Primates love their grooming.

They do it day and night, picking stuff up each other’s bodies, dirt, ticks, the works. Grooming is not only good for hygiene; it helps build relationships, earn social standing, decrease conflict, and exchange resources. Humans, being the big-headed mammals they are, have taken grooming far beyond. We make phone calls, send WhatsApp messages, and comment on the Instagram posts of our friends. These acts of virtual touch shape our networks of relationships: the more we talk, the stronger our social bonds are.

If this talkative grooming is so good for us, releasing endorphins and making us happy, why don’t we do it all the time, with everyone we meet? Well, it turns out the world is not entirely our oyster. Maintaining meaningful relationships takes effort. We need to talk, remember each other’s birthdays, keep track of our shared pasts, and imagine futures. In short, it keeps our neocortex busy. And even mindless spats on Twitter last for a while. These constraints on time and processing capacity limit our number of friends. For decades, researchers have found regular patterns in various types of social interaction; we focus most of our finite attention on a few close friends and family, leaving the rest for progressively more distant acquaintances.

So how general is this pattern, really, given that our modern lives are so inundated with fancy communication technology? Do we distribute our time similarly when meeting face to face or on social media? And how do we end up favoring a few friends (or random strangers online) over others? Are we all the same type of groomers in how we talk?

To answer these questions, we collected as much data on tech-enabled social interactions as we could get our physicist hands on. We got anonymized contact data on 1.9B calls and messages sent by millions of people in Europe. Then there were millions of messages between 100k users of online channels: emails at companies, universities, and research institutions; messages in social platforms, movie recommendation forums, and even dating websites. We used all this data to build egocentric networks of communication; networks of social ties around an individual, where the weight of a tie is proportional to the number of calls, messages, or online interactions between two people.

What did we find? Well, people are remarkably similar in the way they distribute contacts among friends, regardless of the communication channel used. Whether it is with our iPhones or on Facebook, most people have a few best friends and little time for the rest of the world. And yet, there is individual variation: The distribution of weights in an egocentric network, what we call a social signature, is flatter for some, implying these people spread contacts more homogeneously in their networks.

Even more strikingly, this individual behavior is persistent in time; you may continuously make and lose friends in your favorite toxic online platform, and still how you distribute contacts at any given moment (the shape of your social signature) will remain the same. Curious, right?

When looking at egocentric networks growing over time, we noticed something else. The probability of contacting a friend again increases with the number of interactions you’ve already had together, meaning next time, you’ll probably reach out to your best pals rather than that distant college roommate you don’t actually like. It immediately turned our universality senses on, as this rich-get-richer mechanism appears everywhere, from cumulative advantage in social inequality, to proportional growth in economics and biology, and preferential attachment in the growth of techno-social networks.

So we made a model. Something as simple as possible, but with the hints the data was already shouting at us. In the model, people contact each other with probability proportional to their previous activity, but tuned by a parameter we call preferentiality. This parameter measures all constraints in your ability to communicate (available time, neocortex; leading to the total number of contacts you can possibly make) relative to your tendency to select friends at random (a measure of everything we don’t know beyond constraints). As time goes by, people with high preferentiality focus on their closest friends, while the rest share their time among friends more equally.

After lots of math and computations, we adjusted the model to everyone in the data, giving us a single, ‘magic’ number encapsulating the social signature of each individual. In a channel, the preferentiality of a person is constant even if their friends come and go, and the distribution of preferentiality values across an entire population is the same across channels. Whether it’s phone callers, posting millennials, or trolls in Twitter, many modes of off- and online communication show the same types of talkative groomers, and in roughly the same amounts: across channels, 66% to 99% of people preferentially talk to their best friends.

Cool, but what does this all mean? While it might be obvious that people like spending time with their friends and family in the real world, it’s kind of intriguing that the same patterns of behavior appear in platforms with online aliases or near anonymity. Maybe our primate brains are really just wired this way. Or maybe the rich-get-richer phenomenon we see is the effect of different underlying mechanisms, like homophily, emotional closeness, or ease of repeated interactions, all of which end up looking the same at the aggregated level of society.

It’s also intriguing that the shape of social signatures appears to be stable in time, even when the people in our egocentric networks might themselves change. Almost as if preferentiality was a behavioral trait that we take with us wherever we go. Having diverse social relationships correlates with increased longevity, better cognition while aging, and resilience to disease. It then seems crucial to figure out why people have the preferentiality values they have, to help them shape their social circles in a healthier way.

So, are our talkative grooming patterns really part of our personality? Are they stable over longer periods of time, or the same when we move from one channel to the other? What are the consequences of having such a diverse spectrum of groomers for society as a whole?

Sometimes, the coolest papers are those that make you ask even more interesting questions, while trying to answer a simpler one…

Wait, I really need to reply to this message, it’s from a good friend of mine…