In this episode, In today's episode, we are discussing A Thousand Brians: A New Theory of Intelligence by Jeff Hewkins and Richard Darwins. This book focus on all of neuroscience's advances, we've made little progress on its biggest question: How do simple cells in the brain create intelligence?
Jeff Hawkins and his team discovered that the brain uses maplike structures to build a model of the world—not just one model, but hundreds of thousands of models of everything we know. This discovery allows Hawkins to answer important questions about how we perceive the world, why we have a sense of self, and the origin of high-level thought. A Thousand Brains heralds a revolution in the understanding of intelligence. It is a big-think book, in every sense of the word. He proposes a novel theory of intelligence. Hawkins argues that the neocortex consists of thousands of identical mini-brains (cortical columns) that build models of the world through prediction and error correction. This theory offers a new framework for understanding human cognition and has implications for advancing artificial intelligence (AI). The text explores various perspectives on brain function, contrasting Hawkins's model with older theories, and discusses the potential benefits and risks of AI based on this model, emphasizing the importance of incorporating human values into AI development. The text also considers the limitations of current AI and the potential for AGI to surpass human capabilities and preserve human knowledge.

Key Themes and Ideas:

1. Critique of Traditional Brain Models:

• Hierarchical Model: Hawkins critiques the traditional "step-by-step" hierarchical model of brain processing, which he argues is too simplistic. He contends this model "treats sensing as a static one-way process, ignoring how your movements and changing environment actively shape your perception." It also fails to explain how the brain fills in gaps in sensory information to create a coherent picture.
• Left/Right Brain Myth: The popular notion of left-brain/right-brain dominance is dismissed as overly simplistic and misleading, with the source stating "more recent MRI studies show that there's no such thing as left or right dominance." The source also says, "it gives a false impression that specific regions of the brain control specific bily functions when in fact each function is controlled by clusters of neurons networked throughout the brain."
• Semi-Hierarchical Model: Even more nuanced hierarchical models, which acknowledge that sensory information can sometimes "skip steps" in processing, are considered insufficient for capturing the brain's dynamic and interactive nature.

1. The "Thousand Brains" Theory:

• Cortical Columns as Mini-Brains: Hawkins proposes that the neocortex, the seat of higher cognition, is composed of roughly 150,000 nearly identical processing units called cortical columns. Each column operates as a "mini brain" independently, receiving inputs, building models, and making predictions. "Each column receives inputs, builds mental models and makes predictions just as the brain does as a whole."
• Distributed Cognition: Knowledge is not localized in one specific area but distributed across thousands of these cortical columns. "No single column has a complete understanding of any given object or concept."
• Consensus Through Voting: Cortical columns "vote" on the best interpretation of sensory input and the most appropriate response, producing a unified conscious experience from the distributed activity of each column.
• Redundancy and Robustness: This distributed architecture makes the brain more robust and resilient. Cognitive function doesn't depend on any individual column. "Cognition emerges from the parallel redundant processing of all the mini brains working together."

1. The Interplay of the Old and New Brain:

• Old Brain Functions: The older, more primitive "old brain" is responsible for maintaining basic biological functions and reflexes. All sensory information must pass through the old brain.
• Neocortex Dependence: While the neocortex is the site of higher thought, it's entirely dependent on the old brain for interaction with the outside world. The neocortex does not function "in isolation from your old brain."
• Prediction and Error Correction: The neocortex constantly makes predictions based on past experience and current sensory input. "When predictions and inputs match, your neocortex strengthens its existing neuroc connections...When your neurons predictions are wrong, your brain will form new neural connections." This cycle of prediction and error correction is central to learning.

