What's normal? Developing beliefs about natural kinds


From childhood, people tend to view some category members as more representative and informative about their kinds than others. For instance, people tend to think that robins are more representative of birds than, say, penguins. These beliefs about what’s normal shape how people use categories to learn about the world. Although most previous research on concepts has focused on how category knowledge supports inference making, like inferring what a particular girl would like for her birthday (Foster-Hanson & Rhodes, 2021), I found that people also think of many categories prescriptively (e.g., they think that girls are supposed to like certain things, and not others), and these prescriptive beliefs are particularly influential in shaping young children’s concepts. For example, if a person wanted to learn about girls, would it be better to look at a girl who wears an average amount of pink for girls, or one who wears the most pink possible? In recent work, I found that whereas adults often think about gender categories in terms of category averages (e.g., they think of the girl wearing an average amount of pink as most representative and informative about girls in general), young children tend to focus instead on category stereotypes (e.g., they view the girl wearing the absolute most pink as representative, since girls should wear pink; Foster-Hanson & Rhodes, in press).


The tendency to conflate what’s normal with what’s good is a fundamental feature of early social reasoning. For example, young children expect that members of social role categories will do what’s good for the category more than adults do (e.g. reasoning that scientists are curious because they should be; Foster-Hanson & Rhodes, 2019b). But in my work, I show that this tendency actually extends beyond social cognition, reflecting a general bias in concept formation. For instance, the same conflation appears in non-social contexts, like folk biology: Children think about animals in a way that also places a heightened focus on category ideals and beliefs about how animals should be. Although adults tend to view the cheetah that runs the average speed for a cheetah as the most representative of what cheetahs are like in general, young children view the very fastest as most representative (Foster-Hanson & Rhodes, 2019a). And it's from the most ideal animals, such as the two fastest cheetahs, that young children think they can learn the most in general, whereas adults and older children prefer to learn from diverse samples that account for within-category variability (e.g., the fastest and slowest cheetahs; Foster-Hanson et al., 2020).

Children learn about what’s normal for different categories by listening to the adults around them (Foster-Hanson, Cimpian, Leshin, & Rhodes, 2018; Foster-Hanson, Leslie, & Rhodes, in press), so concepts are shaped by cultural ideals from childhood. For example, my colleagues and I found that children’s beliefs about what’s normal for gender categories show a pro-White bias in the United States, with children coming to think of White men, boys, and girls as the most representative of their gender categories by the time they're in third grade, regardless of their own racial group membership (Lei, Leshin, Moty, Foster-Hanson, & Rhodes, 2021). Indeed, we suggest that adults’ intersectional race and gender concepts are also skewed by cultural ideals, centering around what would be ideal for the dominant group in society—i.e., White men (Lei, Foster-Hanson, & Goh, under review).

Given the importance of cultural input in shaping conceptual development, a key focus of my ongoing research is testing how variations in cultural experience shape children’s reasoning about the natural and social worlds. For example, I recently mentored Josie Benítez (a lab coordinator, now doctoral student, in Dr. Rhodes’s lab) on a longitudinal study tracking how variation in the way parents talk to their children about gender stereotypes relates to their children’s developing gender beliefs (Benítez, Foster-Hanson, Brenner, & Rhodes, in prep). I'm also mentoring a master’s student on her thesis testing how explanations shape children’s biological reasoning (Foster-Hanson, Ziska, & Rhodes, in prep). Both projects use unmoderated remote research, a technique I was instrumental in helping to develop (Rhodes, Rizzo, Foster-Hanson, et al., 2020). By allowing children and parents to participate in studies from home whenever it suits them, while recording their speech and behavior through their own webcam, unmoderated remote research is ideal for recording longitudinal data and for investigations of cross-cultural variation, both of which I'll continue in future work. Unmoderated remote research also facilitates inclusion of larger and more diverse samples of participants, as part of my ongoing commitment to robust and replicable open science. For example, I make my data, analysis code, and materials for every project available on the Open Science Framework, and I pre-register all new studies. I also make as much video data as possible available to authorized researchers on Databrary.org, including videos of the full procedure for studies with children.