Review of “A Thousand Brains” by Jeff Hawkins

With so much talk of AI flying about in the news these days, I thought it would be really helpful to look more deeply into the natural, rather than the artificial, version of intelligence. Neuroscientist and tech mogul Jeff Hawkins is a fascinating guy for this exploration via his book A Thousand Brains: A New Theory of Intelligence. I first heard Hawkins discuss this idea on an episode of Ginger Campbell’s Brain Science Podcast about 5 years ago, but I finally read the book this Spring and really enjoyed it. As is usual for my book reviews, I’ll share quotes from each chapter and give some reflections along the way. These come from the Kindle version published in 2022 with a foreword by Richard Dawkins. Enjoy!
 
Foreword by Richard Dawkins

• (p7) Don’t read this book at bedtime. Not that it’s frightening. It won’t give you nightmares. But it is so exhilarating, so stimulating, it’ll turn your mind into a whirling maelstrom of excitingly provocative ideas—you’ll want to rush out and tell someone rather than go to sleep.

 
What a first four sentences! That was very generous of Dawkins but he actually says little else in his short foreword other than give the briefest of summaries of what is to come. We can skip that though and let Hawkins speak for himself.
 
PART 1: A NEW UNDERSTANDING OF THE BRAIN
Chapter 1 Old Brain—New Brain

• (p16) For the past fifteen years, I have led a research team in Silicon Valley that studies a part of the brain called the neocortex. The neocortex occupies about 70 percent of the volume of a human brain and it is responsible for everything we associate with intelligence, from our senses of vision, touch, and hearing, to language in all its forms, to abstract thinking such as mathematics and philosophy.
• (p22) I created the Redwood Neuroscience Institute (RNI) in 2002.
• (p23) after three years of running an institute, I decided the best way to achieve my goals was to lead my own research team.
• (p23) several colleagues and I started Numenta. Numenta is an independent research company. Our primary goal is to develop a theory of how the neocortex works. Our secondary goal is to apply what we learn about brains to machine learning and machine intelligence.
• (p27) there are several pea-size organs in the amygdala, an older part of the brain, that are responsible for different types of aggression, such as premeditated and impulsive aggression. The neocortex is surprisingly different. Although it occupies almost three-quarters of the brain’s volume and is responsible for a myriad of cognitive functions, it has no visually obvious divisions.
• (p28) Nonetheless, the neocortex is still divided into several dozen areas, or regions, that perform different functions.
• (p32) layers are only a rough guide to where a particular type of neuron might be found. It matters more what a neuron connects to and how it behaves. When you classify neurons by their connectivity, there are dozens of types.
• (p32) Neurons in some layers make long-distance horizontal connections, but most of the connections are vertical. This means that information arriving in a region of the neocortex moves mostly up and down between the layers before being sent elsewhere.

 
So that’s a quick background on who Hawkins is and what he has been working on. He went into some more detail about his time in the tech world before this, but I missed highlighting any of that so I should mention here that he made a lot (a lot!) of money as the co-founder of Palm Computing where he co-created the wildly successful PalmPilot. He actually proposed a PhD where he would develop a theory of the neocortex but that was rejected as untenable so he went to work in tech, made his millions, and then went back to study what he was really interested in. Some may see that as the reckless folly of a dilettante, but it definitely provided a high level of resources and independence for his incredibly ambitious goals.
 
Chapter 2 Vernon Mountcastle’s Big Idea

• (p39) [In 1978, in an essay in The Mindful Brain,] Mountcastle proposed that the reason the regions look similar is that they are all doing the same thing. What makes them different is not their intrinsic function but what they are connected to.
• (p39) I hope you can appreciate how unexpected and revolutionary Mountcastle’s proposal is. Darwin proposed that the diversity of life is due to one basic algorithm. Mountcastle proposed that the diversity of intelligence is also due to one basic algorithm.
• (p40) what was Mountcastle’s proposal for the location of the cortical algorithm? He said that the fundamental unit of the neocortex, the unit of intelligence, was a “cortical column.”
• (p40) there are roughly 150,000 cortical columns stacked side by side in a human neocortex.
• (p41) Mountcastle pointed out that each column is further divided into several hundred “minicolumns.”
• (p41) the major expansion of the modern human neocortex relative to our hominid ancestors occurred rapidly in evolutionary time, just a few million years. This is probably not enough time for multiple new complex capabilities to be discovered by evolution, but it is plenty of time for evolution to make more copies of the same thing.
• (p42) there is [also] the argument of extreme flexibility. Humans can do many things for which there was no evolutionary pressure.
• (p42) The fact that we can do these things tells us that the brain relies on a general-purpose method of learning.

