How do you deal with people who have political views that are very different from yours? Do you look for similarities or do you avoid each other? Scientific research on polarisation offers new starting points.

For months, emails, apps and Instagram messages were flying back and forth. Our nanny V. from New York came to visit us in Amsterdam, along with her newlywed husband, a friendly, slightly overweight policeman. They arrived at Schiphol with bags full of gifts. The exact right make-up trinkets for my daughters, a blue tie for my husband, and a carefully printed photo of my recently deceased mother in a picture frame. They even brought a plush New York yellow taxi with squeaky sounds for our dog. They loved everything about Amsterdam, had no specific to-do list, and were in high spirits. In short: the perfect guests.

At the age of seventeen, V. came to New York to earn money for her family, and six years later she joined us. At that time, I discussed everything with her: arguments with her mother in Peru, ‘trouble’ with boyfriends, medical questions. I considered her almost like a fourth child. So, you can imagine my joy when she sent me a photo of her American passport. Finally!

When Barack Obama was elected for a second term, there was no drama between us. We were both foreigners in New York, often making jokes about hysterical Americans, and our conversations were mostly limited to domestic matters.

However, in Amsterdam, at their instigation, we quickly ventured beyond all the usual topics. In New York, immigrants ‘feasted on luxury hotels’, they were given ‘credit cards for unrestricted spending’, all to become future voting cattle for the Democratic Party’. V. was even fiercer than her husband. Everything was blamed on out-of-touch Democrats. But what bothered them the most were the ‘ungrateful immigrants’. Because they were pampered while they themselves had to work hard. Also, ‘police officers were constantly blamed for everything, even though they risked their lives every day. But did you ever hear about that? No, Black Lives Matter is all anyone talks about. What about the lives of police officers?’

Within a few hours, we were entangled in a web of conspiracy theories, where immigrants were scapegoated. And this, while V. had just received her own passport. (This is known as the ‘drawbridge effect’: some immigrants, after naturalization, turn extra fiercely against allowing new immigrants.)

My husband and I did our utmost to steer the conversation away from politics, as it was clear that a meaningful discussion was impossible; there was no agreement on the truth. We don't even live in the same country, yet a political conversation between us was impossible. While V. and I might have had different opinions before, those differences never mattered to our relationship. Now, I wondered how we would get through the days.

Increasingly, people experience feelings of helplessness. Three-quarters of Dutch people believe that polarisation is increasing, according to a recent survey by the Social and Cultural Planning Office. They refer to ‘deteriorating manners and hardening in political and public debate’. They are also annoyed by expressions they perceive as ‘extreme or radical’.

I discuss this with cognitive psychologist Marte Otten, who is involved in polarisation research at the University of Amsterdam. ‘We know that people are not passive processors of information,’ she explains. ‘We have all sorts of hypotheses about what the world looks like and we immediately start working with them. If you make a prediction and test it against the outside world, there is a lot of room to shape that perception itself. We create all kinds of categories and we have ideas about the world; what we then perceive, we fit into those categories and ideas.’

That's not always wrong. But if assumptions and prejudices can influence perceptions themselves, then this opens the door to the creepy idea that people with different assumptions and prejudices actually see a different world. Otten wondered how this works exactly. ‘If humans are a kind of predictive model, where each perception leads to a prediction of what comes next, then it is logical that biases play a role. But I wanted to know if perception itself is already influenced by social knowledge such as biases and stereotypes. If our perception of reality itself is different, then it seems that the end is near. Scary, but also very interesting to investigate,’ says Otten.

A famous experiment is the so-called gun/tool task where participants are shown pictures of either a tool or a weapon. Just before they view the pictures, they briefly see a white or black face. Participants are told that these faces have nothing to do with identifying the object. Yet time and again, it turns out that people who have just seen a black face detect a weapon faster and more often misclassify the tool as a weapon.

‘We have replicated this test to investigate whether participants based on stereotypes really see reality differently. With an unconscious association between black men and violence, do you indeed see the weapons more often?’ When Otten talks about ‘seeing’, she means this literally. ‘The first step is that information enters the retina to then be forwarded to the visual systems in the brain where the visual experience arises.’

