The Brain and Connectionism
After reading "Connectome: The Making of You," I wanted to write something. Since I am not a neuroscience professional, all my knowledge about the brain comes from this popular science book, so please forgive any inaccuracies.
Genome and Connectome
Everyone knows about the genome, which is the complete set of DNA sequences you carry. In simple terms, the genome distinguishes you from other animals and makes you human. However, merely distinguishing you from monkeys is not enough for scientists; we also want to differentiate you from your friends, family, and colleagues. What exactly are your mind, your thoughts, and your memories? Although it is just a hypothesis, I believe the answer lies in the connectome.
The connectome is a structural map of your nervous system, but unlike the genome's "sequence of ATCG," the connectome resembles a "graph" that includes not only your neurons but also how they connect with each other. You can think of it as a topological map of a network. In fact, in the field of artificial intelligence, there is hope that biologists will soon decode the human connectome so that this topological map can be referenced. However, the human brain is incredibly complex, with hundreds of billions of neurons. Currently, scientists have only managed to decode the connectome of the elegant roundworm, a single-celled organism with 300 neurons.
The connectome represents the connections between neurons in the brain, and it is constantly changing. It changes with your life experiences and as you grow. The author makes a vivid analogy—Pompeii. If we could travel back to Pompeii, we would see ancient Romans walking the streets, observing their way of life, much like observing a living brain. Although Pompeii was buried by a volcano, the city was preserved intact, allowing us to examine its vacation villas, fountains on the streets, public baths, bars, and the ubiquitous phallic worship. While you cannot see the living people on the streets, you can still study many details of Roman life. Just like studying Pompeii, scientists also study the connectome by examining images of deceased brains.
From the moment your mother conceived you, your genome was already determined. It will largely shape your personality. As we learned in high school biology, genes express themselves by controlling proteins. During the early development of the embryo, certain proteins control the arrangement and direction of the initial few neurons, determining the original connectome. However, your life experiences and learning behaviors can change your connectome, so it can be said that the connectome is shaped by both nature and nurture. Who are you? You are your connectome, which is unique.
Studying the connectome is also crucial for treating mental illnesses. Although humans have not yet decoded their own brain's connectome, modern technological tools allow us to study the connections in small parts of the brain. Even this small part can be very useful. Stroke occurs due to the chaotic entanglement of certain neurons in the cortex. Can we untangle these entanglements? Or encourage the brain to generate new connections to replace these dysfunctional neurons that are on the verge of dying? For Alzheimer's disease (also known as senile dementia), depression, and schizophrenia, can we also prescribe treatments based on neuropathology? For congenital mental illnesses, studying the genome can help us identify which protein is problematic, but true treatment relies on research into the connectome, as the damage has already occurred, and life has already begun.
Causality and Correlation, the Problem of Phrenology
One of the significant impacts this book had on my perspective (you see, my connectome has also been altered by this book) is its discussion of causality and correlation.
In the history of neuroscience, people have long relied on "phrenology." Why was Einstein smart? Let's weigh his brain. Why is a driver good at navigation? Let's see how big his hippocampus is. We can even divide the brain into several "functional areas"—this area controls speech, that area controls language, and another area controls imagination. If your speech area is large, your language ability is strong; if your imagination area is large, your imagination is rich; if the cortex under your left parietal lobe is thick enough, it indicates you can speak two languages.
This is, of course, incorrect. Simply put, the size of brain regions and the thickness of the cortex cannot be used to assess the excellence of a particular brain function. If two variables have a statistical correlation, we say they are "correlated," and we can calculate a number to reflect the degree of correlation between the two variables, which ranges from -1 to 1, known as the Pearson correlation coefficient. The coefficient between IQ and brain volume is r = 0.33, which is a weak correlation. Although on average, taxi drivers have larger hippocampi, musicians have larger cerebellums, and bilingual individuals have thicker left parietal cortices, this is only an average. For individuals, using size to predict brain function remains ineffective.
