Humans tend to accept beliefs that are coherent with their other beliefs and reject those that do not fit with them.

As a result of this, different people might arrive at different belief structures and the consequences of this on societal belief patterns are not known. In this chapter we show one way to induce possible ‘coherence structures’ for groups of people from survey data and then set them going, influencing each other, within a computer simulation of individual agents.

Of course, people do not perfectly adhere to coherence strategy but are amenable to all sorts of other influences. In the simulation, these other influences are represented as randomness that affects whether to accept or reject beliefs.

The simulation shows that the degree to which agents act randomly (rather than based on coherence) is critical to the outcomes. Low levels of randomness (i.e., strong strive for coherence) result in agents converging on more extreme beliefs. With high levels of randomness (i.e., weak striving for coherence) the simulated agents converge on moderate beliefs.

It is with some intermediate adherence to coherence that the resulting spread of beliefs looks more like what we observe in the real world—spread out or polarised in some beliefs but converged in others. The approach is illustrated using data from the European Social Survey about attitudes in five dimensions.

B. Edmonds, P. Steiglechner, J. Lorenz, F. Kalvas, M.C.L. Batzke, Understanding Mutual Social Influence When People Prefer Coherent Beliefs, in: M.A. Keijzer, J. Lorenz, M. Bojanowski (eds. ) Computational Social Science of Social Cohesion and Polarization, Computational Social Sciences, Springer, Cham, (2026) 207-223.