Inherent aspects of social media that amplify polarization (short version)

If broadcast TV was an opiate of the masses, is social media an amphetamine?

This is an abridged version of a longer piece. If you’re interested in polarization, check out all my podcast episodes related to that.

Is it possible that social media plays a role in amplifying our divides and increasing political polarization?

Much of the thinking around this idea has focused on specific product features, like the use of addictive UI design (as examined in The Social Dilemma), or content-recommendation algorithms that show people increasingly extreme content, or Facebook using private data for political advertising (as examined in The Great Hack).

But what if the focus on product feature choices distracts us from something more fundamental? What if internet communication, by speeding up and distorting human interactions, agitates us and amplifies some of the darker aspects of our social psychology?

For some examples of how there may be inherent aspects of social media that amplify divides and extreme thinking, psychology research supports all of these ideas:

  • Writing down our beliefs makes us more committed to those beliefs.

And all of these things happen more frequently on the internet than they do in traditional in-person encounters. In this piece, I go into detail about the above research and other studies. I bring together ideas that might help us answer the question: If social media contributes to dividing us, what are the psychological processes by which it does that?

An admittedly unscientific Twitter poll that nonetheless captures the beliefs that users of social media have about the technology.

Basics of our tribal psychology

Before getting to social media effects, let’s briefly examine some basic psychological factors that seem to point to the underlying root causes for the tendency of humans to form into aggressive us-versus-them dynamics:

  • Out-group homogeneity effect: We view our own group as being made up of individuals with diverse views and personalities. But we view the other group (the out-group) as a mass of single-minded people: a monolithic, homogeneous group.

Our language plays a big role in the polarization process. To take one example of inaccurate language that increases societal polarization: it’s common to see liberals say things online like, “All Trump supporters are racists.” But the simple fact that there are black Trump supporters, and other people of color who support Trump, should be sufficient evidence that this belief that the other group is all the same (out-group homogeneity effect) is not correct. Our perception of a group as monolithic doesn’t make it true. Our perceptions are not other people’s perceptions. (I could have just as easily used an example of inaccurate, polarizing Trump supporter language: the point is the same.)

And the more one group speaks in all-encompassing, unreasonably hateful, inaccurate ways about another group, the more the other group is going to return the favor.

Before getting to the specific psychological effects, let’s quickly examine how some of these things tend to play out and amplify each other online:

  • Members of one political/cultural group say things like “If you’re a member of this other group, you’re a moron.”

And so on, in an amplifying cycle, with more and more people involved and more and more group-vs-group animosity involved.

Increasing group polarization means that an in-group will tend to reduce the out-group to being homogeneous and represented by their worst members. The other group will return the favor. Things escalate.

To be clear: this is not to say that social media is the primary cause for polarization problems. Looking at history, it’s clear that humans pretty frequently come to be at each others’ throats, with or without help from technology. But what we’re examining is the ways in which internet communication may be an efficient amplifier of our natural tribal tendencies.

Now let’s take a closer look at the processes by which social media may be amplifying such tendencies.

Factor #1: Social media makes us more stubborn

For almost all of history, most conversation was a private affair, ephemeral and quickly fading from memory. It’s new and unusual to have so much of our conversation publicly on display and recorded for others to see later. It’d be surprising if such a big change didn’t have an effect on our minds and society.

In a 1955 study by Girard and Deutsch, they found that writing things down, whether privately or publicly, made people less likely to change their minds. Robert Cialdini’s book Influence summarizes this study:

The students who had never written down their first choices were the least loyal to those choices. […] Those who had merely written their decisions for a moment on a Magic Pad were significantly less willing to change their minds when given the chance. […] By far, it was the students who had publicly recorded their initial positions who most resolutely refused to shift from those positions later. Public commitment had hardened them into the most stubborn of all.

Another study supporting this idea is 2020 study by Strandberg et al. They had people fill our surveys on various political topics and then manipulated some of the polarized answers to be more neutral. 94% of the subjects defended the manipulated answers as their own. Their belief that they’d stated a different opinion made them internalize and defend that opinion. Even more surprising: their attitudes “even persisted one week later.” If we feel a need to defend and internalize things we’ve never said, we must therefore feel a need to defend and internalize things we’ve said, no matter how quickly-conceived or ill-thought-out.