 
I have to admit that this was news to me about the neocortex all looking so similar. Whenever I heard about the visual cortex, auditory cortex, or various regions like Broca’s area, I always thought there must be some slightly obvious differences between the structures. I just never dug deep enough to find this out. But I found Hawkins’ description really persuasive and it also makes sense given the plasticity of brains that are sometimes able to rewire themselves after senses are damaged.
 
Chapter 3 A Model of the World in Your Head

• (p45) What the brain does may seem obvious to you. The brain gets inputs from its sensors, it processes those inputs, and then it acts.
• (p46) From the moment I became interested in how the brain worked, I realized that thinking of the neocortex as an input-leads-to-output system would not be fruitful.
• (p46) My brain, specifically my neocortex, was making multiple simultaneous predictions of what it was about to see, hear, and feel.
• (p49) the neocortex is structured at birth to see, hear, and even learn language. But it is also true that the neocortex doesn’t know what it will see, what it will hear, and what specific languages it might learn.
• (p50) The inputs to the brain are constantly changing. There are two reasons why. First, the world can change. … The second reason is because we move.
• (p51) the brain learns a model of the world by observing how our sensory inputs change as we move.
• (p53) Typically, the number of neurons that are active at the same time is small, maybe 2 percent.
• (p54) Every thought you have is the activity of neurons. Everything you see, hear, or feel is also the activity of neurons.
• (p54) When we learn something, the connections are strengthened, and when we forget something, the connections are weakened. This basic idea was proposed by Donald Hebb in the 1940s and today it is referred to as Hebbian learning.

 
This was more familiar. Neurons that fire together, wire together. And modern neuroscience now sees our brains as prediction machines. (See Clark and Seth, for example.)
 
Chapter 4 The Brain Reveals Its Secrets

• (p56) A high school student can learn the principles of evolution, genetics, quantum mechanics, and relativity. Each of these scientific advances was preceded by confusing observations. But now, they seem straightforward and logical. Similarly, I always believed that the neocortex appeared complicated largely because we didn’t understand it, and that it would appear relatively simple in hindsight.
• (p57) I want to describe several key moments when our understanding took a leap forward, when nature whispered in our ear telling us something we had overlooked. There are three such “aha” moments that I remember vividly.
• (p57) Discovery Number One: The Neocortex Learns a Predictive Model of the World
• (p58) Discovery Number Two: Predictions Occur Inside Neurons
• (p60) Neurons have to figure out how much context is necessary to make the right prediction.
• (p62) Oddly, less than 10 percent of the cell’s synapses are in the proximal area. The other 90 percent are too far away to cause a spike.
• (p62) For many years, no one knew what 90 percent of the synapses in the neocortex did.
• (p63) The big insight I had was that dendrite spikes are predictions.
• (p64) This is a common observation about the neocortex: unexpected inputs cause a lot more activity than expected ones.
• (p64) A prediction occurs when a neuron recognizes a pattern, creates a dendrite spike, and is primed to spike earlier than other neurons. With thousands of distal synapses, each neuron can recognize hundreds of patterns that predict when the neuron should become active. Prediction is built into the fabric of the neocortex, the neuron.
• (p65) Discovery Number Three: The Secret of the Cortical Column Is Reference Frames
• (p67) The movement-related signal we had been searching for, the signal we needed to predict the next input, was “location on the object.”
• (p68) Creating reference frames and tracking locations is not a trivial task. I knew it would take several different types of neurons and multiple layers of cells to make these calculations.
• (p71) The likelihood that a solution is correct increases exponentially with the number of constraints it satisfies.

