The ROI of Thought
The skill of "problem solving" is a myth. No one has problem solving skills. We have cognitive budgets and valuation functions. The goal of education and leadership isn't to build one magic skill but to lower the cost of effortful thought while raising the valuation of the problem. Here are 3 new papers that help illuminate these ideas.
Research Roundup
Wisdom as Waveform
We love to label people. "She is wise." "He is foolish." We treat wisdom like a permanent stat on a D&D character sheet—something you simply have. But wisdom (and for that matter problem-solving) may be less stable traits and more fleeting, high-cost states of mind.
When wisdom is broken into "4 core metacognitive" skills (intellectual humility, recognition of uncertainty, perspective taking, and compromise), people’s ability to use them fluctuate wildly based on context.
For example, “individuals who reported [high] distress at one time point also reported elevated levels of wisdom-related features 3 months later”. They freaked out one day, and that distress forced the "wiser" coping mechanism to kick in later.
Wisdom isn't a holistic gift or genetic endowment. It's a specific, expensive reaction mechanism. Much like courage, it doesn't exist in a vacuum; it only appears when the heat is turned up
Problems are costly; Thinking is free.
Here is a fascinating claim about LLMs: “out-of-the-box reasoning models reflect core features underlying problem and task complexity in human cognition, without requiring any built-in symbolic mechanisms.”
Specifically, “across seven reasoning tasks…the length of the chain-of-thought generated by large reasoning models predicts human reaction times both within tasks—tracking item-level difficulty—and across tasks—capturing broader differences in cognitive demands.”
In other words, AI's internal reasoning process seems to be a good barometer for human mental effort.
There are 3 claims here: all fascinating—one potentially misleading.
- Chain-of-reasoning correlates with human reaction times. (Empirical and cool).
- LLMs produce this complexity without symbolic mechanisms. (A classic debate: do we need explicit symbols, or is "thinking" just tracing trajectories in high-dimensional manifolds? The jury is out, but the data is compelling).
- LLM “reasoning” and human cognition share some underlying complexity.
#1 is the empirical finding of the paper, and so very cool and hard to dispute.
I hadn’t fully engaged with idea #2, but it is a classic debate in cognitive science: does complex cognition require explicit symbolic processing—is it like our conscious thoughts? LLMs are just tracing trajectories in high-dimensional manifolds, and so perhaps not.
However, the last claim (#3) is misleading. LLMs reason and human problem solving share an obvious not-so-hidden cause: the problem complexity itself. Humans and machines might be solving a maze very differently, but it still requires the same series of turns in the end.
Machine problem solving, much like human problem solving, may not reflect one single skill.
"Learning Expectations" is Subjective Utility!
I often complain about "cognitive laziness"—the human tendency to default to the path of least resistance. Let’s be clear, though, cognitive laziness isn't a moral failing; it's a neuroeconomic calculation based on our lived experiences..
We treat "thinking hard" (allocating cognitive control) like an investment, expending more effort if we perceive the task isn’t just doable in the moment but learnable with effort. In a 2-stage experiment, participants' “learning rate in the first block” predicted their willingness to deliberate and effortfully prepare for the second.
When they experienced the benefits of learning (ROI), they invested more cognitive effort. For those that didn’t see the gain, the "subjective utility" of effortful learning didn’t outweigh the metabolic cost of thought.
If students or even colleagues aren't self-motivated "problem-solvers" in the moment, it’s not because they lack a mythical problem solving skill. It's at least in part because their neuroeconomist has run a biased set of numbers and decided your problem isn't worth their metabolic investment.
You can't train them to be "smarter", but if you change their experiences you can train their brains into believing the struggle is actually worth the cost.
Media Mentions
This photo of silent auction items for the Kennedy Human Rights Center isn't terrible good, but if you scan around a bit you will see that one of these items is not like the others!
Magic Johnson and Stephen Colbert, you are making it hard for rank amateurs to fleece New York financiers out of their hard earned filthy lucre...for human rights!
SciFi, Fantasy, & Me
For years, my son has been trying to get me into the world of SCP: a massive, collaborative sci-fi project about a secret organization cataloging the impossible. Imagine my surprise when I saw one of his favorites, "There Is No Antimemetics Division", showing up on "Best SciFi of the Year".
As a genre-snob and a neuroscientist, I was hooked immediately. This is memory horror—a story centered on threats so terrifyingly insidious, they don't even leave a memory trace of their victims. Forget things that go bump in the night; this book is about entities that literally erase themselves and their victims from our awareness.
If you want to wrap your mind around a concept that challenges the very nature of perception and memory—moving past body horror into true conceptual terror—this audiobook is a chilling and brilliantly written ride. Highly recommended!
Stage & Screen
- January 20, Davos: After all these years, they are finally allowing me to speak in Davos at the World Economic Forum.
- February 2, NYC: My latest research on neurotechnologies for cognitive health and more.
- February 10, Nashville: Shockingly, I haven't visited Nashville since I was a little kid. On this trip I'll be looking at why Tennessee and North Carolina appear to have more entrepreneurship than all over their neighboring states combined.
- March 8, LA: I'll be at UCLA talking about AI and teen mental health at the Semel Institute for Neuroscience and Human Behavior.
- March 14, Online: The book launch! Robot-Proof: When Machines Have All The Answers, Build Better People is will finally be inflicted on the world.
- Boston, NYC, DC, & Everywhere Along the Acela line: We're putting together a book tour for you! Stay tuned...
- Late March/Early April, UK & EU: Book Tour!
- March 30, Amsterdam: What else: AI and human I--together is better!
- plus London, Zurich, Basel, Copenhagen, and many other cities in development.
- April, Napa: The Neuroscience of Storytelling
- June, Stockholm: The Smartest Thing on the Planet: Hybrid Collective Intelligence
- October, Toronto: The Future of Work...in the Future