Determining when the information enters consciousness is done using so-called continuous flash suppression. ‘During the test, we expose participants on one half of their visual field to the picture while showing a flickering abstract Mondrian-like pattern on the other half. In such a setup, people are only aware of the flickering Mondrian. They do not see the weapon or tool, although that object does land on the retina. Then we slowly turn off the flickering Mondrian. This creates a transition point where people become aware of the tool or weapon. This way, we can precisely measure how easily or difficult certain information enters our consciousness: if we only need to turn the flickering Mondrian away a little, it indicates that our consciousness is sufficient with subtle visual input to come to an experience. If we first have to make the flickering Mondrian completely invisible before people see the object, then it seems that a large dose of visual information is needed.’

In this way, Otten tests whether stereotypes about skin colour and weapons affect visual processing. If there were such an influence, pictures of weapons would enter consciousness faster in participants after seeing a black face than after seeing a white face. But that was not the case: the faces did not influence the speed at which people became aware of the objects. ‘Fortunately! A great reassurance,’ laughs Otten.

At first glance, it may seem unimportant when exactly an image enters consciousness, because in the end prejudice remains. But in practice, it can make a difference. ‘A few seconds can make all the difference.’ Otten explains that in another version of this experiment in the US, police officers extensively trained in correct use of firearms did not tend to identify innocent objects as weapons. Their choices to shoot or not in a computer simulation were driven by what they really saw, not by what they thought they saw. ‘That's why it's important to know this. Because if we share a perceptual reality together, then you can still do something about it. Then there is still room for this kind of intervention.’

In common parlance, we use words like polarisation and segregation interchangeably to denote a multitude of undesirable situations, while in the extensive scientific literature, efforts are made to distinguish between the concepts and their gradations. Usually, five different forms of polarisation are identified. Paul Dekker, professor of Civil Society at Tilburg University, describes an increasingly popular classification in the anthology Political Polarisation in the Netherlands.

Political polarisation starts with ‘issue polarisation’: with ‘stronger opposing views’, while in the stronger case of alignment, these oppositions begin to ‘broaden’ into an ‘ideological polarisation’, where different topics are connected. Recently, this is often referred to as package politics. If you are concerned about the climate, you probably also want to welcome asylum seekers and impose a higher wealth tax. And if you are concerned about asylum seekers, you also want to be able to eat a meatball and the climate is not that urgent. When opposing views also ‘coincide with social, economic, and cultural differences’ and people ‘come together spatially and organizationally’, we are talking about ‘the societal anchoring of political polarisation’.

This is the situation in the United States. In the case of ‘affective polarisation’, emotions towards each other become even stronger. ‘People become more positive about people with the same views and reject people with different views more strongly, emotionally and also morally’. In the extreme case, differences extend so far that people also believe different things are true and real: ‘facts polarisation’. This is exactly where we found ourselves with V. and her policeman.

Our initial response consisted of asking many questions and looking for commonalities. They told us about the great danger police officers in New York are exposed to. His mother no longer dares to take the subway to work because crime has increased significantly in recent years. His family grew up in Manhattan, but they can no longer afford to live there. Too expensive. Every day, he has to travel a long distance to work. And because the Mayor of New York wants to make the city bike-friendly (‘another leftist hobby!’), parking costs have skyrocketed, the city is gridlocked, and he can no longer use his car. While the subway is dangerous. We nodded sympathetically. Still, we wondered if they were blaming the right people for their problems. Because what did immigrants have to do with this? And that story about pampered migrants in luxury hotels with unlimited credit cards: was this really true? ‘Some say this is a conspiracy theory,’ V. admitted.

I happily realized how good it was to ask questions. And fortunately, we are outsiders, who foster a certain openness. In America itself, many Democrats and Republicans simply no longer talk to each other, at least not about politics. In Europe, however, this is all still quite manageable, I thought, keeping in mind the German sociologist Steffen Mau. In his recent book “Trigger Points: Consensus and Conflict in Contemporary Society”, Mau writes that most Europeans positioned themselves in the political centre thirty years ago, and that remains the case today. It is mainly the politicians who, as ‘polarisation entrepreneurs’, introduce and shape division. They exploit polarisation dynamics through emotional trigger points. Following the European elections, the political centre has indeed held its ground, but the differences on both the right and left have clearly increased. Is this all the work of politicians who cleverly use emotional trigger points?