 
And so, Hawkins argues that his theory of the neocortex as comprised of neurons that predict the world using distal synapses (a synaptic connection located at the far end of a neuron's dendrite, which can influence the neuron's activity by triggering action potentials) and reference frames (more on this in the coming chapters) is very likely to be true because of all the constraints it fits within. In other words, there is so much data that fits with this theory. I’m not expert enough to tell you if there are other data that confound this, but it seems pretty persuasive as presented.
 
Chapter 5 Maps in the Brain

• (p74) [The] simple observation, that we perceive objects as being somewhere—not in our eyes and ears, but at some location out in the world—tells us that the brain must have neurons whose activity represents the location of every object that we perceive.
• (p75) If the quantity of a needed resource, such as food, is increasing, then [bacteria] are more likely to keep moving in the same direction. If the quantity is decreasing, then they are more likely to turn and try a different direction. A bacterium doesn’t know where it is; it doesn’t have any way to represent its location in the world. It just goes forward and uses a simple rule for deciding when to turn.
• (p75) Now consider the advantages afforded to an animal that knows where it is
• (p76) Being able to navigate the world is so valuable that evolution discovered multiple methods for doing it.
• (p76) In mammals, the old brain parts where these map-creating neurons exist are called the hippocampus and the entorhinal cortex.
• (p77) [In the neocortex, there are] place cells: neurons that fire every time the rat is in a particular location
• (p77) [and] grid cells, which fire at multiple locations in an environment.
• (p80) The mapping mechanisms in the neocortex are not an exact copy of ones in the old brain. … It is as if nature stripped down the hippocampus and entorhinal cortex to a minimal form, made tens of thousands of copies, and arranged them side by side in cortical columns. That became the neocortex.
• (p86) Not all of the cortical columns are modeling objects. What the rest of the columns are doing is the topic of the next chapter.

 
Fascinating. Now I understand why the London Cab driver’s hippocampus is the brain region that grows during their training.
 
Chapter 6 Concepts, Language, and High-Level Thinking

• (p87) Vernon Mountcastle proposed that every column in the neocortex performs the same basic function. For this to be true, then, language and other high-level cognitive abilities are, at some fundamental level, the same as seeing, touching, and hearing.
• (p88) Mountcastle didn’t propose what the common function is, and it is hard to imagine what it could be, so it is easy to ignore his proposal or reject it outright.
• (p88) In more abstract terms, we can think of reference frames as a way to organize any kind of knowledge.
• (p90) The hypothesis I explore in this chapter is that the brain arranges all knowledge using reference frames, and that thinking is a form of moving.
• (p91) Reference frames are also the means for achieving goals. Just as a paper map allows you to figure out how to get from where you are to a desired new location, reference frames in the neocortex allow you to figure out the steps you should take to achieve more conceptual goals, such as solving an engineering problem or getting a promotion at work.
• (p92) In the rest of this chapter, I will first describe a well-studied feature of the neocortex, its division into “what” regions and “where” regions. I use this discussion to show how cortical columns can perform markedly different functions by a simple change to their reference frames.
• (p92) Your brain has two vision systems. If you follow the optic nerve as it travels from the eye to the neocortex, you will see that it leads to two parallel vision systems, called the what visual pathway and the where visual pathway.
• (p93) There are what and where regions for seeing, touching, and hearing.
• (p93) Cortical grid cells in what columns attach reference frames to objects. Cortical grid cells in where columns attach reference frames to your body.
• (p95) How can cortical columns create models of things that we can’t sense?
• (p95) The trick is that reference frames don’t have to be anchored to something physical.
• (p95) The second trick is that reference frames for concepts do not have to have the same number or type of dimensions
• (p102) Mathematicians manipulating equations, explorers traveling through a forest, and fingers touching coffee cups all need maplike reference frames to know where they are and what movements they need to perform to get where they want to be.
• (p102) the politician imagines what will happen if they do these things. Their goal is to find a series of actions that will lead them to the desired result: enacting the new law.
• (p108) Reference frames provide the substrate for learning the structure of the world, where things are, and how they move and change. Reference frames can do this not just for the physical objects that we can directly sense, but also for objects we cannot see or feel and even for concepts that have no physical form.

 
Wow. That is a bit mind-blowing but it makes sense from an evolutionary perspective. Each incremental step along the way is easy to see and offers advantages. The body makes do with what it already has, extending this into new functions. And our embodied minds allow us to make models of the world, both physical and abstract. Now we have everything in place for Hawkins’ big idea.
 