Even if all musicians have more developed cerebellums, we still cannot judge a person's musical ability based on the size of the cerebellum. We can only say that this person practices music diligently, leading to a more developed cerebellum. In other words, when studying brain function, we ultimately seek causality rather than correlation. Correlation can be useful at times; for example, the correlation between a person's strength and muscle size is r = 0.7, which is a high correlation. Why is this the case? We still need causality: muscle cells do work when they contract, and the more muscle cells there are, the greater the strength. You can think of muscles as a factory; the more workers there are, the higher the output. However, the brain is not structured like that; it resembles a high-tech company, where efficiency depends more on the collaboration and cooperation among employees. Facebook has only 400 employees in Seattle, while Instagram, which it acquired for $1 billion, had only 13 employees at the time of acquisition. The function of the brain does not depend on size but on the connectome.
"To dig for causality rather than correlation" is a good saying for me. In my entrepreneurial product development process, I often need to analyze a lot of data, such as how many people open this webpage daily, how many people click this button daily, where users go after clicking, and what the completion rate of this step is. Currently, there is a prevailing atmosphere in the industry known as "Data-driven." "Data-driven" is good; it allows us to understand how users use your product in unprecedented ways. However, we do not know how high the correlation is between "users clicking a button" and "users liking this feature," let alone causality. For example, based on your user feedback, your product may have a low rate of negative reviews, but this may not be because your product is excellent; it could be that your product is so poor that users are unwilling to complain about it.
The scientific method should be to uncover the causality behind the correlation, that is, the true voice deep within the users. Great product managers can often see through this at a glance, but most of us, like you and me, probably cannot rely on such mystical skills. How can we prove a hypothesis through experimental data? The answer you learned in middle school is to "set up a control group." Unfortunately, we cannot control multiple parallel universes, and there are too many interfering variables when users use a product, making it very difficult to identify real problems during product data collection, as data itself does not express opinions; it is people who express opinions. "Data-driven" is very important, but "The fact behind the data" is even more important.
The real world is extremely complex, with too many variables. The environment constructed in the laboratory is much simpler, which is why social sciences are so difficult, and we have not yet derived many social laws as concise and universal as the laws of physics.
The Hypothesis of Memory
According to the author's explanation, memory is a synaptic chain, and recalling is the neural conduction between synaptic chains. Scientists have discovered an interesting type of neuron, which they call the Jennifer Aniston neuron (group). When subjects view images or video clips of Jennifer, a specific area of the brain is activated, while no other information that does not feature Jennifer can activate it. Scientists believe this neuron represents Jennifer Aniston. We have so many neurons that we can use them to represent any concept or thing in the world. However, this is not entirely the case. Suppose you have a memory in your brain of kissing your first love; such a beautiful memory is something you would likely not forget. Suppose this memory occurred next to a red house, with a blue sky above and green grass below, where your first kiss took place. When you recall it, it is likely that the neuron representing the blue sky first fires, followed by the neuron representing the green grass, triggering a series of memories.
When a subject sees Jennifer Aniston, it is likely that their retinal neurons receive stimulation from photoreceptor cells, then neurons representing blue eyes and blonde hair fire, followed by a series of connected neurons, allowing them to recognize Jennifer Aniston.
How is memory established? When two neurons are repeatedly activated in sequence, a strong connection is formed between them, known as the "Hebbian rule." The prerequisite for this is that there is already a weak connection between the two neurons. However, before they form a strong connection, these two neurons do not know that they will be linked together at some point in the future. The brain operates according to the "Darwinian theory of evolution," dynamically creating some random connections. If the connection between these two neurons is not activated, the weak connection is eliminated; conversely, it is strengthened. This is like a river; where the water flows, it creates a channel, and the channel, in turn, helps the water flow more efficiently.
I once read a book called "How to Learn Effectively." This book highly praises the learning method of "analogy" and, based on the author's personal experience (the author is extremely impressive), makes some hypotheses about the brain. Because analogy allows you to connect two concepts or skills with similar properties (which I believe essentially overlap in some neurons), it is somewhat like establishing strong connections between groups of neurons. This helps you better remember something or understand a concept.
Let's look at two brilliant analogies from Li Xiaolai:
First, "achieving financial freedom" is like "breaking free from gravity to reach the second cosmic velocity." We must increase our acceleration in earning money while eliminating unnecessary financial "gravity," such as mortgages and car loans, and we may even need to learn how to launch a rocket, discarding excess parts at specific times.
