A Thousand Brains, a Thousand Points
A very interesting read which provided a fresh perspective on how the brain works. It is so miraclous to learn that each cortical column serves is able to process so much and create a model of the world on its own. Jeff Hawkins proposed that Mountcastle’s universal algorithm and Francis Crick’s framework revolves around reference frame where the brain can learn structures, manipulate objects (mental and physical) and create movements (relative movements). The key to learning is through movements, and each cortical column has a set of cells equivalent to grid cells and another equivalent to place cell, and another equivalent to head direction cells to construct a “map” and navigate through information. Each column learns a predictive model and all of them within the brain will come to a common consenus through voting process. This is highly interesting as I hope to incorporate this into hardware in the future.
Part 2 and 3 of the book covers machine intelligence and human intelligence, where Jeff Hawkins covered about consciousness of machines, future of machine intelligence (ML), risks of machine intelligence, risks of human intelligence and estate planning. However, I cannot help but feel that these two parts are more strong in terms of language. Jeff Hawkins himself stated that these two parts are by-no-means what it is, and are just providing points for discussion.