Identifying the necessary conditions for large populations to enhance cumulative culture

Why more people don’t always mean better ideas

From the spread of farming to the rise of smartphones, humans have advanced by sharing and improving ideas over generations. A common assumption is that bigger populations automatically fuel this kind of cultural progress: more people should mean more ideas and faster innovation. Yet past studies have delivered a puzzle—sometimes larger groups do better, sometimes they do worse. This paper asks a simple question with big implications: under what conditions do large groups truly help ideas accumulate and improve over time?

Many minds, good and bad ideas

The authors focus on “cumulative culture,” the process by which useful behaviors and technologies are gradually refined, far beyond what any single person could design alone. Theory suggests that larger groups should be an advantage because they contain more varied solutions and more chances for clever combinations. But there is a catch. Along with more good ideas come more bad ones, and people have limited memory and attention. If we cannot quickly spot and copy the best options in a crowd of mixed-quality solutions, the apparent benefit of a big population may vanish—or even turn harmful.

Testing how people learn in small and large groups

To unpack this, the researchers ran two large experiments with a combined 941 university students. Participants played a computer game in which they designed virtual arrowheads on a grid. Each design earned a hidden performance score based on features like size and symmetry. Over ten rounds, people either worked alone or in groups of three or six. In group conditions, players could see other people’s arrowheads and their scores before making a new design. This setup let the authors observe how well different kinds of social learning helped groups move toward better and better arrowheads across rounds.

Focusing attention in a sea of options

Experiment 1 asked whether the key to unlocking big-group benefits is “attention filtering” – the ability to focus on one high-performing example instead of trying to weigh everything at once. In one condition, participants could choose a single arrowhead from their group to watch being drawn again, naturally steering them toward the best-scoring design. In another, they were forced to view every group member’s arrowhead in random order. The results were clear. Social learning beat solo learning overall, but group size only helped when attention filtering was allowed. In the six-person condition where people could home in on just one top solution, performance climbed higher than in three-person groups. When everyone had to process all designs, large groups lost their edge, and improvement stalled despite access to more ideas.

Letting the world hold our memory

Experiment 2 turned to a different aid: external records. Here, all group members again had to view each arrowhead, but in one condition the designs and their scores disappeared before the next round, while in another they were stored in a visible “gallery” at the bottom of the screen during design. This persistent record acted like a shared notebook or whiteboard. Once again, social learning beat working alone in every case. Crucially, a benefit of larger groups emerged only when the external record was available. With that extra visual memory, participants in six-person groups could better compare options, focus on high scorers, and sometimes blend features from several designs, leading to more effective arrowheads over time.

What this means for innovation in everyday life

Taken together, the studies show that large groups do not automatically generate superior technology or solutions. Instead, big populations help culture accumulate only when people have tools to manage information overload—either by filtering attention to a few promising models or by offloading details to external aids like diagrams, notes, or digital displays. For everyday life, this suggests that systems that highlight quality (such as ratings, recommendation algorithms, or citation counts) and that provide clear shared records (such as collaborative documents or online repositories) are crucial if we want large communities—whether research teams, companies, or whole societies—to turn many minds into steadily improving ideas.

Citation: Walker, B., Fay, N. Identifying the necessary conditions for large populations to enhance cumulative culture. Sci Rep 16, 10090 (2026). https://doi.org/10.1038/s41598-026-40973-x

Keywords: cumulative culture, social learning, population size, attention filtering, external representations