Using social media means frequently writing down our beliefs. There is a pressure to express an opinion on events and topics. And many of these statements of ours will be very quickly conceived and barely thought out. These are the kind of thoughts that, in a pre-digital world, might simply be the starting point of us building more nuanced and less polarized views. Social media, by inducing us to make all sorts of statements, especially publicly and especially simplistic takes, may be hardening our beliefs, making us resistant to listening to others and to changing our minds.

For many people, changing your mind is an indicator of weakness. People use terms like “flip flopping” or “waffling” to insult mind-changers. It’s possible social media, by having us write down and publicly commit to ideas, is making us more stubborn.

Factor #2: Social media promotes negative emotions

There’s a lot one can get upset about these days. The internet gives us awareness of so many events from across the country and the world, in a way that’s unprecedented in human history. At any given point in time, there are many horrible things happening on earth. This has always been the case but now we can easily learn about many of these things, and learn about them in detail. We perhaps aren’t naturally well equipped to deal with so many sources of anger, fear, and sadness.

And we seem to be hard wired to pay attention to negative emotions. From an evolutionary point of view, negative emotions are associated with threats, so it makes sense that we respond more to those than more relaxing, positive emotions. It’s been shown that social media messages with more emotional language get more attention and shares. It’s also been shown that anger is more influential than other emotions for spreading messages online.

As the saying goes: “A lie can travel around the world and back again while the truth is lacing up its boots.” One reason a lie, or just an angry take, can travel fast is that it isn’t constrained by the need for nuance and context; it can tell a simple good-versus-evil story that appeals to our emotions.

Fake news stories are at their most harmful when they have an emotional quality to them, because they become effective tools for arousing anger and violent responses. For example, there have been many instances in India of violence caused by fake news. To take another example: there have been riots in Minneapolis caused by fake news.

As anger proliferates online, social media allows one political group to view the most angry, unreasonable views of the other group. Due to the out-group homogeneity effect, we tend to view the other group as more uniformly angry and unreasonable than they are. This causes us to become more angry than is warranted, and that anger sparks more anger in the out-group, and so on.

Social media allows in-group members to view angry messages from the out-group, which in turn sparks more anger, and so on.

Social media, with its lack of nuance, speed-of-action, and promotion of passionate takes, makes angry over-reactions and misinterpretations common. This ramping up of anger in one group in turn makes the other group more angry (or at the very least more likely to view the first group’s concerns as silly and unserious).

Factor #3: Social media is distant and dehumanizing

A 2017 study by Abe Rutchick showed that it’s easier to kill when we’re physically distant from the killing. If it’s easier to kill at a distance, it stands to reason it’s easier to be unreasonable or insulting or threatening when interacting with people many miles away.

Audio/video call tools, whether the telephone or Zoom, are also long-distance communication tools, but they force a more traditional human interaction. We are more cognizant of the other person’s humanity due to their voice or their facial expressions. Seeing and hearing other people activates our best social instincts.

Also, pure-text communication leads to more misunderstandings, and hence more anger, than we’d otherwise have.

Factor #4: Social media breeds insults, which amplify group grievances

Due to the distant nature of the internet, and due to our increasingly polarized landscape, insults are common online. And the internet allows you to easily find insults to your group; some group members promote such insults in order to rile up their own group.

Social psychologist Karina Korostelina is the author of ‘Political Insults: How Offenses Escalate Conflict.’ In my interview with her, she said that the internet had likely increased human conflict due to greater creation and perception of insults.

A 1967 study by Robert Abelson et al showed that when subjects were insulted during the course of a discussion, that “increased the extremity of [their] initial attitude position”, in a type of boomerang effect.

Next time you’re insulted online, study how you feel. Do you feel an instinctual us-versus-them anger? Even if you logically know that the person insulting you doesn’t represent their entire large group, and even if you know there are also mean and unreasonable people on your own side, there can be that instinctual feeling of “this entire group is as bad as this one person” when we have such interactions.

Factor #5: Social media fosters familiarity, which can breed contempt

Political polarization might be part of a broader pattern of social media just generally increasing our dislike of each other. In a 2007 paper Less Is More: The Lure of Ambiguity, Michael Norton’s research team found that familiarity, as the saying goes, can breed contempt.

Knowing little about the people around us can have value as a social lubricant. Without reason to believe otherwise, we tend to assume that other people are like us in various ways. The more we know about someone, the more likely we are to discover something we dislike. This becomes even more true in an increasingly polarized society.

