Worth a read: A Thousand Brains by Jeff Hawkins

I first taught computer modeling in the long-ago era before the Apple II, and I would start the class by bringing in an object from my then-young son’s toy box. I would hold it up and ask the class, “What is this?”

Model of Ferrari

I could always count on one student, invariably male, to volunteer, “That is a Ferrari!” Which would allow me to say, “Wrong! It is a model of a Ferrari.” Models are simplified representations of a more complex reality, and distinguishing models from reality was a key part of that class. From basic spreadsheets (which were pretty crude then) to hyper-realistic computer-generated imaging found everywhere in today’s cinema, computers have brought us numerous new models of the world around us.

The British statistician George Box (1919–2013) would frequently say something on the order of “all models are wrong, but some are useful.” I cannot sit inside that toy Ferrari and drive it down the street, but I can use it to find a real one in a parking lot, and it can inspire male fantasies on a subject other than sex.

In this blog I have frequently used a conceptual model I call “four ethical brains” to think about four different ways in which our brains are “wired” to handle the questions we term “ethical” or “moral.” When I first introduced that model, I said that I chose four because “two are too few and ten are too many.” Ethical decision-making is not a simple binary like a cartoon angel on one shoulder and a devil on the other. But there are “vectors” of similar writings on ethics through the ages that can be grouped in multiple ways, and there are indeed likely different regions of our own brains “arguing” over difficult questions facing us. I have used this model frequently to suggest one way to “drive” a “moral conversation” about difficult topics.

One thousand brains

Jeff Hawkins is the inventor of the Palm Pilot, the first commercially successful “personal digital assistant” in the 1990s, a device which evolved into the cell phone. Shut out of the phone business, Hawkins turned his attention to neuroscience, founding the Redwood Center for Theoretical Neuroscience and its commercial spin-off Numenta. Hawkins outdoes my brain model by several orders of magnitude in his new book, A Thousand Brains: A New Theory of Intelligence. [1]

The title understates his theory by several more orders of magnitude because Hawkins views the neocortex, that many-folded covering on top of the mammalian brain as consisting of many thousands of “thousand-brain cortical columns” of neurons, running perpendicular to the surface. These structures are largely responsible, he posits, for what we call mammalian intelligence. This type of “thinking” is distinct from that of reptiles and amphibians, which have no comparable “new brain” (the “neo” in “neocortex”). In Hawkins’ model, each one of this massive number of interconnected neuron structures individually “models” a tiny aspect of the larger reality in which we live our lives.

The human brain’s neocortex (from A Thousand Brains, p. 12.)

Hawkins starts with common objects as examples, such as the coffee cup at his fingertips, to start this explanation. One bundle of neurons, which is connected by its long axons and the synapses between nerve cells eventually to a small portion of my eye’s retina, creates a “reference frame” of both what it “sees,” and what it expects to see next. At the same time, a different portion of the brain, connected to my fingers, “feels” small portion of the cup’s handle or surface, itself a new reference frame.

Add many thousands of these, plus more thousands nerve bundles that have nerve impulses always flooding in from the nose or the ears, and together they “vote” on the presence of the cup and where they expect the cup is moving right now. This, Hawkins says, is “learning,” with more nerve bundles and interconnections forming updated and stronger “memories” constantly. Alternatively, when bonds between neuron bundles weaken, we begin to “forget.”

Fitting in with the other theories

At its own level of definition, down at the level of neuron electrochemistry, Hawkins model fits quite well, as the book’s forward by Richard Dawkins notes, with the Daniel Dennett model of brain consciousness as an “expectations generator.” In that model, Dennett sees our brain and its attached senses evolving over billions of years as a larger entity to “guess the future.” The evolutionary “winners” are the creatures that “guess better more often” through evolved senses and brain structures, at least long enough to reproduce in their ecological niche.

Similarly, evolutionary biologist David Sloan Wilson calls this “an elaborate mechanism of environmental assessment,” focusing on the evolution of multiple, and sometimes redundant, sensory organs. [2] In their book The Accidental Homo Sapiens, paleo-anthropologist Ian Tattersall and zoologist Rob DeSalle also key in on the prediction capabilities of these neurons: “Put simply, because we can imagine more choices than our binary brethren can, we can make more choices.” [3]

These are also all, by the way, models of the brain, each simplifying our complex brains’ functioning in their own academic discipline’s language.

Going beyond coffee cups

The research of Hawkins and his teams has focused on using biological and computerized modeling to attempt to replicate this type of “reference framing,” in order to demonstrate the feasibility of their description of brain function. This has been a long and only partially successful process. Our computers have yet to catch up to our amazing mass of brain matter, although interesting, and sometimes disturbing, experiments on brain neurons continue, with scientists recently reporting the growing of rudimentary “eyes” using brain stem cells.

Jeff Hawkins assets that his model is extensible to non-sensory “reference frames” as well, such as language and the human understanding of bigger picture “concepts” like Shakespeare and economics. While the “old brain” beneath the neocortex has physical structures that map to specific bodily functions, Hawkins sees the neocortex as being more of a “general purpose” device, adaptable to knowledge of many kinds. There is plenty of evidence demonstrating how portions of the brain can be “re-mapped” in, say, a blind person from visual processing to perform alternative sensory and memory functions.

This book does not delve too deeply into the “hard problem” of consciousness, but Hawkins does spend a chapter suggesting why his “thousand brains model” is compatible with our current and many conjectured understandings of self-awareness. He does come down, similar to Daniel Dennett, on the side that much of our “consciousness” is a constructed illusion, which likely troubles many readers. Hawkins writes:

“The brain’s model of the world includes a model of our self. This leads to the strange truth that what you and I perceive, moment to moment, is a simulation of the world, not the real world. One consequence of the Thousand Brains Theory is that our beliefs about the world can be false. I explain how this can occur, why false beliefs can be difficult to eliminate, and how false beliefs combined with our more primitive emotions are a threat to our long-term survival.” (pp 5-6)

Machine learning and artificial intelligence

Because Hawkins has turned much of his attention to the practical and commercial implications of brain research, he spends much of the latter part of the book talking about how his model of brain functioning can be translated into a more rich “machine learning” opportunity than our current attempts at artificial intelligence. Just as in the first part of his book, Hawkins avoids the deep chemistry and math of “reality” in favor of easily understandable and practical examples.

Even if you have not modeled every part of that Ferrari correctly, sometimes “good enough” will get you a marketable product. My first job in the automobile industry had me programming very crude computers, by today’s standards, to model the complex interactions of dozens of conveyor systems in car and truck assembly plants. Even then, just one “Aha!” insight from a simulation, for instance realizing that manufacturing defect rates are more like Poisson random variables than classic normal probability distributions, can save a company many thousands of dollars if you model well before installing a factory full of equipment.

Meanwhile, your brain is even capable of imagining yourself in that Ferrari while puttering along in your Kia Soul. Sometimes the model is more satisfying than the reality.


  1. Hawkins, Jeff. A Thousand Brains: A New Theory of Intelligence. Basic Books, 2021.
  2. Wilson, David Sloan. Darwin’s Cathedral: Evolution, Religion, and the Nature of Society. University of Chicago Press, 2010, p.83.
  3. Tattersall, Ian, and Rob DeSalle. The Accidental Homo Sapiens: Genetics, Behavior, and Free Will. Pegasus Books, 2019.

Related posts:

For additional posts on probability, volition and ethics, follow the Dice icon back or forward where it appears.

Prior Dice  Next

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.