nshepperd comments on The Useful Idea of Truth - Less Wrong

77 Post author: Eliezer_Yudkowsky 02 October 2012 06:16PM

You are viewing a comment permalink. View the original post to see all comments and the full post content.

Comments (513)

You are viewing a single comment's thread. Show more comments above.

Comment author: nshepperd 03 October 2012 03:34:07PM 2 points [-]

You're using an AIXI-like notion of senses and actions related by a turing machine or other computation? Isn't such a model annoyingly incomplete in that it can by design provide no explanation for how the input and output streams themselves came about?

Comment author: shminux 03 October 2012 04:32:25PM *  0 points [-]

provide no explanation for how the input and output streams themselves came about?

Right, you postulate input streams and use the feedback from our intentions to act back to the input streams to define outputs. Whether this is more complicated than postulating external reality is a separate discussion.

Comment author: nshepperd 03 October 2012 05:50:43PM 2 points [-]

Um, I don't think I'm concerned with "complexity" as such. It's more like...

Consider the Newtonian universe. "This place" is a euclidean space full of little billiard balls bouncing back and forth. You can point at some subset of the universe and say "that's me". If I ask "why am I seeing [something]?" you can answer that with "because that thing there is a brain, the one computing your experience of consciousness, and it is attached to eyes which happen to be looking at [something]". I guess, it's reductionist.

In the AIXI/instrumentalist model, "sensory input" doesn't seem to reduce to anything else, and the self is "taken for granted". Doesn't that bother you?