Second, "compared to climbing a mountain, entrepreneurship is more like breaking ice." Climbing means you are very clear about where your goal is; you just need to climb step by step according to the plan. However, entrepreneurship is not like that at all. If we must compare, I believe "the college entrance examination" is more like climbing a mountain: knowing how to solve a problem means you can score more points, mastering a knowledge point means you can score more points, and the general types of questions will be announced in advance, with a clear goal of achieving a high score. Breaking ice is different; you might break the ice after just a few hits, or you might hit for a long time without breaking it. Until you actually break the ice, unless you have special tools, you cannot know how thick the ice is.
How the Brain Modifies the Connectome
For the brain, "remembering" is important, but "forgetting" is also crucial. Suppose your brain has two neuron cell groups representing Brad Pitt and Angelina Jolie. After learning that they got married, you need to connect these two neurons. Suppose one day, I say suppose, these two public figures divorced, and you must sever this connection (you might create a new connection related to "divorced").
The brain modifies the connectome through four methods: re-weighting, re-connecting, re-wiring, and re-establishing. How these four "re" methods work is not elaborated here; interested friends can purchase this book.
It is precisely because of these four "re" methods that our brains are plastic, and the connectome can change. The ability to change is age-related; children's brains are more plastic than adults. However, there are exceptions; in the three months immediately following a stroke, the brain will re-strengthen its plasticity, making recovery possible. We need more research to explore the mechanisms of the brain in this regard.
Brain Democracy
"Group decision-making" is often more accurate than "individual decision-making," and "accuracy" is crucial for the brain. The conduction between neurons is not singular but diverse. By controlling receptors for the transmission of electrical and chemical signals, neurons can receive various "opinions" from other neurons. Some neurons only accept positive potentials, while others only accept negative potentials (equivalent to a dissenting vote). Some neurons have very low thresholds and will activate as soon as one neuron transmits a potential (some bills can pass with just one-third approval), while others have very high thresholds and require all connected neurons to transmit potentials to activate (some bills require unanimous approval to pass). Very interesting. The impulsiveness of adolescents is speculated to be related to this.
The Difficulties of Brain Research
For a long time, we were unable to see the connectome of the brain. Optical microscopes could not clearly visualize the connections between neurons. Later, with the advent of electron microscopes, the resolution was sufficient, but the quality of the slices became a headache. Slices are continuous electron microscope images, and stacking many slices together can yield a three-dimensional image. Because the slices are thin, they inevitably suffer damage and contamination during preparation.
After some time, scientists in several laboratories developed better slicing tools, allowing us to obtain high-definition images of brain neurons.
The current problem is that we have obtained high-definition connectome images but cannot analyze them. Just one cubic millimeter of brain tissue slice generates data in the PB range, and the data volume for a mouse brain is a thousand times that, while the human brain's data volume is another thousand times that of the mouse brain. 1 PB = 1024 TB. In 2015, the total active data volume for Taobao in one day was 50 TB, processed by a Chinese internet giant.
It is impossible to process connectome images manually. The current progress involves attempting to use machine learning methods to train computers to assist humans in recognition, but the accuracy is still not high enough. What is very simple for humans, such as distinguishing between two neurons in an image, is incredibly difficult for computers. Even if we can train the reliability of machine learning models to a high level (e.g., 99%), the number of errors it makes in the face of such vast data is still too many.
Unlike the entire field of neuroscience, the goals of connectome research are very clear, and the author is confident about the steps to achieve these goals. It can be said that the key limiting factor for the progress of connectome research is "technological tools." The development of technology often relies on the advancement of tools; without microscopes, we would not have discovered bacteria or seen sperm. Without telescopes, we would not have seen Jupiter's moons or disproved the geocentric theory.
Unfair Advantage
Let me tell a joke: an economist says, "Look, there’s $20 on the road!" Another economist replies, "Don’t be silly; that’s impossible; otherwise, someone would have picked it up already." This joke is a satire of the Efficient Markets Hypothesis (EMH). This is a controversial assertion that states there are no fair and reliable investment methods that can ensure returns above market levels.
Unreliable methods to beat the market do exist; perhaps you randomly bought a stock, and it went up. Unfair methods to beat the market also exist; if you work at a pharmaceutical company, you might know in advance that a certain new drug has passed clinical trials. However, if you buy the company's stock based on this non-public information, you might face insider trading charges. Both methods of beating the market do not meet the EMH's requirements of being "fair" and "reliable." This assertion suggests that such methods do not exist. Investors dislike this assertion; they believe they succeed through intelligence, but EMH claims their success is either due to luck or cheating.
