Last time, in The Neocortex Builds a Model of the World, I wrote about how the brain builds a representation, or model, of every thing we know—including objects, words, ideas, etc. I've also written about the columnar structure of the neocortex with its hundreds of thousands of tiny columns that each utilize a similar learning algorithm to build its knowledge.
All of this is based off of Jeff Hawkin's "Thousand Brains" theory of intelligence. At this point, you might be left wondering why this theory refers to a thousand brains and not just one. So here's my final piece on our series on intelligence.
A thousand models
The Thousand Brains theory states that the brain stores information as models. However, there isn't just one model for everything you know. It turns out that almost every cortical column builds its own model of complete objects or ideas. Given that there are a few hundred thousand columns in the neocortex, that means there are thousands of models for every thing you hold in your brain.
That isn't to say that each model is perfectly complete. Each of our cortical columns are connected to different parts of our bodies from our fingertips to our eyes to our feet. That means that each column has an incomplete picture of the nature of an object or idea based on its connections.
In one of his talks, Jeff uses the example of a coffee cup. Say you placed three fingers on a coffee cup and didn't move them. A single point on a coffee cup could be interpreted as a ball, soda can, or mug given their similar features. However, one column feels a rim and can thus rule out the ball as the object. Another column feels the handle and can thus rule out the ball and the soda can. These columns talk to each other, vote on which object it is, and conclude that the object you are touching is a coffee cup.
Jeff claims that this is similar to how the retina in your eye works: each patch of retina is connected to its own column and together they vote to recognize objects in your visual field and construct a coherent image.
This is an incredibly simplistic example, but I hope the point gets across: your brain is a like a democracy: each column votes based in its available information and together your brain perceives and acts accordingly.
Where is knowledge stored in the brain?
After learning that this could be how the neocortex works, I realized that this answers one of my most fascinating questions about the brain: where is knowledge stored?
If the Thousand Brain theory is true, knowledge about the world is stored distributed throughout your entire brain. Jeff uses an excellent analogy of a city's water system to illustrate this concept.
A city's water infrastructure is incredibly complex with many components from pumps to pipes to filters, etc. Where is the city's knowledge about how to maintain the system stored? If you asked individual people within the city, they would be able to give you partial answers based on their experience and expertise. An engineer might talk about the pumps and chemistry behind the water system. A plumber might talk about a home's pipe infrastructure.
No one person can tell you about the city's entire water system maintenance, but collectively the city knows how to maintain its water.
That's what distributed knowledge in the brain is like: knowledge is stored everywhere all throughout the neocortex.
That's it for now on my series on intelligence. I hope you were as interested in this topic as me. As always, I leave out a lot and try to give you just the most important/interesting information, so if you have a follow-up question feel free to email me or comment below!
Thanks for reading.
⚡️P.S. If you're new here and want to read more of the Synapse Newsletter twice a month, subscribe below!⚡️