losslessly refactored; more reuse without loss of information.
2
1
Replying to
That’s roughly Schmidhubers compression progress hypothesis. I think the staircase is lower bounded by accumulation of incompressibles. Like collecting prime numbers as you count. My own current tack is to interpret the process as time experience.
Replying to
"accumulation of incompressibles" (nice) and not-yet-and-maybe-never-compressibles
Very Schmidhuberian. I want something more neural or winner-take-all mass flow and more phenomenological and more (deep refactorable) ontological and less bitstringy and computationy. But yeah.
1
I suppose I started it with the combinatorics and kolmogorovs
1
Show replies

