Your Brain Is a Prediction Machine, Not a Computer ================================================== Sam and Sophie break down Jeff Hawkins's theory that the brain isn't a logic processor—it's constantly predicting what happens next. They talk invariant representations, why AI has failed so far, and what it means to be truly intelligent. ---------------------------------------- SAM: Hey there, welcome back to 7 Minute Books. I'm Sam, and today we're talking about by Jeff Hawkins and Sandra Blakeslee. Sophie, I have to say, this book completely flipped how I think about my own brain. SOPHIE: Oh, absolutely. And hi, everyone. This is one of those books that you finish and suddenly everything feels different. Hawkins is a Silicon Valley engineer and neuroscientist, and he's basically arguing that we've had it all wrong for decades. SAM: Right, the big idea is that the brain is not a computer. Which sounds obvious when you say it, but we've been treating it like one forever. SOPHIE: Exactly. The dominant model in AI and neuroscience has been that the brain processes information, follows rules, and outputs responses, like a biological CPU. But Hawkins says that's why AI has failed at things a toddler can do. SAM: Yeah, we can build a computer that beats a chess grandmaster, but we can't build one that ties a shoelace or recognizes a cat the way a kid does. That really stuck with me. SOPHIE: And his alternative is the Memory-Prediction Framework. The brain's main job isn't to process data, it's to predict what's going to happen next based on past experience. SAM: So it's a prediction machine. That's the phrase he uses, and it's so elegant. He says intelligence itself is defined by this ability to anticipate. SOPHIE: Right. And the physical structure that makes this work is the neocortex. It's remarkably uniform, whether it's handling vision, hearing, or touch, it uses the same basic algorithm. SAM: That blew my mind. The same process for everything? How does that work? SOPHIE: Through invariant representations. Think about recognizing a chair, you can do it from any angle, any lighting, any style. Your brain doesn't store one static image; it stores the 'chairness' across all variations. SAM: Oh, that's beautiful. So the brain is constantly compressing experience into these abstract patterns, and then using them to predict. SOPHIE: Exactly. And it does this through a cortical hierarchy. Information flows up from the senses, becoming more abstract, and predictions flow down from higher areas. SAM: So when a prediction matches reality, the brain dampens the signal. You don't notice the feeling of your clothes on your skin until I mention it, right? Because your brain predicted it. SOPHIE: Yes! And when the prediction fails, like a fly lands on your arm, you suddenly pay attention. That's learning. The brain adjusts its model. SAM: I love the example he gives about picking up a cup of coffee. Your brain doesn't calculate the exact position and then execute a command. It predicts the feel and weight, and then adjusts in real time. SOPHIE: Right. That feedback loop between prediction and sensation is what makes movement smooth. You can catch a ball without thinking about every muscle. SAM: And this has huge implications for AI. Hawkins is really critical of the current approach, massive data, brute force, because it's based on the wrong principles. SOPHIE: He co-founded a company called Numenta to build software based on this framework. It's called Hierarchical Temporal Memory, and it learns patterns in streaming data to make predictions and detect anomalies. SAM: So instead of needing millions of examples, you could learn from a few. That's more like how we actually learn. SOPHIE: Exactly. And he also talks about how memory works. We don't store perfect recordings; we reconstruct memories every time we recall them. That's why they change. SAM: Déjà vu makes so much sense now. Your brain makes such a strong prediction that it feels like a memory. SOPHIE: Yeah. And there's this whole idea that creativity isn't a mysterious spark, it's finding novel connections between existing patterns. Making unexpected predictions. SAM: That's empowering. It means intelligence isn't fixed. You can build a richer model of the world by seeking new experiences and challenging your assumptions. SOPHIE: The book also talks about the older brain, the limbic system. The neocortex learns, but the older brain provides motivation and values. Without it, we'd just be aimless pattern matchers. SAM: So true intelligence needs both. The predictive power plus a guiding sense of what matters. SOPHIE: Right. And that's why we feel pleasure when a prediction succeeds and pain when it fails. That feedback loop is how we learn. SAM: Honestly, the part that got me was when he says we're not as smart as we think, but we have the potential to be much smarter. It's humbling and exciting at the same time. SOPHIE: Yeah. A master chess player has a different model of the board than a beginner. The difference is the richness of the internal model, not raw intelligence. SAM: So the takeaway for me is, stop thinking of your brain as a computer. Start thinking of it as a prediction engine that you can actively refine. SOPHIE: And if you want to go deeper, the whole library is over at 7minutebooks.com/app, with over 6,000 fiction and nonfiction titles you can read or listen to in any language. It starts at $2.99 a month, $9.99 a year, or $19.99 for lifetime access. SAM: That's a great deal. Anyway, this book really makes you see your own mind differently. SOPHIE: It does. asks us to see ourselves as prediction machines, living in a constant state of anticipation. And in that, it finds the very essence of being human. We'll see you in the next one.