As so often in science, it depends on which studies you consider. An important question in the polarisation debate is to what extent different groups still encounter each other. How precisely defined are our bubbles? The worst example is American society. As early as 2004, American journalist Bill Bishop described ‘the big sort’, also the title of the book he co-authored with sociologist and statistician Robert Cushing four years later. Bishop demonstrated how Americans had divided themselves into separate geographical, economic, and political communities over the preceding three decades. He already pointed out the significant danger posed by this ‘big sort’. Because the more we live in our own bubbles, the greater the risk of (dangerous) polarisation. Unfamiliarity often breeds contempt. The question is: to what extent the Netherlands is following in the footsteps of the United States?

Segregation has traditionally been measured at the neighbourhood level, where residential location is the determining factor for measurement. You look at a neighbourhood and measure where people with different characteristics (education, income, wealth, political preferences, cultural background) live. Subsequently, you conclude that if they live separately from each other, they are also unlikely to encounter each other frequently. In this case, there is segregation, which subsequently increases the risk of polarisation. However, the question remains how much information neighbourhood research precisely provides about segregation. After all, people can live in a very mixed neighbourhood but then commute every day to their not-so-mixed workplace and in their private lives interact only with their not-so-mixed friends and family. Therefore, the social network approach is a good complement to such geographical studies.

Yuliia Kazmina is a Computational Social Science PhD candidate at the University of Amsterdam and conducted such research in the Netherlands, mapping socio-economic segregation (based on income). The social network approach is different because it does not measure segregation in a geographical area but in ‘networks of social opportunities’. In short, it measures segregation based on social capital. Everyone has relationships: with people in the household, neighbours, family members, colleagues, or classmates. These relationships determine (among other things) the individual's social network. For example, if in your family and school there are many people who earn well, you are more likely to have someone in your network who can help you find a well-paid job at some point.

The CBS knows who your neighbours are, who you work with, which school you attend, who you live with, who your family members are, and how much you earn. So, this data isn't new, but the way researchers study it is. Eelke Heemskerk, Professor of Social Sciences at the University of Amsterdam, supervisor of Kazmina, and co-author of the study, explains: ‘We collaborate with the CBS and Leiden University under the name POPNET. Normally, due to privacy concerns, researchers cannot just access CBS data. Rightfully so. Therefore, we temporarily housed our own facilities and researchers at the CBS. The CBS has a highly secure environment, allowing us to safeguard citizens’ privacy. All data is anonymized.’ The study mapped the social network of 17.2 million Dutch people. ‘If all those individuals have relationships on five levels, you're talking about 1.3 billion connections. You need a computer with a lot of processing power for that,’ Kazmina knows.

The results are striking. ‘We found twice as much segregation with this model as in normal research based on geographical data. Moreover, we found a higher degree of segregation in larger cities, whereas it is often assumed that there is less segregation in cities due to greater socio-economic mixing in neighbourhoods’, says Kazmina. But in a large city, geographical segregation may be less pronounced, but segregation of social networks may be greater. The fact that you encounter people from a different socio-economic group every day does not necessarily reflect the composition of your individual social network. ‘Such studies can therefore be a good complement to other forms of research because they can demonstrate “hidden segregation” beyond the neighbourhood,’ says Heemskerk.

Among others, due to these new studies, Han van der Maas, Professor of Psychological Methodology at the University of Amsterdam, specialized in polarisation, doubts the conclusions of Steffen Mau about politicians as ‘polarisation entrepreneurs’. ‘Mau is, of course, a sociologist, so this is a different approach in psychology, but compared to the research he bases himself on, countless others show that polarisation in society is indeed increasing’. In this case, the concerns of three-quarters of Dutch people about polarisation would thus not only be fuelled by politicians but would also stem from society itself.