Chapter 7 The Thousand Brains Theory of Intelligence

• (p111) Today, the most common way of thinking about the neocortex is like a flowchart. Information from the senses is processed step-by-step as it passes from one region of the neocortex to the next.
• (p111) It is presumed that a similar process—going from simple features to complex features to complete objects—is also occurring with touch and hearing.
• (p111) This view of the neocortex as a hierarchy of feature detectors has been the dominant theory for fifty years.
• (p115) Our proposal of reference frames in cortical columns suggests a different way of thinking about how the neocortex works. It says that all cortical columns, even in low-level sensory regions, are capable of learning and recognizing complete objects.
• (p117) Knowledge of something is distributed in thousands of columns, but these are a small subset of all the columns.
• (p118) Complex systems work best when knowledge and actions are distributed among many, but not too many, elements. Everything in the brain works this way.
• (p118) each column is a complete sensory-motor system, [which distributes the workload] just as each water department worker is able to independently fix some portion of the water infrastructure.
• (p119) Scientists have long assumed that the varied inputs to the neocortex must converge onto a single place in the brain where something like a coffee cup is perceived. This assumption is part of the hierarchy of features theory. However, the connections in the neocortex don’t look like this.
• (p119) we have proposed an answer: columns vote. Your perception is the consensus the columns reach by voting.
• (p122) We propose that [the] cells with long-distance connections are voting.
• (p122) cells that represent what object is being sensed can vote and will project broadly.
• (p123) If you could look down on the neocortex, you would see a stable pattern of activity in one layer of cells. The stability would span large areas, covering thousands of columns. These are the voting neurons. The activity of the cells in other layers would be rapidly changing on a column-by-column basis. What we perceive is based on the stable voting neurons.
• (p125) The voting layer wants to reach a consensus—it does not permit two objects to be active simultaneously—so it picks one possibility over the other. [This is why you] can perceive faces or a vase [in the famous illusion], but not both at the same time.
• (p131) The Thousand Brains Theory is [currently] a framework; it is like finishing the puzzle’s border and knowing what the overall picture looks like. As I write, we have filled in some parts of the interior of the puzzle, whereas many other parts are not done.

 
I have to say, this has revolutionized the way I think about my conscious self and what is humming along in the unconscious background waiting to bubble up when enough “votes” come together to sway my perceptions and predictions. It makes sense from an evolutionary perspective and it makes sense as a robust design full of redundancies. It is not dissimilar to the way bees form a superorganism, which can act as a whole based on information gathered by individuals who dance more or less vigorously the more or less certain they are of the value of what they have found. And it fits with Lynn Margulis’ theory of symbiosis where life is not composed of individuals competing to the death with one another, but is instead composed of groups of groups of groups all the way up and down the chain of life, able to emerge as higher-level entities wherever a group has managed to evolve effective cooperation towards flourishing. Hawkins’ description of the neocortex sounds like that to me—countless individual neurons shaped by millions of years of evolution to come together and enable what we experience as a complex single self.
 
This Thousand Brains theory, with its innumerable copies of the same neuronal mechanisms, also seems to ask tough questions for people proffering some versions of evolutionary psychology and mental modularity. And it may be hard to square with the theory that we have specialized mental immune systems in our brains. But those are questions for another day. For now, let’s quickly look at the rest of this book and how Hawkins himself applies his idea.
 
PART 2: MACHINE INTELLIGENCE

• (P136) From here on in the book, I am going to describe a future that is different than what most people, indeed most experts, are expecting. First, I describe a future of artificial intelligence that runs counter to what most of the leaders of AI are currently thinking, and then, in Part 3, I describe the future of humanity in a way you probably have never considered.

 
Okay, this sounds very interesting.
 
