Summary: A new study identifies semantic knowledge—the mental map of how concepts connect and function—as the central cognitive engine behind human innovation and cumulative cultural evolution. The research shows that understanding relationships between things is a strict precondition for producing meaningful inventions; without that internal map, human creativity reduces to random guesswork even when social learning is available.
Combining an agent-based computer model of cultural change with a large behavioral experiment involving 1,243 participants, the investigators demonstrate that semantic knowledge both directs exploration toward useful solutions and multiplies the value of social learning. The findings reshape how we understand cultural inheritance: previous generations pass on not just artifacts but a conceptual toolbox that makes future innovation possible.
Key findings
- Semantic knowledge as the overlooked engine: While many theories emphasize social learning and motivation, this study highlights an internal cognitive resource—semantic knowledge—that encodes how objects and concepts relate and how those relations can be applied. This knowledge is the mental scaffold that lets people combine components in purposeful ways.
- Large-scale experiment with real and abstract stimuli: More than 1,200 participants played a specially designed computer game that required creating novel “innovations” by combining items. One group worked with familiar, meaningful items (for example, rocks and branches); another performed the same task using arbitrary, abstract symbols stripped of semantic content.
- Human performance collapses without meaning: Participants who could use semantic information reliably found serviceable, high-functioning combinations. Participants deprived of semantic context performed at the level of random baseline agents—no better than uncalibrated bots—even when they could observe other players’ results through social learning.
- Synergy between semantic knowledge and social learning: When people had both a semantic understanding of component relations and access to peers’ solutions, groups produced roughly twice as many unique innovations as groups that only had social learning. Semantic knowledge thus amplifies, refines, and accelerates cumulative cultural change.
- Transmission of a conceptual toolbox: The results suggest cultural inheritance transmits more than finished artifacts or techniques; it transfers an active blueprint for how the world works—an organized set of concepts and relations that makes purposeful innovation possible.
- The paradox of strong priors: The researchers note that dense, well-developed semantic maps can sometimes impede breakthrough discovery. Strong conceptual priors may bias attention away from counterintuitive or unconventional solutions, a question the team plans to explore in more complex, real-world settings.
Source: Karolinska Institutet
This study asked a fundamental question: what makes humans especially good at inventing new ideas and technologies? The evidence points to semantic knowledge—our stored understanding of object properties, functions, and the relationships among concepts—as a decisive factor. By biasing search toward plausible combinations, semantic knowledge offers an efficient filtering mechanism that spares the brain from millions of fruitless trials.
In the experiment, participants who could draw on real-world semantics quickly converged on useful combinations. Participants working with meaningless abstract symbols, in contrast, behaved like shallow explorers and rarely found functional solutions. Crucially, having access to others’ solutions was not enough to overcome the lack of semantic grounding: social learning becomes powerful only when learners can interpret and build on what they observe.
Björn Lindström of the Department of Clinical Neuroscience at Karolinska Institutet summarizes the implication: “Semantic knowledge is our cognitive toolbox. It helps us understand which elements can work together and why.” The research team combined these behavioral results with an agent-based cultural-evolution model that reached the same conclusion: agents endowed with internal knowledge maps innovate far more effectively than agents that combine components at random.
Beyond demonstrating the essential role of semantic representations, the study highlights how these representations interact with social transmission to produce cumulative culture: ideas are not only passed on but are amplified and improved across generations when learners inherit both artifacts and the concepts that explain them.
The team’s next steps will explore how semantic knowledge functions in richer, more realistic environments and under what conditions strong priors hinder rather than help radical innovation. Understanding when conceptual scaffolding becomes a constraint is important for designing educational systems, research environments, and collaborative platforms that foster genuinely novel discovery.
Funding: The project was conducted by researchers at Karolinska Institutet and Vrije Universiteit Amsterdam, financed by the European Research Council (ERC) and the Knut and Alice Wallenberg Foundation. No conflicts of interest were reported.
Key questions answered
A: Copying surface-level behavior without understanding why it works is empty mimicry. The experiment showed that participants facing abstract, meaningless tasks performed no better than random agents even when they could observe peers. Social learning becomes effective only when learners can decode the underlying rationale and adapt it—functions provided by semantic knowledge.
A: Semantic knowledge acts like an internal map that predicts how properties interact. Seeing a rock and a branch, your mind immediately evaluates likely uses based on prior knowledge. This preselection filters out implausible combinations, focusing exploration on promising avenues and saving cognitive effort.
A: Yes. Strong semantic priors create entrenched expectations about what is reasonable. While these priors are invaluable for everyday innovation, they can blind innovators to counterintuitive or unconventional ideas—an empirical paradox the researchers plan to investigate further.
Editorial notes
- This article was edited by a Neuroscience News editor.
- The journal paper was reviewed in full by editorial staff.
- Additional context was provided by the reporting team.
About this research
Author / Contact: Press Office, Karolinska Institutet. Source: Karolinska Institutet. Image credit: Neuroscience News.
Original research: “Semantic knowledge guides innovation and drives cultural evolution” by Anil Yaman, Shen Tian, and Björn Lindström. Published in PNAS. DOI: 10.1073/pnas.2530750123. (Closed access)
Abstract
Semantic knowledge guides innovation and drives cultural evolution
Cultural evolution enables ideas and technologies to accumulate across generations, with humans achieving particularly complex and open-ended cultural change. While social learning transmits innovations, the cognitive processes that generate them have been unclear. Classical models often treat innovation as random variation, which fails to explain human cultural complexity. We propose that semantic knowledge—the associations that link concepts to properties and functions—guides human innovation and underpins cumulative culture.
Using an agent-based model and a large behavioral experiment (N = 1,243), the study finds that semantic knowledge focuses exploration on meaningful solutions, increases innovation success, and supports generalization from prior discoveries. Semantic knowledge also synergizes with social learning to amplify innovation and accelerate cumulative change. Participants lacking semantic access performed at chance levels and relied on shallow exploration strategies. Together, these results identify semantic knowledge as a core cognitive process driving human cumulative culture.