In Jaime Settle’s book ‘Frenemies: How Social Media is Polarizing America,’ she showed that Facebook use was associated with an increase in people’s animosity towards members of the other political group. The primary way this seemed to happen was that Facebook allowed people to more easily place their non-close acquaintances (e.g., a hairdresser, a babysitter, a teacher) in in-group/out-group categories and interpret posts through that lens, in ways that would happen less frequently pre-internet. (Here’s my interview of researcher Jaime Settle on these topics.)

Another unscientific Twitter poll of mine. A good percentage of social media users seem to think social media has negatively impacted their relationships even with people on their political side.

Factor #6: Social media removes normal social context

Interactions on social media happen outside of the normal frames of reference that have traditionally defined human relationships. In pre-internet days, we interacted with people based on various frames of reference, like meeting as neighbors, or as close friends, or as fellow churchgoers. Our interactions were guided by that frame.

But online, this helpful context is stripped away. Communication researchers call this context collapse. This can lead to us misunderstanding each other more, making us less likable to each other.

To take a personal example: my wife posted a message about veganism on Facebook, and an old friend of mine responded with something like, “I’m so glad we all have so much free time that we can worry about stuff like this.”

It was out of character for him to be rude. He later apologized and explained that when he made that comment, he’d just come from spending time with a friend whose child had died. When he saw my wife’s post, his immediate reaction was, “How petty to be worried about this with all the tragedy in the world.”

But the way we perceive the world at a specific moment seldom aligns with others’ perceptions and priorities at that point in time. Even our own perception of what’s right or important can change throughout the day; something you posted online yesterday may seem tone-deaf and embarrassing to you tomorrow. Most of the things any of us do or say will seem petty and unimportant to someone somewhere, especially when compared to issues of life and death.

There’s always something that can be deemed “more important” happening somewhere, so any post can theoretically produce an angry “why are you worried about this?” type of response.

In my example, my friend would likely never have said such a thing to my wife in pre-internet days. And that’s because in-person interactions give us helpful frames of reference, and activate our more charitable social instincts. In-person socializing fosters a natural give-and-take dynamic, while social media communication takes the form of a series of isolated statements.

Without the social context we are accustomed to, we can appear to each other petty, heartless, hard to understand. We will frequently misunderstand each other and assign motives and feelings to each other that aren’t there.

Factor #7: Social media gives power to the more extreme

People who are most active on social media don’t represent most of society. From Pew research: “Most users rarely tweet, but the most prolific 10% create 80% of tweets from adult U.S. users.”

And most Americans don’t follow politics closely. From an October 2020 New York Times article:

What we found is that most Americans — upward of 80 percent to 85 percent — follow politics casually or not at all. Just 15 percent to 20 percent follow it closely (the people we call “deeply involved”).

A 2018 survey about U.S. polarization showed that we’re pressured to conform to the beliefs of people around us. Pressure to conform to one’s group has always been present, but social media increases our anxiety of being judged publicly, in full view of everyone, and of being judged by multiple people in our own group (maybe even hundreds or thousands).

Such anxiety understandably leads to fewer people criticizing their group online. And this dynamic creates the perception that the more extreme and angry views are more common and unquestioned than they are.

Yet another unscientific Twitter poll. Maybe interesting: my Twitter audience is mostly liberal.

If you’re liberal and have felt some pressure to not criticize some liberal thinking that you find unreasonable, then you can perhaps relate to the pressure some moderate conservatives might feel to fall in line as their friends and family become more extreme (and vice versa). We all want to think the best of the people “on our side”: we want to assume they’re good, reasonable people. But this in-group favoritism can lead us astray when the more extreme people hold undue influence.

Factor #8: Social media makes like-minded groups grow more extreme

The term group polarization describes the psychological effect of how like-minded groups grow more extreme over time. There are quite a few studies about group polarization and social media: one 2010 study of Twitter users found that “replies between like-minded individuals strengthen group identity, whereas replies between different-minded individuals reinforce in-group and out-group affiliation.” The 2020 study by Steven Johnson et al showed Facebook led to increased polarization; they wrote “Facebook indeed serves as an echo chamber[…]”

The internet can be viewed as a tool for bringing like-minded people together and giving them spaces to craft and spread ideas. And that’s great in many ways, but it also has negative effects. Before the internet, it would have been hard to get many people together in one place to discuss whether the earth might be flat; now it’s easy. And before the internet, it would have been hard to get a bunch of people together to discuss their interest in pedophilia and child pornography; now it’s easy.

