AI Uncovers Hidden Patterns in the Evolution of Human Emotion and Thought

Although the emotions and thoughts of past generations left no direct trace, cultural artifacts such as paintings, literature, and music serve as “cognitive fossils” that offer insight into historical psychology. Today, artificial intelligence (AI) is enabling researchers to mine these vast cultural records to uncover patterns in the evolution of human emotion and cognition over time.

Since the proliferation of printing in the 15th century, more than 160 million book titles have been published globally, forming a rich foundation for textual analysis. For instance, a team from the University of British Columbia analyzed the usage of the Chinese character "心" for “heart” in pre-Qin texts, demonstrating that ancient Chinese thinkers distinguished between mind and body. Meanwhile, researchers at the École Normale Supérieure in France found that references to love in literature rose significantly alongside economic growth, supporting the hypothesis that material prosperity fosters romantic narratives.

Paintings, too, reflect the emotional currents of their eras. Economists at the University of Toronto trained AI to analyze 631,000 artworks created between 1400 and the present, revealing strong links between artistic styles and historical events. During the warming phase of the Little Ice Age (1500–1700), for example, depictions of fear and sadness declined. In Germany, positive sentiment in art peaked around 1850, fell steadily thereafter, and only recovered after World War II. By detecting facial expressions, weapons, and other visual cues, AI has shown an ability to replicate human interpretations of art.

Despite its power to process massive datasets, AI's insights depend on human-defined labels and algorithmic choices. Researchers caution that cultural data is inherently biased: historically, access to literature and art was limited to elites. Thus, AI can only analyze the expressions of a specific social stratum. Still, AI offers historians and psychologists a valuable tool to augment traditional methods. With improved data annotation and collaboration with subject experts, AI-based studies could open new, systematic ways to explore the interplay between human behavior and social change.