New theory of intelligence could upend AI concepts

Oct 29, 2018, 8:15 AM EDT
(Source: Simon Fraser University - University Communications/flickr)
(Source: Simon Fraser University - University Communications/flickr)

Even though neuroscientists have diverging opinions on what constitutes intelligence in humans and how it is formed, most of our machine learning models borrow from the biological mechanisms of the brain. Any attempt to define a novel unifying framework of human intelligence could, thus, upend our concepts of artificial intelligence and probably those of neuroscience as well.

Technologist, scientist, and co-founder of Numenta, Jeff Hawkins, has proposed a new theory of intelligence, entitled “The Thousand Brains Theory,” which interprets the neocortex, the part of the brain associated with higher-order functions such as conscious thought, language, etc., in a radically new manner, notes Psychology Today.

The conventional belief holds that our brain carries out a series of hierarchical processing steps to recognize an object through its sensory features while the new theory posits that the neocortex learns thousands of complete models of each object in the world in parallel, reports Business Wire.