The brain thinks in Lisp parse trees
spectrum.ieee.org
spectrum.ieee.org
I don't like how this guy (Jeff Hawkins, founder of Palm and Handspring) claims to have figured out the fundamental algorithm being used in the cortex without any research or data. His book "On Intelligence" is next in my reading queue and maybe I'll be convinced after I read it... not counting on it though.
What motivated my wild comparison of HTMs and Lisp is:
- HTMs do not distinguish program and data.
- HTMs are fractal like a Lisp parse tree: sameness of structure at any level.
- HTMs program themselves from input, analogous to Lisp programs' ability to change themselves at run time.
Whether the brain actually uses HTMs or not: as far as I know (as a researcher in neuroscience), the HTM structure fits a simplified description of cortex microarchitecture. As Hawkins says, the field of neuromorphic engineering and computation is in its infancy.