Why Logic Alone Can't Explain the Human Mind (with Tom Griffiths)

My book, Productiveness, is now available. It's about the difference between doing more and doing what matters, and how we lost the ability to tell them apart. I'll send the release date the day it's set. Visit https://mikevardy.com/productiveness to learn more and get your copy today.

Logic is beautiful and it's also almost never enough. That's the tension running through this entire conversation — the idea that we want our decisions to be purely logical because it's simple and concrete, but almost no real situation in a human life actually has the structure logic requires. I've believed for a long time that reason lives somewhere in the nuanced middle between logic and emotion, and I wanted to sit down with someone who's spent a career studying exactly that boundary.

Tom Griffiths is a Princeton professor and the author of the new book The Laws of Thought: The Quest for a Mathematical Theory of the Mind. If you know his earlier book, Algorithms to Live By, this one is the foundation underneath it — the deeper, denser history of how mathematics has tried, and often failed, to fully capture how minds work. We covered logic's limits, David Hume's riddle about induction, why AI systems fail in strange and "jagged" ways humans don't, and how a 200-year-old idea from George Boole ended up, several generations later, inside modern neural networks through his own great-great-grandson.

Six Discussion Points

• Logic only works when you already have all the information you need and the conclusion is certain — which describes almost none of the decisions we actually make as humans.
• Amos Tversky's early work on similarity (a horse is more "like" a camel than a camel is "like" a horse) laid the groundwork for the heuristics-and-biases research he later did with Daniel Kahneman.
• "Productivity," in the cognitive science sense, means a system's ability to generate something genuinely new — which is closer to creativity than to output.
• David Hume's riddle of induction shows we can't logically justify assuming the future will resemble the past — we simply have a habit of expecting it to.
• AI systems demonstrate "jagged intelligence": strong on one problem, mysteriously weak on the problem right next to it, because they arrive at solutions through a completely different process than human brains do.
• Reading old, "outdated" books remains valuable specifically because they capture the right questions — not just today's most efficient answers.

Three Connection Points

• The Laws of Thought by Tom Griffiths
• Algorithms to Live By by Tom Griffiths and Brian Christian
• Stop Managing Your Time, Start Crafting Your Time Instead

If there's one thing to sit with after this conversation, it's that fuzziness isn't a flaw in how we think — it's often the most accurate description available. Tom's book gives you the historical throughline for why that's true, and honestly, it's a book you want to take your time with, not tear through. Give it the same patience the ideas in it deserve.