A Thousand Brains

January 1, 2021
How a collection of identical cells takes in information and generates intelligence. Entrepreneur and computer engineer Hawkins' enthusiasm comes through in TED talks and an earlier bestseller (On Intelligence, 2004), and neuroscientists take him seriously--though readers new to the subject may have a more difficult time digesting the complex information. The author focuses on the neocortex. Only mammals have one, but all animals possess a deeper "reptilian" brain designed to ensure survival and reproduction. The neocortex allows humans to "devote our lives to philosophy, mathematics, poetry, astrophysics, music, geology, or the warmth of human love, in defiance of the old brain's genetic urging" that we should be spending time "fighting rivals and pursuing multiple sexual partners." Hawkins adds that all thoughts and actions result from activity and the connections among neurons. Every element of intelligence--seeing, touching, language, thought--is fundamentally the same. The author's intriguing thousand brains theory maintains that identical structures called "reference frames" occur throughout the neocortex. All take in sensory information "to model everything we know, not just physical objects," and "all knowledge is stored at locations relative to reference frames." Nothing enters our skull but electrical spikes, so this model is a simulation. It's usually accurate, but humans perceive lots of nonsense and false beliefs, which have become threats to our long-term survival. Modern life remains a battle between the neocortex (knowledge) and the old brain (competition, survival). Usefully, Hawkins then applies his theory to machine intelligence. Computers store knowledge but lack reference frames and the ability to model: "Nothing we call AI today is intelligent." Humans are intelligent because we can learn to do practically anything. Computers do one thing, although they do it far better than humans. There is no "deep learning," only access to immense amounts of data, and future intelligent machines will be more like humans, and "success...could be a machine that has the abilities of a five-year-old child." Richard Dawkins provides the foreword. Insightful stuff for readers immersed in the labyrinthine world of neuroscience.

COPYRIGHT(2021) Kirkus Reviews, ALL RIGHTS RESERVED.

January 11, 2021
Hawkins (On Intelligence), inventor of the PalmPilot, explicates his theories of how the brain works in this revelatory survey of human intelligence. He begins with the cell, explaining that individual cells work together in the brain to create intelligence (which, Hawkins writes, spans from “basic sensory functions to the highest forms of intellectual ability”) before moving on to a consideration of how the brain’s neocortex works. Hawkins’s central idea, the thousand brains theory, is that “the entire world is learned: as a complex hierarchy of objects located relative to other objects.” Extending beyond human intellect, Hawkins discusses artificial intelligence, which he writes falls short of human intelligence because of the narrow limits constraining the operations of even the smartest machines. In his most daring section, he argues that serious consideration should be given to using intelligent machines to preserve human knowledge, which would enable information to persist and be distributed throughout the galaxy, long after the death of the last human. The complex concepts are presented as simply as possible, but they aren’t dumbed down and demand focus. Readers who persist will find Hawkins’s study is full of thought-provoking arguments.