The practical evidence supporting and opposing EMH is complex, but its theoretical explanation is simple: if a stock has new positive news, the first investor to know will raise the stock price. Therefore, EMH believes that good investment opportunities do not exist, just as there won't be
In fact, I have also told a joke: "I have an amazing startup idea".
Returning to the author's line of thought, there are unreliable methods and unfair methods in scientific research. Let's take an example of an unreliable method: Alexander discovered and named penicillin because one of his bacterial cultures was accidentally contaminated with penicillin-producing antibodies. For scientists, rather than relying on unreliable luck, it is better to seek "unfair" advantages; the techniques of observation and measurement are the ways to achieve this. Galileo made the world's first telescope himself; "inventing the telescope" allowed him to occupy an unmatched position in the field of astronomical discoveries because he had the equipment to examine celestial bodies, while others did not. The microscope was not invented by Antonie van Leeuwenhoek, but by a master lens maker, yet these microscopes could only achieve magnifications of 20 to 50 times. Van Leeuwenhoek increased this magnification by ten times, and unlike many ordinary microscopes with multiple lenses, his microscope required only one lens. No one knows how he did it because he kept the manufacturing method a secret. He created a microscope that was more useful than all his competitors, and then, as you know, he became the father of microbiology.
If you are a scientist who needs to buy equipment and are good at securing funding, you might get better equipment than your competitors. But if you can manufacture equipment that money cannot buy, then you have a more decisive advantage.
Suppose you think of a brilliant experiment. Has anyone done it? Check the literature and find out. If no one has done it, you need to think carefully about why not. Maybe it’s because it’s not a good idea at all, but it could also be due to a lack of necessary technical means. If you have the opportunity to obtain the right equipment at this time, you may be able to conduct this experiment before anyone else.
The competition in scientific research is often a competition of technology; people tend to remember those who make significant discoveries, while the experimental instruments behind them receive little attention. In a war, people remember the generals who achieve great feats, but rarely focus on the technological inventions behind them: guns, fighter planes, atomic bombs. The inventors of these killing machines continuously change the face of war, something no general can achieve alone.
In fact, in the previous section, the instruments for preparing slices also underwent a similar "arms race." Everyone hopes to establish their own unfair advantage, which reminds me of the "technological barriers" that many investors in internet startups favor. When I read about "better to use than all competitors" and "getting ahead of everyone else," I felt it was very much like entrepreneurship. Entrepreneurs starting businesses and investors investing are essentially the same; both are investments. Investors put in funds, while entrepreneurs invest in human resources, knowledge, and of course, time. After all, when you immerse yourself in an industry, you temporarily give up opportunities in all other industries.
Who Are "You"
Based on connectomics, the author introduces "human cryopreservation" and "mind uploading" in the last part of the book and seriously discusses the feasibility of these two methods of achieving immortality. This reminds me of a game called "SOMA," where human thoughts can be scanned, saved, and imported into a robot. It also brings to mind "Sword Art Online: Underworld," which features a plot about "mind copying" and descriptions of "artificial consciousness."
Many artistic youths still believe that humans have a soul in addition to their physical body. However, many intellectuals in the sciences do not agree with this view. After reading this book, you should be more convinced that the 21 grams of soul they speak of does not exist. What you think and feel now is a neural impulse; your personality and memories are a connectome.
For centuries, scientists have shaken our beliefs about the soul. Physicists say, "You are a collection of atoms, moving and colliding according to the same physical laws as all other atoms in the universe since the birth of the universe," while biologists say, "You are a machine made up of cells and special molecules like DNA, fundamentally no different from machines made by humans, just more complex."
However, computer science forces us to reconsider this view. We can now accept the concept of "information and its carrier"; if you destroy a computer but the hard drive remains intact and is connected to a new computer, the information is still preserved. We can argue, "I am not my atoms; I am the arrangement pattern of my atoms," or "I am not my neurons; I am the connection pattern of these neurons."
Some might say that when you import old data into a new computer, the soul of the old computer has reincarnated into the new one. They even say, "Information is the new soul." We have come full circle and returned to the starting point; the self is ultimately a non-material existence.