1. Cortical Column Function in Detail:

• Vertical Arrangement: The cortical columns are vertical arrangements of neurons.
• Uniform Function: Despite varied functions, each operates on the same basic principles. "This uniformity suggests that the brain processes every type of information using a standard set of rules."
• Prediction and Reaction to Change: They predict expected inputs and react when those expectations differ. When "multiple neurons in a cortical column receive the same unexpected input, they fire as one, sending signals to other columns throughout the brain."
• Model Building and Reference Frames: Each cortical column builds models of the world, using reference frames. "To make accurate predictions cortical columns model objects and their positions in three-dimensional space using reference frames."
• Learning Through Prediction: Neurons within the columns learn to predict future inputs. "Each neuron tries to maximize its impact on other neurons while minimizing its own energy consumption." They improve predictions by constantly updating synaptic connections.

1. The Nature of Abstract Thought:

• Extension of Models: Hawkins argues that the capacity for abstract thought is based on the same modeling principles of physical objects and spaces. Abstract concepts are represented by "multi-dimensional reference frames."
• Moving Through Mental Maps: Abstract thinking involves moving through abstract reference frames, just as one physically navigates the real world. "When you recall a memory...you mentally traverse the associated reference frame."
• Embodied Cognition: Hawkins' ideas connect to embodied cognition, a school of thought arguing that "cognition isn't just a property of the brain, but instead emerges from continuous interactions between your brain, your body, and the world."

1. Implications for Artificial Intelligence (AI):

• Mimicking the Neocortex: To achieve Artificial General Intelligence (AGI), Hawkins argues that AI research should shift to mimic the structure and function of the neocortex and use "architectures that create reference frames and mental models in a way similar to how the cortical columns of the neocortex process information."
• Continuous Learning: AI systems should learn continuously by interacting with their environment. "These AI systems would learn by actively exploring and interacting with their environment, building and refining models based on sensory input and motor feedback."
• Reference Frames for AI: The use of reference frames in AI systems could enable the machines to build "models of knowledge and relationships, rather than relying solely on statistical probabilities."
• AI Doesn't Need Human Traits: The old brain, with its primitive drives, could be omitted when creating AGI. "Old brain functions such as survival instincts and emotions could be omitted or replaced with an old brain equivalent."
• Physical/Virtual Presence: AI needs a physical or virtual presence in the world to develop intelligence and motivation. "By equipping AI with a variety of sensors and the ability to direct its attention as we do, researchers can give AI the necessary basis for learning and model building in the real world."

1. Dangers of Intelligence:

• AI Risks Overstated: Hawkins believes the risks of AI are overstated, whereas the risks of human intelligence are more immediate. He argues "AI's risks are often overstated, while human intelligence with its primitive drives and capacity for self-deception poses a more immediate and demonstrable threat to the world."
• AI is Constrained: "No matter how smart an AI becomes, the physical limitations of the real world would still constrain its ability to act."
• Human Threat: Human actions are driven by primitive desires and "the old brain has an immense amount of power over what we perceive and how we act."
• Flawed Mental Models: Humans create flawed mental models of the world, which can spread throughout groups. "The most insidious flawed mental models are those that spread from person to person as self-propagating ideas."

1. Future Possibilities:

• Versatile and Adaptive AI: AGI based on the neocortex will be versatile and capable of tackling complex real-world problems. "This flexibility will allow AGI to tackle complex real world problems such as climate change or economic recessions that require an understanding of multiple domains."
• Accelerated Knowledge: Knowledge acquired by one AI system can be easily transferred to others.
• Preservation of Human Legacy: AGI could carry human knowledge and achievements into the future. "By creating intelligent digital offspring that can carry our legacy into the future, the human race may one day disappear but our achievements and insights won't be lost."
• No Digital Immortality: Hawkins doesn't believe that digital upload of the human brain is possible.

Conclusion:

Hawkins' "A Thousand Brains" presents a compelling and well-supported theory of intelligence based on the distributed function of cortical columns. He provides not only a model of how the brain works, but insights into the nature of the self, and a roadmap for developing more sophisticated AI. He also offers a unique perspective on the risks of both AI and human intelligence. His ideas challenge traditional views and offer a new framework for understanding cognition, AI, and the future of human knowledge.