Chapter 8 Why There Is No “I” in AI

• (p139) the biggest reason that today’s AI systems are not considered intelligent is they can only do one thing, whereas humans can do many things. In other words, AI systems are not flexible. Any individual human, such as you or me, can learn to play Go, to farm, to write software, to fly a plane, and to play music.
• (p141) I know it sounds like it should be easy, but no one could figure out how a computer could know something as simple as what a ball is. This problem is called knowledge representation. Some AI scientists concluded that knowledge representation was not only a big problem for AI, it was the only problem.
• (p143) Deep learning networks work well, but not because they solved the knowledge representation problem. They work well because they avoided it completely, relying on statistics and lots of data instead.
• (p143) they don’t possess knowledge and, therefore, are not on the path to having the ability of a five-year-old child.
• (p145) Truly intelligent machines, AGI, will learn models of the world using maplike reference frames just like the neocortex. I see this as inevitable. I don’t believe there is another way to create truly intelligent machines.
• (p146) Today we are building dedicated AI systems that are the best at whatever task they are designed to do. But in the future, most intelligent machines will be universal: more like humans, capable of learning practically anything.
• (p148) Due to dramatic reductions in cost and size, general-purpose computers became one of the largest and most economically important technologies of the last century. I believe that general-purpose AI will similarly dominate machine intelligence in the latter part of the twenty-first century.
• (p149) My proposal for what qualifies as intelligent is based on the brain. Each of the four attributes in the following list is something we know that the brain does and that I believe an intelligent machine must do too.
• (p150) 1. Learning Continuously
• (p150) 2. Learning via Movement
• (p151) 3. Many Models
• (p152) 4. Using Reference Frames to Store Knowledge
• (p153) Once AI researchers understand the essential role of movement and reference frames for creating AGI, the separation between artificial intelligence and robotics will disappear completely.

 
This makes so much sense to me and is largely absent from the AI discussions in the news today. As far as I can tell, the world is being fooled by overconfident tech bros with their smooth-talking machines whose actual capabilities are being hidden inside a black box. And so much money and energy are being sunk into this that it is distorting the environment and the economy while accelerating the enshittification of everything on the Internet. It’s so frustrating.
 
Chapter 9 When Machines Are Conscious

• (p156) Since there isn’t even an agreement on what the word “consciousness” means, it is best to not worry about it.
• (p158) our sense of awareness, what many people would call being conscious, requires that we form moment-to-moment memories of our actions. Consciousness also requires that we form moment-to-moment memories of our thoughts.
• (p162) I don’t have an explanation for why pain hurts, or why it feels the way it does and not like something else. This doesn’t bother me in any deep way.
• (p164) Fears and emotions are created by neurons in the old brain when they release hormones and other chemicals into the body. The neocortex may help the old brain decide when to release these chemicals, but without the old brain we would not sense fear or sadness. Fear of death and sorrow for loss are not required ingredients for a machine to be conscious or intelligent.
• (p166) At some point in the future, we will accept that any system that learns a model of the world, continuously remembers the states of that model, and recalls the remembered states will be conscious. There will be remaining unanswered questions, but consciousness will no longer be talked about as “the hard problem.” It won’t even be considered a problem.

 
Ugh. Hawkins starts by admitting we don’t have an agreed upon definition of consciousness but then he goes on to craft a very narrow definition of it and blithely dismiss the hard problem. He’s out of his depth here and there’s no need to harp on it. But I would say that “fear of death and sorrow for loss” may not be required for Hawkins’ very limited definitions of consciousness and intelligence, but I believe they are absolutely required for wisdom and the kind of sentience that matters.
 
Chapter 10 The Future of Machine Intelligence

• (p168) older parts of the human brain control the basic functions of life. They create our emotions, our desires to survive and procreate, and our innate behaviors. When creating intelligent machines, there is no reason we should replicate all the functions of the human brain. The new brain, the neocortex, is the organ of intelligence, so intelligent machines need something equivalent to it. When it comes to the rest of the brain, we can choose which parts we want and which parts we don’t.
• (p168) Intelligence is the ability of a system to learn a model of the world. However, the resulting model by itself is valueless, emotionless, and has no goals. Goals and values are provided by whatever system is using the model.
• (p169) The recipe for designing an intelligent machine can be broken into three parts: embodiment, parts of the old brain, and the neocortex.
• (p174) The neocortex must be attached to something that already has sensors and already has behaviors. It does not create completely new behaviors; it learns how to string together existing ones in new and useful ways.

