I've had a lot of interesting ideas lately about the nature of intelligence (a very important topic to me). I haven't had the patience to write it in a "Clear" understandable manner, but here are some of the notes I jotted down. I don't know how correct it is, but many theories start off as some sort of intuition. It appears logical to me, and I don't know if human brains work exactly like this, but it appears that this is one possible way to look at it.
The following paragraph basically says that the brain uses a statistical brute force approach to predicting outcomes, though this idea hasn't been completely formed or refined even slightly yet.
----
The time required to predict that A will lead to B is inversely proportional to expected utility. The brain is wired so that memories of predictions have a higher confidence rating when these memories are more rapidly retrieved. New connections are made to each particular memory every time another memory is conceptually mapped to it. These other conceptually linked memories add to the confidence rating of a prediction in memory. The brain does not know or understand these connections. However, the connections are able to be made because a statistical trend can be derived from the fact that more connections will equal more evidence (other conceptually linked memories can be treated as evidence), and more evidence will equal more accurate predictions. Memories that have more conceptual connections will statistically be more likely to be retrieved more rapidly, and memories that are retrieved more rapidly are statistically more likely to be correct. Using these principles, it may be possible to determine that the brain calculates expected utility using "Time taken to process" as one of the input variables
----
Alright, will probably not make sense to nearly everyone who reads this, as the language used is vague and I am not sure how to make it more "Readable" yet without turning it into a novel, but small things like this can be looked at like puzzle pieces, and when enough pieces are found, the big picture becomes clear.
----
Saturday, November 21, 2009
Subscribe to:
Post Comments (Atom)
Ben,
ReplyDeleteI making the following comments on the same order of ambiguity as you made:
Like Roger Penrose (across his popular books) delicately and purposefully using the term ‘struggling’ I would like to tell you that I am also struggling with the whole concept of wiring in the brain. I know it is a highly utilitarian concept, everything seems to be wired and wire-connected, take it from nerve connections to wiring of equipment and the web connection (physically/virtually) but in my view this utility fails up to an effective limit. This effective is the grey margin that can break the whole notion of predictability, the latter always is begging for a statistical ensemble to support its claims. The serious alternative could be how would prediction be possible without running your extensive data mining, all sorts of algorithmic somersaults with pre- and retrodiction, etc. for a bit of clarity I am hinting to something similar to what PW Anderson said under Less Is More.
As much as emergent behavior do require a kind of participating population, (interpose its dynamics too) and almost certainly a kind of statistics is at play, nevertheless, there are quite anomalous cases that its mere observation does defeat the whole statistical treatment. I’m not finished but leave this comment for now. Regards, Alireza
Thank you for your comment :)
ReplyDeleteThis is not Ben, however! I have been straying away from the neurological theories of the mind and focusing more on functional theories of collective minds, which I hope to turn into something practical in the next year or so.