When we find others who think like us, it lends credibility to ideas and experiences that we otherwise might have doubted, or only temporarily considered.

In mainstream media, the focus has been mostly on extreme rightwing ideas, like the fact that half of Trump supporters apparently believe there is a pedophilic sex-trafficking operation run by top Democrats, or the fact that many Trump supporters believe the 2020 presidential election was rigged.

But the predominant focus on extreme rightwing ideas misses the point that these dynamics may be affecting many people across the board. If you’re a liberal, this means recognizing that many liberals have unreasonable and extreme views. For a few examples:

In some anti-cop and black-rights Facebook groups I’ve spent time in, there is widespread sharing of fake news, misleading framings of events, conspiracy theories, and hateful speech towards cops and white people (a few Facebook examples), just as there are such things in many pro-Trump Facebook groups.

From Facebook group called “Fuck the police.” Reactions included 10 hearts, 3 thumbs up, and 3 laughs.

A surprising number of people have “burn it down” answers to societal problems. In a study by Michael Bang Peterson and colleagues about destructive so-called “need for chaos” worldviews, 24% of people surveyed agreed that “society should be burned to the ground,” and 40% agreed that “we cannot fix the problems in our social institutions, we need to tear them down and start over.” Interestingly, the study authors related this to social media usage, saying that “a segment of the American electorate that was once peripheral is drawn to ‘chaos incitement’ and […] this segment has gained decisive influence through the rise of social media.”

There are many “burn it down”-type views these days, from both conservatives and liberals.

Such types of views were found in significant degree amongst both Trump and Bernie Sanders supporters. And in Jennifer McCoy’s research on political polarization, she found that Bernie Sanders supporters, as a group, had more Manichean, black-and-white thinking than did Trump supporters as a group (discussed in this podcast).

If we’re going to think seriously about the factors that may be amplifying our divides and increasing extreme ideas, we must be willing to examine how our own in-group favoritism may lead us to give a pass to unreasonable people in our group.

This does not mean believing that all political groups are equal in their degree of extreme thinking: this is unrelated to making such false equivalencies. (With the complexity of our world and our psychology, it’s hard to imagine how any effect would equally affect two very different groups.) It does mean being open to examining effects at an individual human level, even if you believe one side is worse. If our goal is understanding unreasonable group-versus-group animosity, part of that is realizing that we’re all individuals, that our political groups are not monolithic. All of us are capable of being misguided or deceived in various ways.

Factor #9: The internet speeds everything up

The internet can be seen as an accelerant of social interaction. Before the internet, it was hard to get content in front of people. Now it’s very easy. This ease of sharing has many obvious benefits but it also results in us being inundated by a deluge of information with no time to process it well.

It’s possible that the period of American history when broadcast TV was our dominant media (roughly the 1950s to the 1980s, pre-cable TV news) had a calming effect due to most people being influenced by only a few fairly monolithic and aligned sources of information. If TV was an opiate of the masses in that way, perhaps social media can be seen as an amphetamine of the masses, fracturing us with many competing worldviews, and amplifying our reactions.

The internet allows us to quickly align, react, and organize in a way that previous media did not: the Arab Spring protests and George Floyd protests were hastened and amplified by social media communications. That speediness can have positive aspects (ability to quickly share information, to organize), but it can also lead to over-reactions and instability, like a train going too fast around curves. As noted, violence caused by false information is increasingly common.

A post about fake news reports causing riots in Minneapolis.

And such over-reactions in turn have reactions. Violent behavior caused by extreme beliefs or misunderstandings will increase feelings of animosity in the opposite political party, or push more moderate people away from a group. (Here’s an interview about the impacts of violent protests/riots on voting behavior.)

The internet serves to wire us all together more and more tightly, and one result of this is that as a societal body, our nerves are more sensitive, more frayed. We’re all more on edge.

(Thoughts on other potential accelerants: people are spending more and more time online. And Covid has increased our time spent online while depriving us of the tempering influence of normal social interactions.)

Factor #10: Social media is perceived as an important battleground

Despite most social media content being made by a small percentage of people, social media seems to be often perceived as being an important arena of conflict, a place where important societal battles are being fought.