 
This seems accurate but I would be more careful with the words being used. Hawkins seems to think we can disaggregate “intelligent behavior” into “behavior” (driven by emotions) and “intelligence” (guided by the neocortex) and somehow arrive at a component of intelligence all on its own without any emotions. But without goals and values, how exactly is something going to be intelligent? The simple dictionary definition of intelligence is “the ability to acquire and apply knowledge and skills”. Lacking all motivation, an agent will not “acquire” or “apply” anything. In that case, the agent is merely a dumb tool that something else with intelligence can use. Any actions that AI agents take were given to it by some human, wittingly or unwittingly. In his solo podcast lecture titled “What is Intelligence?”, the intelligence researcher Scott Barry Kaufman made a specific point of introducing the roll of motivation to his new theory of personal intelligence (start watching at 41:29 for more). He did this for humans because different levels of motivation cause real differences in the perceived intelligence of one person compared to another. But if we imagine removing all motivation—something that is impossible for a living person, but is absolutely the case for AI systems—then all of the real intelligence goes away as well. The implications of this are actually close to what Hawkins says next.
 
Chapter 11 The Existential Risks of Machine Intelligence

• (p186) It is one thing for bad people to use intelligent machines to do bad things; it is something else if the intelligent machines are themselves bad actors and decide on their own to wipe out humanity. I am going to focus only on the latter possibility, the existential threats of AI.
• (p187) arguments are being made without any understanding of what intelligence is. They feel wildly speculative, based on incorrect notions not just of what is technically possible, but what it means to be intelligent.
• (p191) The goal-misalignment threat depends on two improbabilities: first, although the intelligent machine accepts our first request, it ignores subsequent requests, and, second, the intelligent machine is capable of commandeering sufficient resources to prevent all human efforts to stop it.
• (p192) Intelligent machines will not develop misaligned goals unless we go to great lengths to endow them with that ability.
• (p192) We don’t let a single human, or even a small number of humans, control the world’s resources. We need to be similarly careful with machines.

 
Agreed. And note that the “we” in these prescriptions are human beings with actual motivations, intelligence, and moral responsibility. Don’t forget that whenever Silicon Valley executives try to fob off their liability onto some kind of “conscious AI”.
 
PART 3: HUMAN INTELLIGENCE

• (p197) How we act in the coming years will determine whether our sudden rise leads to a sudden collapse—or, alternately, if we exit this period of rapid change on a sustainable trajectory. These are the themes I discuss in the remaining chapters of the book.

 
This last section of the book is the most speculative from Hawkins so I’ll go through it quickly.
 
Chapter 12 False Beliefs

• (p200) everything we do perceive must be fabricated in the brain.
• (p201) it doesn’t feel as if we are living in a simulation. It feels as if we are looking directly at the world, touching it, smelling it, and feeling it.
• (p207) Believing the Earth is flat does not cause you to act in a way that spreads your belief to other people. Viral models of the world prescribe behaviors that spread the model from brain to brain in increasing numbers.
• (p210) There is only one way, that we know of, to discern falsehoods from truths, one way to see if our model of the world has errors. That method is to actively seek evidence that contradicts our beliefs.
• (p210) Actively looking for evidence to disprove our beliefs is the scientific method.

 
Yeah, okay, but there’s a lot of philosophy out there dedicated to this. It’s great that Hawkins recognizes this is an issue that we need to address, but this chapter was very shallow.
 
Chapter 13 The Existential Risks of Human Intelligence

• (p216) Life is based on a very simple idea: genes make as many copies of themselves as possible.
• (p216) However, what is good for genes is not always good for individuals.
• (p217) So why aren’t we collectively lowering the population? Because the old brain is still in charge.
• (p218) The simple and clever solution is to make sure every woman has the ability to control her own fertility and is empowered to exercise that option if she wants to.
• (p222) two common false beliefs that are existential threats are denying the danger of climate change and belief in an afterlife.
• (p225) Each of us has an old brain that causes us to behave in ways that are detrimental to our species’ long-term survival. Are we doomed? Is there any way out of this dilemma? In the remaining chapters, I describe our options.

 
You won’t get any arguments from me about educating women and the dangers of climate change and imaginary afterlives. Like a good engineer, Hawkins lays out clear problems that he wants to solve.
 
