Ferrari, S., Lammers, W., Wenmackers, S. (forthcoming) How the structure of scientific communities could impact the public uptake of uncertain science. Philosophy of Science.

The present model is an abstract ABM designed for theoretical exploration and hypotheses generation. Its main aim is to explore the relationship between disagreement over the diagnostic value of evidence and the formation of polarization in scientific communities.
The model represents a scientific community in which scientists aim to determine whether hypothesis H is true, and we assume that agents are in a world in which H is indeed true. To this end, scientists perform experiments, interpret data and exchange their views on how diagnostic of H the obtained evidence is. Based on how the scientists conduct the inquiry, the community may reach a correct consensus (i.e. a situation in which every scientist agrees that H is correct) or not.

We consider scientific communities where each scientist employs one of two characteristic methods: an “adequate” method (A) and a “superior” method (S). The quality of methodology is relevant to the epistemic products of these scientists, and generate credit for their users. Higher-credit methods tend to be imitated, allowing to explore whether communities will adopt one method or the other. We use the model to examine the effects of (1) bias for existing methods, (2) competence to assess relative value of competing methods, and (3) two forms of interdisciplinarity: (a) the tendency for members of a scientific community to receive meaningful credit assignment from those outside their community, and (b) the tendency to consider new methods used outside their community. The model can be used to show how interdisciplinarity can overcome the effects of bias and incompetence for the spread of superior methods.

The model that simulates the dynamic creation and maintenance of knowledge-based formations such as communities of scientists, fashion movements, and subcultures. The model’s environment is a spatial one, representing not geographical space, but a “knowledge space” in which each point is a different collection of knowledge elements. Agents moving through this space represent people’s differing and changing knowledge and beliefs. The agents have only very simple behaviors: If they are “lonely,” that is, far from a local concentration of agents, they move toward the crowd; if they are crowded, they move away.

Running the model shows that the initial uniform random distribution of agents separates into “clumps,” in which some agents are central and others are distributed around them. The central agents are crowded, and so move. In doing so, they shift the centroid of the clump slightly and may make other agents either crowded or lonely, and they too will move. Thus, the clump of agents, although remaining together for long durations (as measured in time steps), drifts across the view. Lonely agents move toward the clump, sometimes joining it and sometimes continuing to trail behind it. The clumps never merge.

The model is written in NetLogo (v6). It is used as a demonstration of agent-based modelling in Gilbert, N. (2008) Agent-Based Models (Quantitative Applications in the Social Sciences). Sage Publications, Inc. and described in detail in Gilbert, N. (2007) “A generic model of collectivities,” Cybernetics and Systems. European Meeting on Cybernetic Science and Systems Research, 38(7), pp. 695–706.