Scholar Alexandra Carstensen on Early Diversity in Abstract Thought

By Catherine Xie, Teacher

Imagine you are presented with two images: a chicken and a patch of grass. After you are presented with a third image, a cow, which of the first two images do you pair it with? In other words, do you see the cow as belonging with the chicken or the patch of grass? According to cultural psychological research, your answer may provide some insight into your cultural background. When education psychologist Lian-Hwang Chiu and his colleagues presented this scenario to Chinese and American children in 1972, they found that Chinese children were more likely to categorize the cow and grass together, whereas American children were more likely to group the cow with the chicken. To explain this difference, Chiu and his research collaborators highlighted a cultural tendency that is one of the cornerstones of cultural psychology: Collectivistic societies (like China) emphasize relationships more than individualistic societies (like the United States). Accordingly, American children were more likely to classify the images based on category membership, grouping the cow with the chicken, while Chinese children more often made classifications based on relationships, grouping the cow with the grass.

This study is one of the inspirations for Stanford psychologist Alex Carstensen’s investigations into how culture, in addition to language, can play a role in abstract reasoning. A postdoctoral fellow in Michael Frank’s Language and Cognition Lab, Carstensen and her research assistants have been running a study at Bing since summer 2019, and she shared some of her findings with Bing teachers and staff in February at the school’s staff development day. Adults are good at abstract reasoning, easily being able to, for example, complete an analogy. Interestingly enough, American toddlers, and even infants, show sophisticated early abilities in relational reasoning (distinguishing between same and different pairs or groups of objects), but some of these abilities get worse over time before improving. Thus, their relational reasoning abilities can be represented by a U-shaped curve—dipping down before increasing again.

So why is there this dip in children’s ability to reason about relationships?

To explore possible causes for this phenomenon, Carstensen investigated two perspectives on relational reasoning: the relational shift “paradox” view and the rational learner “paradigm” view. The former suggests that language learning causes both the decline and reemergence of relational understanding by directing children’s thinking away from relations temporarily before later providing support for relational reasoning. The latter theory takes a broader view, suggesting that experience with language and the environment may vary across groups of children and create corresponding variation in learned biases for either relational or object-based reasoning.

To test for variation in relational reasoning across cultural and linguistic contexts, Carstensen and her colleagues examined early relational reasoning abilities in American and Chinese children. They considered the role that relational word acquisition may play: relational words, such as “speak,” “give” or “marry,” encapsulate relationships between entities. Because Chinese culture has a greater focus on relationships, and children learning Mandarin Chinese learn more relational words than American children learning English, the researchers thought that Chinese children may display relational reasoning abilities earlier than American children. They carried out their study in three parts. In the first experiment, 3-year-old children saw four pairs of blocks that were placed atop a music-making toy. Of the four pairs, two pairs consisted of identical blocks, while two pairs consisted of different blocks. In the “same” condition, the toy was activated and played music only when the identical blocks were placed on it. In the “different” condition, the opposite pattern was observed. The researchers then presented novel pairs of blocks to the children and tested their ability to select the “correct” pair to activate the music-making toy, depending on which condition they were in (see Figure 1). They found that while U.S. children’s selections appeared to be random, preschoolers in China correctly selected the pair exemplifying the relation (same or different) that had worked before, demonstrating successful inference about same-different relations.

The results from the first study led to their next question: Do these findings point to distinct developmental trajectories of relational reasoning, or are Chinese preschoolers simply further ahead on a similar developmental trajectory? Examining children’s performance from 18-to 48-months, they found that there are distinct developmental trajectories: Children in the United States perform well initially, and then decline to chance during this period (corresponding to the first half of the U-shaped curve) but not in China, where they show gradual linear improvement over the same period. Psychologists still don’t have a complete understanding of the time course for the U shape, but American adults perform this task with high accuracy, demonstrating a recovery sometime after age 4.

The paradigm view predicts that cultural variation in relational reasoning is the result of learned biases (rather than differences in attentiveness or ability). To test this prediction, researchers presented 3-year-olds with a task that was similar in structure to the one in the first study, but with a key difference: The children were given a forced choice between a relational and object-based solution, both of which were reasonable based on the examples the researcher had shown (see Figure 2). In this ambiguous scenario with no clear correct answer, they had to decide whether it was the recurring object or the relationship (that is, two different objects) that activated the music. The researchers found that Chinese children tended to select the relational match, while U.S. children preferentially selected the object match.

Based on their findings across the three studies, Carstensen and her colleagues concluded that there are indeed distinct developmental trajectories of relational learning across cultures. This difference suggests that certain kinds of reasoning are privileged in different cultural contexts.

This study lays the groundwork for future research that can potentially disentangle the specific features of the learning environment, including language, that influence children’s abstract thinking. Carstensen’s research invites us to contextualize children’s abstract reasoning within a culturally diverse society and appreciate a stunning truth about children’s learning: that it is dynamic, complex and ever-evolving.