Celeste Kidd, UC Berkeley: On attention and curiosity, how we form beliefs, and where certainty comes from

November 22, 2022

RSS · Spotify · Apple Podcasts · Pocket Casts

Celeste Kidd is a professor of psychology at UC Berkeley. Her lab studies the processes involved in knowledge acquisition; essentially, how we form our beliefs over time and what allows us to select a subset of all the information we encounter in the world to form those beliefs. In this episode, we chat about attention and curiosity, beliefs and expectations, where certainty comes from, and much more.

Below are some highlights from our conversation as well as links to the papers, people, and groups referenced in the episode.

Some highlights from our conversation

“We are very different from our primate relatives, our evolutionary relatives, in that humans specialized to a degree that is not seen in any other animals. […] That’s enabled by the way that our curiosity works, the way in which we select information.”

“It’s very possible, I think even likely, that babies are able to form a representation that says, ‘Oh, this is random.’ And if you concluded that this was random, you’re actually done. That’s a higher order representation that’s not captured by the unigram or the transitional model. […] It’s an important thing to understand, I think, for understanding all the rest of human cognition. To what degree can we represent random? We’re not very good at generating it. We’re not very good at recognizing it. I think whether or not, and to what degree, adults have a representation of random is still an open question.”

“When you reach some threshold level of certainty, when you are very certain, our systems are designed to disengage at that point. And once you have become certain about something, it’s very hard to motivate a person to revise that belief.”

“What we were interested in understanding is: what is driving people’s certainty? Is it the vertical probabilities about how certain you should be given the strength of the evidence and some awareness of how big the hypothesis base is? Or is it something else? It turns out it’s something else. It turns out that, like many aspects of higher intelligence, we’re doing it through the use of heuristics, and certainty appears to be driven (for this type of conceptual reasoning) by recent feedback. So the best predictor of whether or not you will say you’re certain is whether or not you’ve gotten the past few right—even if, and this is weird, you start getting them wrong!”

“I have a hot take. I’m asked all the time, and actually sometimes it’s even an assumption of a grant that I’m applying for, that kids are more curious than adults. And I don’t know that we know that. I won’t go so far to say it’s not true. I will say I have a hard time seeing it like that.”

Referenced in this podcast

Thanks to Tessa Hall for editing the podcast.