The ability to recognize, understand, and regulate emotions is an important component of human capital: the skills, knowledge, and abilities people possess that can contribute to economic and other outcomes . People with greater emotional intelligence exhibit stronger stress management, academic performance, and workplace functioning. These skills may be particularly important in the modern job market, where employers increasingly seek workers who can communicate effectively, respond to conflict, and manage interpersonal relationships.
Despite growing evidence on the importance of social and emotional skills, less is known about the childhood experiences that develop these skills and the norms that reinforce them. Textbooks and children’s books are one potential influence: through both text and images, they can model emotional experiences and convey expectations about which emotions are appropriate to display.

In this paper, the authors examine the emotions children encounter in educational materials across formal and informal settings. They analyze three decades of public-school textbooks from California and Texas alongside a century of award-winning children’s books commonly found in homes, libraries, and schools. They also incorporate data on household purchases and public library inventories to examine whether consumer demand may shape the emotions represented in children’s books.
The authors use artificial intelligence tools to identify eight primary emotions in both text and images: anger, calm, confusion, disgust, fear, happiness, sadness, and surprise. They also examine whether emotional portrayals differ across types of books, states, grades and subjects, as well as characters’ gender, age, race, skin color, and whether they are human or non-human.
Children’s educational materials present a broad range of emotions in words but a much narrower range in pictures: calm and happiness account for roughly 90% of emotions shown in images.
Their analysis reveals a systematic divergence between the emotions children read about and those they see in images:
- Text presents a broad range of emotions, while images overwhelmingly show happiness and calm. Happiness (29%) and sadness (20%) are the most common emotions in text, with other emotions also appearing frequently. In images, by contrast, calm and happiness account for roughly 90% of emotions shown. This mismatch persists even within the same page: when text describes anger, fear, sadness, or surprise, those same emotions only account for on average 8-13% of image emotions on the same page.
- Nearly half of pages show no overlap between emotions in text and images. Regardless of the emotion described in the text, images are most likely to show calm.
- The pattern persists across time, contexts, and identities. The divide between emotions represented in text and images remains stable across decades and appears in both children’s literature and textbooks, across subjects and grade levels, and across character identities (e.g., race, gender, age, skin color, and whether the character is human or non-human.)
- Consumer demand may help reinforce the emphasis on positive imagery. Household purchases and public-library collections show that purchasers are less likely to choose books whose covers depict emotions traditionally considered negative, even when comparing books with similar characteristics.
The authors test whether these results depend on their AI-based method by comparing their findings with alternative approaches to classifying emotions and with classifications made by human annotators. Across these checks, they recover the same central result: text displays both positively and negatively categorized emotions, while images overwhelmingly show happiness and calm.
The findings highlight how children’s educational materials may help shape emotional skills and norms. While text exposes children to a broad range of emotions, images reinforce a much narrower range centered on happiness and calm. This pattern may limit the emotions children see modeled during important years of emotional development, pointing to the value of representing a wider range of emotions in both words and images.







