The research is published in PNAS.
The role of semantic knowledge
Classical theories of human cultural progress typically treat human innovation as a random trial-and-error process: people come up with variety of ideas, successful ones are selected and transmitted across generations, and society gradually advances. However, this new study shows that people do not explore ideas randomly. Rather, our structured understanding of the world acts as a mental map, guiding us toward meaningful discoveries.
The findings are based on large-scale behavioural experiments involving over 1,200 participants, who completed a computerized innovation task aimed at discovering new items by combining existing ones from their inventories. Participants with access to familiar semantic information about the items consistently outperformed those working with abstract items lacking real-world meaning. Furthermore, the participants working with abstract items performed no better than chance by relying on shallow exploration strategies.
“Mental associations of the items narrow down the vast number of possibilities,” explains VU computer scientist Anil Yaman, adding, “They provide a direction for exploration that results in more discoveries than relying on lucky guesses.”
Combining knowledge with social learning
Prior work has largely focused on social learning - adopting innovations from others- as the catalyst for human progress. However, this study found a powerful interaction: combining semantic knowledge with social learning amplifies progress. With internal semantic knowledge, an individual does more than simply copy a peer’s idea; they grasp why it works and utilize this knowledge for a new discovery.
"From the invention of the earliest tools to the development of artificial intelligence, human progress relies on building on past successes," Yaman noted. "While we know that passing ideas to others is crucial, the internal spark that creates those ideas in the first place has remained largely unexamined."
Ultimately, the study demonstrates that our ability to make progress depends on the structured knowledge we construct in our minds. Next, the researchers plan to investigate how we form this knowledge in different domains and how we transfer it across domains in complex, real-world situations.
About the researchers
The study Semantic knowledge guides innovation and drives cultural evolution was co-led by Anil Yaman from the Department of Computer Science at Vrije Universiteit Amsterdam, together with Shen Tian and Björn Lindström from the Department of Clinical Neuroscience at Karolinska Institutet, bridging the fields of computational intelligence and cognitive science.