Chapter 14 Merging Brains and Machines

• (p226) There are two widely discussed proposals for how humans could combine brains and computers to prevent our death and extinction. One is uploading our brains into computers, and the other is merging our brains with computers.
• (p232) “Uploading your brain” is a misleading phrase. What you have really done is split yourself into two people.
• (p234) Similar to the “upload your brain” scenario, there are extreme technical challenges that have to be overcome to merge with a computer.

 
So, neither of these looks very promising any time soon. If that’s the case…
 
Chapter 15 Estate Planning for Humanity

• (p237) Up until now, I have been discussing intelligence in both biological and machine form. From here on I want to shift the focus to knowledge. Knowledge is just the name for what we have learned about the world.
• (p237) In this and in the final chapter, I explore the idea that knowledge is worthy of preservation and propagation, even if that means doing so independently of humans.
• (p239) Estate planning is something you do during your life that benefits the future, not yourself.
• (p241) I am going to discuss these ideas further by describing three scenarios we might use to communicate with the future.
• (p241) Message in a Bottle
• (p242) Leave the Lights On
• (p249) Wiki Earth

 
Sure, this seems like a good idea as humanity gets closer to extinction. But we don’t feel that pressure yet. So, for the same reason that 25-year-olds rarely have wills, this is unlikely to happen.
 
Chapter 16 Genes Versus Knowledge

• (p255) In this final chapter, I discuss three methods we might pursue to prevent our demise. The first method may or may not work without modifying our genes, the second is based on gene modification, and the third abandons biology altogether.
• (p255) Become a Multi-planet Species
• (p257) For humans to live on Mars on a permanent basis requires intelligent machines to help us.
• (p261) Environmentalism is not about preserving nature, but about the choices we make. As a rule, environmentalists make choices that benefit future humans.
• (p261) do we want our future to be driven by the processes that got us here, namely, natural selection, competition, and the drive of selfish genes? Or do we want our future to be driven by intelligence and the desire to understand the world?
• (p262) Modify Our Genes
• (p263) imagine we learn how to modify our genes to eliminate aggressive behavior and make a person more altruistic. Should we allow this?
• (p265) Leaving Darwin’s Orbit
• (p265) The ultimate way to free our intelligence from the grip of our old brain and our biology is to create machines that are intelligent like us, but not dependent upon us.
• (p270) I want to make the case for knowledge over genes. There is a fundamental difference between the two, a difference that makes preserving and spreading knowledge, in my opinion, a more worthy goal than preserving and spreading our genes.
• (p270) Genes are just molecules that replicate.
• (p270) Knowledge is different. Knowledge has both a direction and an end goal.
• (p270) There is a direction to knowledge. Knowledge of gravity can go from no knowledge, to Newton’s, to Einstein’s, but it can’t go in the opposite direction.
• (p271) In addition to a direction, knowledge has an end goal.
• (p271) There are plenty of other things we don’t understand: What is time? How did life originate? How common is intelligent life? Answering these questions is a goal, and history suggests we can achieve it.

 
Oofa. First of all, environmentalists don’t just care about future humans. Many of us care about life in general, where humans are deeply enmeshed and interconnected. (See my magazine article “When the Human in Humanism Isn’t Enough”.) Then, it seems like Hawkins stopped reading evolutionary biology after discovering Dawkins’ selfish genes. But there are many stories out there focusing on the role that cooperation has played in the evolution of life through its major transitions (c.f.: Prosocial, Survival of the Nicest, or Mutual Aid). So, we probably don’t need to take such drastic measures as genetic modification (whose outcomes are unknowable) in order to improve society. And finally, as for knowledge having a direction and a goal, this is just at odds with the fundamentals of epistemology. Truth may be a goal, but it seems unlikely we’ll ever know we have it. That is why knowledge can go (and has gone) in lots of different directions as new observations roll in. As the future unfolds, I think we ought to keep valuing the genetic codes of life that have got us this far. They took billions of years to develop and there is a lot of intelligence built in them that we are very unlikely to have fully grasped (seeing as how we have no other examples yet of how life proceeds).
 
That’s all for now. As a quick summary, I thought the first third of this book was filled with great science. It is a potential tour de force for understanding our minds and grappling with the confusing situations that AI research and hype are currently bringing upon us. After that, the rest of the book could use some better philosophy. I actually wrote something to help with that which was just published so I’ll share that next week.