What researchers studied

Large-scale computational content analysis using large language and multimodal models, alternative NLP and computer-vision pipelines, and human validation to compare emotional representation in text and images across public-school textbooks from the last three decades and influential children's literature spanning roughly a century; the study also analyzes household purchases and public-library holdings as suggestive evidence about demand. Public-school elementary textbooks and influential children's literature spanning decades of educational and literary material, supplemented by household book-purchase records and inventory data from 26 Seattle Public Library branches

What they found

  • Text exposed children to a relatively broad emotional repertoire, including happiness, sadness, anger, fear, confusion, disgust, surprise, and calm.
  • In images, happiness and calm together accounted for roughly 90% of detected character emotions, while negatively valenced emotions were rarely pictured.
  • Nearly half of pages showed no overlap between the emotions conveyed in the text and the emotions conveyed in the images on the same page.
  • The text-image mismatch was qualitatively robust across alternative classification approaches and persisted across time, contexts, and character identities.
  • Household purchase and public-library inventory patterns were consistent with greater selection of books whose covers showed happy or calm characters, although the study cannot establish exactly why purchasers preferred those books.

What the study does not prove

  • This is a September 2026 EdWorkingPaper and should not be described as a peer-reviewed journal article unless a later publication is independently verified.
  • The study documents emotional representation in educational materials; it does not measure whether the text-image mismatch causes changes in children's emotional development, behavior, academic performance, or well-being.
  • Primary emotion classifications were model-generated, although the authors validated them against human annotators and alternative non-generative methods; calm and happiness were especially difficult to separate.
  • The consumer-demand analysis cannot distinguish whether purchasers genuinely prefer positive imagery, believe it is more appropriate for children, or are simply more attracted to smiling faces, so demand should be treated as one possible mechanism rather than the sole cause of the pattern.

Evidence strength: Strong descriptive and computational evidence about representation, with multi-method robustness checks and human validation; not causal evidence about child outcomes; EdWorkingPaper.

Why this matters for families

Children may read stories about fear, sadness, anger, and confusion while repeatedly seeing faces that look happy or calm. This study suggests educational materials often present a narrower visual model of emotion than the stories themselves, but it does not establish downstream effects on children's development.

Noor interpretation

How Noor translates the evidence into practice

The Noor-relevant lesson is that emotional intelligence cannot be reduced to telling children to stay calm or positive. Educational materials may expose students to complex feelings in words while visually modeling a much narrower range. For educators who care about emotional intelligence, it is useful to normalize accurate recognition and language for frustration, confusion, disappointment, fear, and uncertainty alongside happiness and calm, without implying that this study proves a specific emotional-learning intervention will improve academic outcomes.

Noor Lyra Educators can use emotionally varied examples, discussion prompts, and reflective language when students encounter challenge: name confusion before resolving it, distinguish frustration from inability, and model that difficult emotions can coexist with productive learning. Keep this as an instructional and communication principle rather than a clinical or therapeutic claim.

Read the original source

Noor links to the original or authoritative source so families can distinguish the evidence itself from our interpretation.

Open original source →

DOI: 10.26300/hgn9-n385

Research notes

emotional intelligencecurriculumchildren's bookssocial emotional learningtextbookseducation researchlearning environment

No single study determines a student's plan. Noor uses research as one input alongside the learner's goals, observed performance, academic context, and response to instruction.