The third perspective emerges from physicists and complex systems research communities. They explore social phenomena from the perspective familiar to them: systems and interactions within the systems. It argues that societies consist of complex individuals, thus bearing similarity to phenomena more familiar to physicists: atoms that create a larger system of matter. The idea is that, similar to studying atoms and how they establish larger systems, researchers can study individuals and how they establish larger communities and what kinds of patterns emerge in them. Buchanan (2008) uses the metaphor of the social atom to illustrate this kind of approach. Individuals, like atoms, follow simple rules, but emerging from those rules are complex patterns. Methodologically, this perspective is a tightly connected simulation model (e.g., Conte et al., 2012). Agent-based simulations are methodological approaches where researchers can explore the implications of different parameters â both at an individual level and system level â and explore phenomena that emerge through running a simulation. Thus, Epstein and Axtell (1996) argue that to explain a phenomenon, or to build a model, you need to be able to simulate the phenomena â to grow it in an artificial society.
The early work by Axelrod (1980a,c) focused on the Prisoner's dilemma, a commonly used tool to reflect world politics and economics. In the Prisoner's dilemma, two players need to choose each turn if they defect or if they collaborate. There is a modest reward if both collaborate, a high reward if only one is defecting and no reward if both players choose to defect. These papers report a tournament to understand how to win the Prisoner's dilemma game. The tournament was run using computers, where each competitor wrote their own strategy in a format that the computer could understand. Each competitorâs submission was then run against other submissions to determine which model was most successful. After the tournament, the winning strategy was tit-for-tat, a simple approach where one defected only in cases where the opposite side had already defected in the previous turn. Beyond these findings, they demonstrate how through simulating different situations, an optimal strategy could be found to the challenging problem and, thus, share a light to the various fields using the Prisoner's dilemma.
The simulation models have been used beyond abstract and theoretic models as well. Nowak et al. (1990) sought to use an agent-based simulation to study polarisation processes. They first argue that, overall, individual actors following simple rules may be used to grow larger, social-level phenomena. This emergence, however, then influences back to the individual actors. As they put it, `Laws operating on lower levels of social reality may have unforeseen, seemingly emergent, consequences for higher levels, which, in turn, will affect the social environment facing lower level units' (Nowak et al., 1990, 362). They then draw some social science frameworks to understand how attitude change may occur and simulate the impact of those rules in an artificial society. Their results indicate that groups polarise: `As the simulation went on, the frequency of attitude changes decreased, because individuals within larger subgroups were less likely to change their attitudes. After some number of simulation steps, the modelled society reached a state of equilibrium in which there were no further changes' (Nowak et al., 1990, 368). The benefits of their approach to study polarisation include the clear theoretic link. Individuals' actions are based on theory-driven models and approaches, thus allowing discussion of what seems to cause the polarisation tendency. Beyond that, as they have full control over parameters, they can explore the parameter space and expand understanding of conditions where polarisation tends to occur in groups. For example, they report that a minority opinion with 10% mind share initially was either totally erased or strongly reduced in the simulations, but with 20% mind share, total erosion did not occur. This type of analysis would be highly difficult in observational studies or even in laboratory setups.
Again, this perspective provides insights into the potential of computing technologies to reconsider how social research can be conducted. The research has also evolved beyond abstract simulation models thanks to greater access to data (e.g., Eagle and Pentland, 2006). Pentland (2014) argues that, together with the idea of modelling human behaviour and increased data access, artificial societies may evolve to socio-physics: studying social behaviour as it was controlled through general laws.check for a good quote? Unlike the previous two perspectives, this perspective highlights not methods nor data but a paradigm of conducting social science research. The description of individuals as social atoms may be suitable for those arriving at social science questions from physics and related disciplines, but social scientists may struggle with such conceptualisations â at least I do. Furthermore, these ideas are deeply rooted with the idea that social science research should be revolutionised through a more systematic study of societies through methods and paradigms familiar to physicists (e.g., Conte et al., 2012). Thus, from this perspective, computational social science seeks to be a new research field, trying to somewhat distance itself from traditional social science approaches. Rather, it seems to adapt methods familiar to physicists and complexity scientists to social science research problems.