‘What he says is partly true. Often, politicians are very similar in terms of voting behaviour, but during elections, they still have to distinguish themselves to win votes. You see the same in journalism. If you keep saying all the time that the truth lies in the middle, then no one clicks on your pieces.’ Tom van der Meer described in his recent book “Valueless Politics” how important it is for politicians to clearly explain the political choices citizens have and on which values they are based. A healthy form of polarisation is essential for this.

Han van der Maas notes that there is a lot of scientific research on polarisation. ‘There is no university without a polarisation lab or social cohesion hub nowadays. And so many different disciplines are involved. Even the experts lose track.’ Typically, such research deals with societal data, such as that from the CBS, or through conducting surveys. ‘Often it’s not very theoretical: just collect data and see if you find something out, or discover that X correlates strongly with Y. It's not a bad method in itself. But the topic is important enough to bet on different horses.’

Van der Maas’ horse is the so-called cascade model of attitude change, which considers polarisation as a complex system. In the cascade model, he uses fundamental principles from statistical physics to look at both the interactions that occur within the individual itself and the interactions that occur simultaneously between different individuals. ‘If you mathematically combine these two network representations of individual and social networks, you get a network of networks: each person has their own inner network and is also a node within a social network.’

This quickly becomes very complicated. But with this model, which combines two different levels, it can be explained why attitudes are sometimes very difficult to influence and how they sometimes become unstable and can therefore change quickly. ‘The model offers a new explanation for polarisation and options to escape from this.’

If a few assumptions are made, polarisation in society can be modelled according to principles of statistical physics in a so-called ‘agent-based model’ for two dimensions, one of the individual and the other one of the group. One of those assumptions is that an opinion stemming from the internal, individual network becomes a continuous variable in the social network, depending on the information field in which you find yourself. Internally, there may be many reservations in forming your opinion, but externally, you ultimately vote for or against something. Another assumption is that polarisation in a social network arises from lack of contact. If there is contact, there is a possible gateway.

‘A key role here is the concept of “bounded confidence”,’ explains Van der Maas. ‘The idea is that people who differ too much in opinion will refuse interaction, preventing any approach. It has been extensively shown that this can lead to polarisation in two or more groups.’ Bounded confidence is therefore built into the model. ‘Suppose the opinions resulting from all those individual internal networks can have two flavours: conservative and progressive. If this model is then “run” (on a computer), it indeed shows that progressives and conservatives no longer interact with each other and polarisation occurs.’ They move further apart. Immediately arises the question of what exactly can be done with this knowledge. Does it provide a basis to combat dangerous polarisation if enforcing contact is practically difficult?

Indeed, the cascade model suggests several exits. The most important one is counterintuitive. Van der Maas: ‘It is possible to temporarily have a person with a progressive opinion adopt a conservative opinion in the model. A disturbance of the opinion towards conservative can be stable for some time. However, we see that after some time, the opinion bounces back to the more stable progressive value. In the temporarily conservative state, however, this person suddenly becomes a conversational partner for the “real” conservatives. The opinions have moved closer together, and there is sufficient trust, enabling influence. The information that the “transformed conservative” subsequently exchanges in their own circle is then progressive in nature. And if this person also pays more attention to the subject, the real conservative may move significantly towards progressive. Of course, the same applies if the roles are reversed.’

It reminds me of the A12 in the summer of 2023, when some lawyers from the Zuidas came to protest against fossil subsidies. Suddenly, that seemed socially acceptable for a moment. Would activists be wise to put on a suit and temporarily embrace a conservative opinion? Van der Maas laughs. It's not that simple, of course. ‘But these are interesting outcomes. In the next five to ten years, I expect that we will gain deeper insights into how these processes interact exactly in these models. Hopefully, we can then also translate this into interventions.’

How do you create movement in the bubbles? Scientific research into polarisation offers new starting points, especially when viewed through a lens of complexity. Yet, in a polarised society, it remains extremely difficult to deal with friends whose political views you find unpalatable. In our case, we kept searching for common ground with V. and her husband. We talked a lot about dogs; the plush taxi was a big hit.

Published in De Groene Amsterdammer on July 10, 2024. Translated by the Institute for Advanced Study.