How Do We Understand the Brain and Intelligence? — A Recommendation for “The Idea of the Brain: A History”
After working in the field of AI for many years, my interest in the brain has grown increasingly.
I have witnessed countless AI researchers tirelessly pursuing AGI (Artificial General Intelligence), and after several decades, OpenAI has finally approached the threshold of AGI. Meanwhile, there are still many well-respected experts who believe that the path taken by ChatGPT is not viable.
Yet, each of us carries an AGI-level brain within our skulls. If we were to start with the brain to explore the mysteries of AGI, would that be a shortcut? It’s worth noting that many innovations in the AI field have been inspired by neuroscience.
Thus, I began to fantasize that brain science research might be the essential path to exploring intelligence.
To verify my conjecture, I recently read a popular science book titled “The Idea of the Brain: A History”. The result was both astonishing and disappointing to me.
The research in neuroscience has not been as smooth as I had imagined, and I even came to believe that AI research might be a shortcut to uncovering the secrets of intelligence.
A Tortuous History
The book introduces the development of brain science research from the 17th century to the present day in chronological order. From baseless speculation without any physiological foundation to experimental research based on electrochemical transmission mechanisms, it took hundreds of years for people to finally understand the underlying mechanisms of the human brain. During this period, animals used in experiments, humans, and patients who underwent neurosurgical procedures made significant contributions to the discovery of truth. While we are glad for the progress of science, the ancient bloody scenes are shocking.
The process of exploring the brain is like blind men feeling an elephant; people are misled by partial appearances, leading to various conjectures that seem correct but are far from the truth. This is different from the development of physics. Although Einstein overthrew Newton’s classical mechanics, classical mechanics has not become obsolete and can still be used to explain most phenomena in life. Early guesses about the working principles of the brain were often limited by the scientific progress of the time and seem absurd today. By reading this book, you can learn about those interesting yet regrettable historical moments.
The Curse of Complexity
What disappoints me is that brain science research still knows very little about the brain and intelligence to this day. The author candidly admits this.
At a lower level, we have almost figured out how visual perception is generated and how visual neurons extract features layer by layer to form abstract concepts. People who know computer vision will find this part familiar, indicating that AI is on the right track.
But this is only up to the level of perception. Beyond perception, how is memory stored, how are emotions generated, how does thinking proceed, and what is the nature of consciousness? The more we delve into these questions, the more difficult they become to answer, and even research becomes unfeasible.
A lot of brain science research focuses on a specific area, such as a certain location in a brain region, to explore what kind of response these neurons will produce under specific external stimuli, thereby inferring the function of that part of the brain region. Unfortunately, people’s experimental results often contradict each other and cannot be integrated into a complete theory.
There is also a lot of brain science research that uses fMRI and other detectors to monitor large-scale physiological activities in the brain, using machine learning methods to associate them with external manifestations, thereby revealing the functions of different brain regions.
Unfortunately, no matter which type of research, the conclusions drawn do not make much of a splash. At most, they add a little bit of understanding to the existing cognition. But even this existing cognition may be based on incorrect theories.
The author of “The Idea of the Brain: A History” repeatedly conveys to us the crux of the problem — complexity. The brain is a super complex entity that has evolved naturally. Natural evolution means there are no simple and clear rules, and it is likely a patch on top of another patch, just like a large software system that no one knows how it runs. And super complexity means that the research on it exceeds the understanding capacity of the human brain. Just as we cannot understand deep neural networks, even if we draw a complete connection map of the brain, we cannot understand its operating mechanism. For humans, we call this “emergence.” But is there a higher level of intelligence that could understand deep neural networks as easily as we understand 1+1=2? I think this is a question worth considering.
How Do We Understand the Brain and Intelligence
Finally, like the author of this book, I also want to look forward to the future.
Brain science research is still very important, but cracking the mystery of intelligence from here is probably not possible. The reason is the curse of complexity mentioned earlier.
I prefer to approach AGI from the perspective of AI and then exploring the nature of intelligence under known network structures. By designing different network models and then conducting experiments with the methods of neuroscience, we can explore the working principles of cognition, memory, thinking, and consciousness in AI. This method is far more efficient than traditional neuroscience and can avoid ethical restrictions to some extent.
But the worst-case scenario is that, just like we cannot understand deep neural networks, even if we have AGI, we may not be able to understand how it operates. One solution that comes to my mind is to let higher-level AI understand lower-level AI. Then let them try to explain it to us from a perspective that humans can understand. However, when that time comes, making AI listen to humans may become our wishful thinking.
In short, no matter how the future is, optimistic or pessimistic, we must first live a good life now. For anyone who is interested in the brain, I highly recommend reading the book. The book is rich in content, clear in structure, and smooth in language. It is definitely a rare popular science masterpiece.