In a paper by Shannon McGregor about news coverage during the 2016 election, she examines how journalists cover social media. She explains, “I find that despite social media users not reflecting the electorate, the press reported online sentiments and trends as a form of public opinion that services the horserace narrative and complements survey polling and vox populi quotes” and that “journalists worried about an over-reliance on social media to inform coverage.”

Part of this is related to budgetary cuts to newspapers and journalism in general; it’s much easier and cheaper to cover social media than it is to do in-depth journalism.

Social media being covered as news could help explain why older Americans seem to be most polarized. If the younger generation is more accustomed to social media culture, perhaps they’ve become more accustomed to some of the aspects of social media discussed in this piece and are less agitated by it, while older Americans may see extreme or hateful social media posts and believe that the opposite political group has become more extreme and unreasonable than it actually is.

What can we do about this?

When reading this, you may have been thinking, “No matter how polarization happens or what social media’s role is, there’s still a lot to be angry about right now.”

And yes, there is. I am not arguing to not be angry, to not work towards political goals you see as important. But we can still attempt to achieve our goals while keeping in mind that how we interact with people, the language that we use, plays a role in either bringing us closer together or driving us farther apart.

And working to speak carefully and build bridges will likely do more to achieve our political goals than will any us-versus-them thinking. For one thing, consider how close the 2016 and 2020 American elections were. A little bit of good will and empathy can go a long way to nudging a few middle-of-the-road people over to your side. Inaccurate, polarizing language and good-versus-evil narratives seem unlikely to do so.

And to be clear: I am not saying you should behave like a saint and be nice to everyone online. The path to decreasing political polarization is building more bridges between the most reasonable and least extreme people. Remember that most of the population isn’t that extreme and wants to find ways to work together (for more on that, read the full version of this piece). If someone is extreme, unreasonable, and rude, go ahead and criticize them or mock them if you want (although ignoring them is probably better).

One reason I was motivated to write this piece is that it seems like the leaders and influencers I would have expected to make more attempts to build bridges and bring us together haven’t been doing a good job. In other countries that have fallen apart, there were probably many ordinary people, people like you and me, assuming that the people in charge would take care of things, that it wasn’t their responsibility, as a non-leader, as a regular citizen, to fix things.

But maybe the only way to avoid worst case scenarios is by all of us realizing that no one is coming to save us, that it’s up to all of us to consider the role we all play in these group dynamics. Because there are powerful forces at work, tribal forces deep inside of us, and maybe it will take an unusual shift in societal awareness to turn the tide against these oft-repeated self-destructive processes.

Here are some ideas for ways we can better use social media to help decrease unreasonable anger and polarization:

  • Speak carefully. Avoid speaking in inaccurate and all-encompassing “all Trump supporters are x” ways. Focus your criticisms and ire on the people you know are directly at fault (e.g., political leaders), and not the mass of citizens whose motives and beliefs are varied and hard to know. This will make your language more persuasive and less likely to cause anger.

If social media does work to fracture society, we’ll need more education about how that happens. A lot of the work in this area is focused on teaching people how to differentiate fake news from real news, or focused on specific features or algorithms.

But if it’s true that online communication tools have inherently polarizing effects, then we’ll need to broaden the scope of this education. We may need more “political literacy” in the form of educating people about how, despite perceptions, most people have much in common. And we may need more “psychological literacy” in the form of educating people about our social and emotional instincts and how those can lead us to some dark places.

In early 2021, I talked to a high school class about these topics. Their teacher recognized that these topics were especially important for children, who must navigate a complex and contentious environment that poses threats to their emotional well being.

I think we need more of this kind of teaching in schools. We need more people to recognize that these topics are important for society; we need more citizens to start thinking about these dynamics early.

Who am I?

I’m Zachary Elwood. I have a psychology/behavior-focused podcast called People Who Read People where I’ve recently been discussing political topics with various experts. I’ve also done some research into online deception and fake accounts; my research has been featured in the New York Times, The Washington Post, and Buzzfeed. And I’m a former professional poker player who’s written several books on poker tells (i.e., poker behavior).

Reminder that this is an abridged piece; full version here.

Thanks to Molly Elwood, Austin Schaefer, Daniel Steinberg, and Shawn Kilburn for comments and edits on this piece.

From Pew Research polling of citizens’ ideological views: one measure of our increased political polarization.

Have psych podcast “People Who Read People.” My research into online deception featured in NYT, WaPo, more. Wrote books on poker tells (translated 8 languages).