The cognitive scientist asking how kids solve problems
Roman Feiman is discovering sophisticated reasoning in young children
Even very young children are capable of sophisticated reasoning and problem-solving, says Roman Feiman. But some types of reasoning are more difficult and may require explicit instruction. Annie Brookman-Byrne finds out more.
Annie Brookman-Byrne: What type of reasoning do you research?
Roman Feiman: Humans have an incredible ability to think and reason about things they’ve never experienced, even things that no one could ever experience. I could tell you that the planet Zoltar is full of friendly intergalactic rodents who’ll bring you cookies when they meet you, unless you wear a watch, which for some reason they’re terrified of. This is all made up, of course, but a minute ago you’d never once thought about Zoltar in your life, and now all of a sudden you know not to wear a watch if you ever go there (unless you hate cookies!).
“Humans have an incredible ability to think and reason about things they’ve never experienced, even things that no one could ever experience.”
The same kind of reasoning underlies all of our thinking, whether it’s about an imaginary scenario or a potential real-life situation – the idea that something will happen, unless something else does. Imagine you’re a kid and there’s a new girl at school who you’ve heard is pretty unfriendly, unless you offer her candy. Or, imagine there’s a stock that’s declining in value and looks like it’s going to zero, but your financial analyst says it will skyrocket if the CEO of the company resigns.
Because you can reason like this, you know what to expect under which circumstances. Your reasoning can then also help you decide how to bring about the circumstance that you like. Whether you’re thinking about Zoltar, or a new classmate, or picking stocks, you use a common system of reasoning to make inferences and decisions. I want to understand how children start to do this sort of thing.
This kind of thinking is applicable to anything. Once you can do it, you can solve all sorts of problems. And once we understand how kids develop the ability to think and problem-solve in this structured way, we can start to learn more about what we can do to help them reason better, and at an earlier age.
ABB: How do you study this in children?
RF: In a lot of my research, I set up simple little games that test whether kids can reason in a certain way. For example, with 18-month-olds, we play a kind of hide-and-seek game. We hide a toy in one of two buckets, out of the kid’s view, so they don’t know where we’ve put it. Then we show them that one of the two buckets is empty. To find the toy on their first try, they have to reason by exclusion: The toy is in one of these buckets. It’s not in this one, so it’s got to be in the other one. Time after time, these kinds of games show that before they even start preschool, before anyone’s taught them how to do it, very little kids can already engage in some pretty amazing and sophisticated reasoning.
But I can only identify those abilities if I set up the test just right. I have to make it fun, and I have to make sure I don’t sound like a robot when I’m explaining the game. It has to make sense that we would play a game like this, and that I would ask such questions in the first place (like “Where’s the toy?”). I have found that if I don’t set up the test correctly, kids won’t do nearly as well, and I’ll underestimate what they’re capable of, not recognizing how much I’ve confused them.
I’ve taken this lesson to heart when teaching students, and when studying how adults think, too. It’s really easy to confuse someone, or lead them to the wrong answer, and that doesn’t mean they’re incapable of finding the right answer. Usually, it just means I didn’t teach the content well or ask my questions clearly enough. If you really want to see what someone is capable of or teach them something new, you need to make everything else as easy and intuitive as possible.
“If you really want to see what someone is capable of or teach them something new, you need to make everything else as easy and intuitive as possible.”
ABB: What ideas are you exploring next?
RF: How good are ordinary people at reasoning, really? Most psychologists think the answer is “not great,” because so many studies show people making all kinds of mistakes on little logic games. My guess is that a lot of those mistakes are due to how the tests, games, and questions are set up – if researchers make sure that nothing gets in the way, they’ll find that people are often capable of reasoning really well in cases where, in the past, poorly designed tests suggested that they weren’t.
At the same time, some mistakes probably run deeper, and reflect certain kinds of reasoning that actually are foreign and difficult for most people, no matter how they are tested. I’m excited about trying to figure out which is which: What sort of reasoning is pretty intuitive for everyone? And what needs to be taught from scratch because it’s not the way people are naturally wired to think? This could help us understand the best ways to teach kids.
Footnotes
Roman Feiman is a cognitive scientist at Brown University in the US. Roman studies how children learn to reason, to express their reasoning in words, and to understand the reasoning of other people. Roman is a 2022-2024 Jacobs Foundation Research Fellow.
Brown Language and Thought Lab
Roman on X
Roman on Bluesky
This interview has been edited for clarity.
Interviewee:
Interview by:
