Beliefs About Beliefs
For this week's Research Roundup we're exploring how beliefs alter cognition.
Gamified Humility
Here’s a revealing “paradox”: parents who rate themselves as highly humble are more likely to have overconfident, less humble children.
In the study, parents “completed two primary measures of intellectual humility: a self-report measure…and a behavioral measure”. It turns out what they said and what they did “were not correlated” at all.
With so much (fair) criticism of LLM hallucinations and overconfidence, it’s important that we are honest about our own human shortcomings, and celebrate our relative superpower in meta-uncertainty: assessing our internal uncertainty.
The parents who admitted the limits of their knowledge—"I don't know, let's figure that out"—had children who were more accurate in assessing their own knowledge. That’s wonderful and actionable!
Humility is not simply a trait you possess, it's a behavior you can practice…and role model. Our children, our teams, and our colleagues learn from our actions, not our self-professed virtues.
In an age where AI can provide answers instantly, knowing everything (much less pretending to know everything) isn't an advantage; being “Robot-Proof” is possessing the curiosity and intellectual humility to ask better questions and readily admit what we don't know.
The parents in this study who demonstrated humility weren't just teaching a virtue; they were equipping their children with the essential mindset to thrive in a world of intelligent machines.
We The People
One of the most powerful tools for leadership isn't a grand strategy; it's a single, two-letter word: "we".
When we describe a mistake with "Sometimes we drop things" instead of "Sometimes she drops things" even young children perceive us as kinder, more compassionate, and part of an inclusive group.
Using "we" (even more than a "generic you") seems to signal that we see a problem not as an individual failing, but as a shared human experience. Subtle cues like this might have large surprising impacts on psychological safety and collective problem-solving.
New research we are conducting at The Human Trust on building a "civic immune system" includes exploring literal "us vs. them" language shifts and it’s ties to zero- or negative sum beliefs.
Mind Meld
What makes a team smart? It’s not a room full of geniuses.
In contrast to individuals, a group's success in collective problem solving is “not strongly correlated with the average or maximum individual intelligence of [individual] group members”. The highest raw brainpower didn’t drive collective intelligence.
The smartest teams were the ones who were good at collaborating. This correlated with 3 factors:
- “average social sensitivity” of members
- “conversational turn-taking”, and
- “proportion of females in the group”.
Teams where a few people dominated the conversation were consistently dumber than teams where everyone contributed. (Key causal note: making everyone talk doesn’t make teams smart either.)
It’s our beliefs about our collaborators and our own intellectual humility that drives collective intelligence. In my work on “Augmented Collective Intelligence”, the focus is on building systems that don't just provide answers but that actively foster the collaborative dynamics that make human teams (and cyborg collectives) more than the sum of its parts.