On winter evenings, millions of starlings fly into Rome from the surrounding countryside to roost. Before they settle, they perform. Thousands of birds wheel over the rooftops as one dark shape, folding, stretching and turning faster than any single bird could announce a change.
Nobody leads a murmuration. Each starling reacts only to the birds around it, which raises a simple question with a surprisingly exact answer: how many of them does it watch?
Counting the neighbors
A team of physicists in Rome mounted synchronized cameras on a roof near the city’s main railway station and filmed at dusk. Stereo images let them rebuild the three-dimensional position of every bird in flocks of several thousand. (Ballerini et al., 2008Ballerini et al. (2008), “Interaction ruling animal collective behavior depends on topological rather than metric distance: Evidence from a field study”, PNAS doi.org/10.1073/pnas.0711437105 All sources)
The expected answer was a radius: a starling would track everyone within a certain range. The data disagreed. Each one coordinated with a fixed number of neighbors, six or seven, whether those birds flew close or far apart. When a predator attacks and the flock thins out or squeezes together, every individual keeps the same few in view, so the group stays intact. A rule based on distance would let it tear. (Ballerini et al., 2008Ballerini et al. (2008), “Interaction ruling animal collective behavior depends on topological rather than metric distance: Evidence from a field study”, PNAS doi.org/10.1073/pnas.0711437105 All sources)
Why seven
Five years later, engineers at Princeton, working with the Rome researchers, asked why the count landed where it did. They modeled each flock as a network and measured how well birds reached agreement despite noisy, uncertain information. Tracking more neighbors improved consensus, but every extra bird cost attention. Six or seven gave the best balance between cohesion and individual effort. (Young et al., 2013Young, Scardovi, Cavagna, Giardina & Leonard (2013), “Starling flock networks manage uncertainty in consensus at low cost”, PLOS Computational Biology doi.org/10.1371/journal.pcbi.1002894 All sources)
I’ll admit a bias here: seven has always been my favorite number. The bird version rests on measurement, though, and it comes with a caveat. The optimal count shifted with a flock’s shape, especially its thickness, rather than its size.
The human limit
People have a famous limit of their own. In 1956, psychologist George Miller described “the magical number seven, plus or minus two” as the span of items we can hold in mind at once. Later work trimmed it. Nelson Cowan’s review puts the central capacity closer to three to five meaningful chunks for young adults. (Cowan, 2010Cowan (2010), “The magical mystery four: How is working memory capacity limited, and why?”, Current Directions in Psychological Science doi.org/10.1177/0963721409359277 All sources)
Whichever figure wins, attention works like a budget. A starling can’t track the whole flock, and no person weighs every voice in a room. Influence travels through whoever occupies that small circle. A guest at a crowded event judges the night by the few tables in view. A team member forms an opinion of a project from the handful of colleagues they hear from most.
Curate your seven
The useful move is to choose that circle on purpose. Name the sources, coworkers and numbers you actually respond to, then ask whether they represent the whole group or only its nearest corner. Anyone designing a launch, a venue or an org chart can assume each guest or employee will see only a handful of others, and make sure the visible ones are worth following.
A murmuration looks like one mind. Up close, it’s thousands of birds, each watching seven, and that turns out to be enough.
Sources
- Ballerini et al. (2008), “Interaction ruling animal collective behavior depends on topological rather than metric distance: Evidence from a field study”, PNAS doi.org/10.1073/pnas.0711437105
- Young, Scardovi, Cavagna, Giardina & Leonard (2013), “Starling flock networks manage uncertainty in consensus at low cost”, PLOS Computational Biology doi.org/10.1371/journal.pcbi.1002894
- Cowan (2010), “The magical mystery four: How is working memory capacity limited, and why?”, Current Directions in Psychological Science doi.org/10.1177/0963721409359277