When homing pigeons fly together to find their way home, they don't need a leader or a complicated decision-making process—they just average out the routes their companions take. Scientists at the University of Massachusetts Boston have discovered this elegant navigational strategy, and the finding overturns some longstanding assumptions about how social animals learn from each other.
The research, published in eLife, matters because it reveals that sophisticated outcomes—like finding progressively better routes home—don't always require sophisticated thinking. Understanding how animals navigate together is more than academic curiosity: it touches on fundamental questions about how groups make decisions, how knowledge spreads through populations, and what cognitive abilities are truly necessary for complex behavior to emerge.
Shoubhik Banerjee, a Ph.D. student in senior author Albert Kao's lab, led the investigation alongside postdoctoral researcher Fritz Francisco. They built on a 2017 study that had shown pairs of pigeons could continuously improve their homing routes over multiple generations. That earlier work proposed the mechanism was cumulative cultural evolution—chains of birds exploring different options and choosing better ones. But exactly how that improvement happened remained a mystery.
To crack it open, the team set up an ingenious experiment with pigeons released 8.4 kilometers from home. They created "chains" of five generations, where each new naive bird was paired with an experienced bird that had completed the previous leg of the journey. Every twelve flights, the experienced pigeon was retired and replaced with a new naive one, passing knowledge along like a game of telephone played over miles of sky. Solo birds and fixed pairs of the same birds served as control groups, each completing the same total of 60 flights.
The experimental chains significantly outperformed both solo and fixed pair controls by the fifth generation—a dramatic result that demanded explanation. The researchers didn't assume the answer was complex. Instead, they developed seven different learning mechanisms ranging from the simplest to the most cognitively demanding. The simplest involved no knowledge at all: birds simply averaged the routes of their companions. The middle tier assumed birds could recognize who was more experienced. The most complex required birds to actively evaluate their own and their partner's performance, something that aligns with what scientists call the mechanistic criteria for cumulative cultural evolution.
When they ran the numbers, something surprising emerged: all seven strategies produced route improvements. Even the simplest mechanism—pure averaging—replicated the experimental results. This suggests that homing pigeons likely rely on the most straightforward approach: they fly a path, their companions fly theirs, and the group's actual trajectory becomes a blended average of those individual routes. Repeated over generations, this simple process produces increasingly efficient paths home.
The finding reshapes how scientists think about animal cognition and collective behavior. It suggests that "wisdom of crowds" reasoning—where averaging individual decisions produces surprisingly good outcomes—may be far more common in nature than previously suspected. Rather than requiring birds to track who knows best or consciously evaluate performance, natural selection has favored a mechanism so simple that it barely qualifies as learning at all. Yet from that simplicity emerges something that looks indistinguishable from cultural evolution. The discovery opens new questions about which animals use similar strategies and how such elegant solutions evolved across the natural world.