The analogy called on most often to explain artificial intelligence is the brain. It starts with the name neural network, and terms such as learning, memory and attention are borrowed from the brain as well. Yet how much do we know about the original that was borrowed from? Lisa Feldman's Seven and a Half Lessons About the Brain (The Quest, 2021) is a book that makes you open that original again.
It is a general-interest book on the brain. The chapters are short enough to read one at a time on a commute. Rather than leading with technical terms, it starts from everyday sensations and traces how the brain works.
An artificial neural network is a computational model that mathematically imitates the connections among the brain's nerve cells. Yet because the names are the same, an impression has spread that the two things run on the same principles. This book shakes that impression head-on. The view that the brain, before it is a device that takes in and processes information, is an organ that allocates energy to keep the body alive runs through the whole book.
This view connects directly to today's debates over artificial intelligence. With the electricity used to run large models now on the agenda of climate technology, looking closely at an organ that works within a limited budget is instructive in engineering terms as well. The way self-driving systems sketch out road conditions in advance and correct that picture with sensory input resembles how the brain sets up predictions and reduces the error.
The book's strength is in its compression. Within a short length it clears away the familiar diagram in which reason and emotion are stacked in layers, and builds up an image of an organ that predicts and adjusts. Compared with general-interest brain science books, which tend to be thick doorstops or to slide into the language of self-help, this one leans neither way.
Its limits come from the same place. Being thin means the layers of evidence are not stacked thick. A reader who starts to wonder why each topic is summed up the way it is, and what the counterarguments are, will find it hard to be satisfied with this book alone. The points of contact with artificial intelligence are likewise left for the reader to connect.
The audience to recommend it to is clear. People who read artificial intelligence articles and presentation decks every day and still pause for a moment at the sentence "it resembles the brain," and people who want to know what that analogy costs before using it in product plans or policy documents. It also serves as an entry point for readers opening brain science for the first time. It is better to decide what to read after this book before you start reading it.
The moment a person who blamed three-o'clock irritability on personality starts looking back at the hunger of that hour and the previous night's sleep, what the book set out to say has largely been conveyed. What remains is the sense that while we were absorbed in putting intelligence into machines, the intelligence of our own bodies was left unattended.
The book takes the reader back to the point where they paused for a moment at the sentence "it resembles the brain." And then it brings up sleep and meals.
