How did you come to be studying violence?
In the 1980s, I trained in political economy, the central tenet of which is rationality - the idea that you can understand human behaviour through a cool calculation of costs and benefits. I found that ideas valid for a lot of economics, but as a political scientist, I found it very limiting, I wanted to work on a subject that involved more emotions than those generated by the question of prices. When I finished my PhD and started teaching at Harvard, Hindu-Muslim violence was once again breaking out in many parts of India. And research into this kind of violence left a lot to be desired. So, in the summer of 1990, I went off to Kashmir, which was in the middle of a civil war, and I could see very clearly the role of emotion in politics.

How did it change your intellectual outlook?
The emotional intensity I encountered in the Kashmir valley not only overwhelmed me, but also aroused deep intellectual curiosity. Until that time, my fieldwork had dealt with the politics of economic policy. Compared with that experience, the loss, pain and suffering I witnessed in the Kashmir valley appeared to belong to an entirely different realm of life, a realm not easily amenable to standard political-economic enquiry. I saw children who had lost their limbs, families that had lost their children, and yet people were willing to fight a war of insurgency. Their inner selves, they argued, were at stake, and they were prepared to suffer a lot of pain for dignity, identity and self-respect.

The intellectual challenge became obvious: can we create a science of pain, suffering, loss and death experienced in ethnic or communal conflict? In a way, this subject is close to my heart. I want to come up with some practical ideas about peace. There is far too much violence in the world. I am also a bit of a rebel and I find it extremely hard to accept academic orthodoxy. Intellectually, if you give me an inch, I will take a mile.

Are you saying that when it came to studying human behaviour, rationality was not a sufficient basis?
Let me put it this way: you cannot base your research into ethnic politics on a cool calculation of costs and benefits, which is the assumption a lot of people made through the 1990s. It is not irrelevant, but neither is it central. I am not saying that ethnic politics is entirely irrational, but emotions play a much greater role than people have assumed.

Have you ever witnessed violence yourself or been close to it?
Yes. When I was 13 there was a very big riot in the town of Aligarh, where we were living. It didn’t affect me in a profound way at the time because I didn’t see anyone killed in front of my eyes. I saw buses being burnt and the police everywhere. We couldn’t go to school for two weeks, so we played cricket instead, which was great. There were people being killed but I was kept away from it.

The thing I remember about Aligarh was that it was a very segregated town. Before we moved there I had grown up with Muslims for much of my childhood, but in Aligarh, I realised for the first time in my life that I had no Muslim friends and that the Muslim children were not part of my cricket team. I remember other things such as seeing a Hindu nationalist being very enraged and driving off saying he had to teach Muslims a lesson. Things like that left a big impression on me. I didn’t come to study violence because of something in my childhood, but when I started working on it these childhood memories did start to come back.

So what is the key to predicting which communities will turn violent and which will remain peaceful in times of ethnic unrest?
It comes down to how the cities or villages are structured, and the networks that people form across religious or ethnic divides. In India, I have identified two types of civic network, which I call the associational and the everyday. The everyday type covers things such as Hindu and Muslim children playing together and their families and friends visiting each other or eating with each other, or taking part in festivals together. The associational type involves the two groups being members of the same trade unions, sports clubs, political parties or business organisations. Associational structures go beyond neighbourhood warmth, and in times of unrest they are much more robust. They can be a serious constraint on the polarising strategies of political elites. Places with strong networks of this kind are very likely to remain peaceful.

Why are trade unions, political parties or business associations more likely to result in a stable community?
Because they bring together a solid common interest or ideology, rather than being simply social. They are more than just a way for people to entertain themselves or understand each other. Everyday interactions can easily be ruptured by criminals and gangs, but associations represent shared interests, and there is more incentive to preserve them.

It is partly to do with self-interest, but it is more than that. Sometimes it has to do with a shared ideology. When a trade union or a political party fights for its cause, they are not always fighting for electoral dividends; they are fighting for a vision of the kind of India they want to build. People will go to great lengths, and will sometimes sacrifice their own lives, to protect a certain vision that they have built for their organisation or their society. Those shared ideologies are important enough for the people to fight to preserve them whatever ethnic group they belong to. And that is what keeps the peace.

Some researchers have suggested that a mixed community is more likely to stay peaceful if the people from the different ethnic groups know each other well. Does your work back this up?
Yes, but it is not the whole truth, because sometimes it goes the other way. Although I have not properly tested that hypothesis, I found that in the three most unstable, highly segregated towns I studied in India, the children have such awful views about the other communities. Muslims boys in those towns never interacted with Hindu boys and they believed the worst about each other. But I also found that interaction itself can lead to enmity. For example, where there is intermarriage between Sikhs and Hindus or Muslims and Christians some people see it as threatening the ethnic or religious purity of the community and have hit back. That would be a very interesting question to ask: under what conditions does knowing each other actually promote bitterness?

How can strong community networks prevent politicians from exploiting ethnic differences to further their own ends, in the way Slobodan Milosevic did in the Balkans?
There is no doubt that large-scale ethnic violence cannot take place unless politicians protect criminal gangs. The gangs need political support so that the police don’t arrest them. When a politician finds not only that the neighbhourhoods are integrated but also that whole organisations are, he knows it is a lost cause to try to exploit the differences. Violence is always sparked by something, and the key with integrated organisations is that they prevent the spark from becoming a fire. They work together to kill rumours, cooperate with the state, or identify those making mischief. It is in their interests to maintain the peace because they will be hurt by violence. A Slobodan Milosevic may in the end succeed in breaking such integrated networks, but it is a lot easier to create violence if communities are segregated. The odds that normal politicians - and Milosevic was not one - can do so are low.

Could you persuade a segregated city to adopt these civic structures as a way of keeping the peace?
Yes, you could. I am engaged in a project to reconstruct Aligarh. We are trying to improve interaction between the Hindus and Muslims there. Cities can be turned around. Bhiwandi, a previously riot-prone town just outside Bombay, was turned around in the late 1980s, and it remained peaceful even during the awful Hindu-Muslim violence that took placed in India in the early 1990s. Building integrated neighbourhood-level organisations was the key. You need the support of the government and the police, but essentially it is a job that civilians have to do. This would be harder after a civil war, when you need the state to come in a big way.

Communal violence is often triggered by an apparently religious act - throwing a pig into a mosque, for example, or the killing of a cow. How much violence is genuinely religion-inspired?

The symbolism is often religious, but to call it religious violence is usually to oversimplify it. Religion can also be experienced as culture - I have a Hindu name but I am not a practicing Hindu. The religious nature of communal problems is often exploited by people who are secular. It is the same in Northern Ireland - not all Catholics and Protestants engaged in the struggle are really believers.

Is this kind of violence linked with poverty? Are richer places less prone to violence?
After a certain income threshold, the risk of riots does diminish. You’ll still get community activists who want to fight for their cause, but this does not necessarily lead to violence. You could see this all over Europe as incomes rose after the Second World War. The recent riots I have studied in Indonesia, Nigeria and India look similar to the riots that took place in the US around the world wars: St Louis 1917 to 1918, Detroit 1938 to 1939, Harlem in the early 1940s. You don’t get that kind of group rioting in the US today. But in affluent places, you still get hate crimes. Income does not seem to affect these.

Why is group violence predominantly an urban phenomenon?
In a village, everyone knows everybody else. People have personal knowledge of each other, which reduces the likelihood of mass violence. But you do have violence of other kinds - for example, violence between castes.