In this episode, neuroscientist Jeff Hawkins, founder of Numenta and author of 'A Thousand Brains', teams up with Viviane Clay, director of the Thousand Brains Project and cognitive computing expert. They delve into the Thousand Brains Theory, examining the brain's cortical columns and their significance in sensory integration. The duo contrasts this with deep learning, emphasizing a new AI paradigm rooted in human-like learning. They also discuss the concept of affordances, curiosity in technology, and the transformative potential of their findings on our understanding of intelligence.
01:32:15
forum Ask episode
web_stories AI Snips
view_agenda Chapters
menu_book Books
auto_awesome Transcript
info_circle Episode notes
insights INSIGHT
Mountcastle's Theory
Mountcastle theorized the neocortex uses a universal algorithm, replicated thousands of times.
Different cortical areas perform different functions, but share the same underlying circuitry.
insights INSIGHT
Neocortical Reprogramming
The neocortex exhibits plasticity, repurposing areas for different functions if one is impaired (e.g., vision to touch).
This adaptability supports Mountcastle's theory of a common algorithm.
insights INSIGHT
Neocortical Layers
The neocortex's six layers contain diverse cell types with complex connectivity, not just six simple layers.
Numenta researchers are identifying the functional roles of these cell types in creating intelligence.
Get the Snipd Podcast app to discover more snips from this episode
In this book, Jeff Hawkins presents a revolutionary theory of intelligence that explains how the brain creates intelligence through the use of maplike structures to build tens of thousands of models of the world. The book delves into how these models enable us to perceive the world, understand our sense of self, and originate high-level thought. It also discusses the implications of this theory for machine intelligence and the future of humanity, highlighting potential threats and opportunities[1][4][5].
On Intelligence
How a New Understanding of the Brain Will Lead to the Creation of Truly Intelligent Machines
Jeff Hawkins
Sandra Blakeslee
In 'On Intelligence,' Jeff Hawkins, with the help of Sandra Blakeslee, outlines his memory-prediction framework theory of the brain. This theory posits that the brain is a hierarchical, predictive system that uses memory to make continuous predictions about future events. Hawkins argues that current approaches to artificial intelligence are flawed because they do not understand the fundamental principles of how the brain works. He explains how the neocortex, the seat of intelligence, operates through a hierarchical structure, making predictions based on associative memory. The book discusses the implications of this theory for neuroscience, the development of intelligent machines, and our understanding of human behavior and cognition.
Perceptual neuroscience
Vernon B. Mountcastle
This monumental work by Vernon B. Mountcastle provides a detailed survey of the intrinsic operations of the cerebral cortex, including phylogenetic, comparative, and neuroanatomical studies of the neocortex. It covers topics such as the large-scale organization of the neocortex, neuronal histogenesis, synaptic transmission, and the binding problem in perception. The book is a landmark in the literature of neuroscience, offering insights into the thinking of one of the most distinguished neuroscientists and providing a quantitative approach to understanding the cerebral cortex.
Jim talks with Jeff Hawkins and Viviane Clay about the Thousand Brains Project and Jeff's book A Thousand Brains: A New Theory of Intelligence. They discuss Mountcastle's theory of the neocortex's universal algorithm, cortical columns & their structure, learning modules in AI sensory systems, reprogramming of the neocortex, the 6 layers of cortex, mini-columns & macro-columns, the visual cascade, reference frames as essential for knowledge representation, "voting" for perceptual consensus, how the project differs from deep learning & LLM approaches, James Gibson's concept of affordances, the "Jennifer Aniston neuron" idea, current state of the Monty project, solving fundamental problems vs making impressive demos, avoiding "old brain" traits in AI systems, and much more.
Episode Transcript
Perceptual Neuroscience: The Cerebral Cortex, Vernon B. Mountcastle
On Intelligence, Jeff Hawkins and Sandra Blakeslee (2004)
A Thousand Brains: A New Theory of Intelligence, Jeff Hawkins
Monty Project – Open-Source Implementation
Superintelligence: Paths, Dangers, Strategies, Nick Bostrom
Jeff Hawkins is a scientist whose life-long interest in neuroscience led to the creation of Numenta and its focus on neocortical theory. His research focuses on how the cortex learns predictive models of the world through sensation and movement. In 2002, he founded the Redwood Neuroscience Institute, where he served as Director for three years. The institute is currently located at U.C. Berkeley. Previously, he co-founded two companies, Palm and Handspring, where he designed products such as the PalmPilot and Treo smartphone. Jeff has written two books, On Intelligence (2004 with Sandra Blakeslee) and A Thousand Brains: A New Theory of Intelligence (2021).
Viviane Clay is the director of the Thousand Brains Project. She received her doctorate degree in Cognitive Computing and master’s degree in Cognitive Science at University of Osnabrück in Germany, where she focused on sensorimotor learning as a key aspect in intelligence. She brings to Numenta fifteen years of coding experience, along with her background in neuroscience, psychology, and machine learning.
The Jim Rutt Show 
