Hacker News Reader

This reads more like somebody who wants to debate Hawkins than review his book. After reading the review I still don't have a very good idea of what the book is about.

Aside: did anything interesting ever come out of Numenta?

View on HN

>did anything interesting ever come out of Numenta?

Nope.

Except the HTM theory and Hawkins’ talks, which, while perhaps not totally holding up to scientific scrutiny, are inspiring. A bit like prose for the AI/neuroscience-inclined audience.

I'd accept the work as inspirational if the self-proclaimed intent was not to revolutionize cognitive algorithms in the face of those stodgy academics who won't accept the author.

The main idea of the book is the very exciting idea of reference frames which are created by cortical columns and whose function is to model aspects of the world such as physical objects and abstract concepts. Thinking is then moving through reference frames and is directly analogous to moving the body.

Importantly, we use many different reference frames to model, say, a coffee cup (Jeff Hawkins' favourite example!) and they vote between themselves in order to produce a coherent/unitary experience. Hence 'Thousand Brains'.

Reply on news.ycombinator.com