Rationality and Group Behavior
Project: Research › Individual research project
Research areas and keywords
UKÄ subject classification
- Epistemology, Group Behavior
When in groups, people often behave in seemingly irrational ways that lead to highly undesirable collective outcomes. People sometimes blindly follow others. The bystander effect (Latané and Darley, 1969)¬, for example, shows that people in large groups fail to help the victim of an accident that they would otherwise help. Social psychologists have categorized such phenomena and provided explanatory clues for many of them. In a large number of cases, however, the question remains open as to whether these phenomena show that humans are truly irrational.
These phenomena represent real challenges and open fields for investigation for Epistemology, i.e. the study of rationality. Three relevant questions for epistemologists are:
1) Do these phenomena show that people are irrational?
2) What are the factors that trigger them at the group level?
3) How, if at all, can they be prevented or countered?
The present project addresses these general questions for three relevant types of group behavior. Its aim is to investigate them with the main tools available in formal epistemology: logic, probability theory, game theory, and argumentation theory. The ultimate goal of such a study is to get analytical results and develop hypotheses that are tested via computer simulations.
a) Pluralistic ignorance and bystander effects
Pluralistic ignorance is the name for the situation in which everyone believes something but also believes that the others do not believe it (Katz and Allport 1931). This phenomenon pops up very frequently in real life (Prentice and Miller 1993). Pluralistic ignorance can explain bystander effects (Latané and Darley 1968, Latané and Nida 1981). As Proietti and Olsson (2014) show, pluralistic ignorance can even arise among artificial agents that reason according to all norms of logical rationality. This project aims at (i) a detailed analysis of the varieties of pluralistic ignorance and (ii) explaining the mechanisms by which pluralistic ignorance generates bystander effects and similar phenomena.
b) Group polarization
A second area of investigation concerns group polarization, also known as “risky shift” (Stoner, 1961). This is the phenomenon where members of a group polarize with respect to their opinions on a topic after discussing it (Isenberg, 1986). Understanding polarization is an important issue, especially in the era of social media. Indeed, virtual online discussion and political debate (Yardi and Boyd 2010; Sunstein 2003) seem to witness a more pronounced tendency for groups to polarize than actual face to face discussions. In general, polarization speaks against the assumption that debate among informed people should lead to consensus. I intend to investigate this issue with the formal tools from argumentation frameworks introduced by Dung (1995). This part of the project aims at (i) understanding what it is for an agent to rationally update her argumentative framework in a debate and (ii) whether polarization is possible under rational updates.
c) Rationality and social efficiency
According to J. Elster (1989) social norms that glue societies together have two fundamental characteristics. They are “(a) shared by other people and (b) partly sustained by the approval or disapproval of others” (Elster, 1989). Scholars in philosophy and economics are prone to give a game-theoretic explanation of the notion (Bicchieri 2010; Binmore 2010; Gintis 2010). For all these approaches an important requisite for a norm to be considered fair is that it results in a state where no one can get “more” without someone else getting “less”. Actions aimed towards such a state are considered pro-social rational behavior. The key question is whether this behavior, if followed by everybody, always results in social well-being. Indeed, there are cases where pro-social behavior within a community seems, to the contrary, to cause inefficiencies, low quality, and detrimental collective outcomes (Origgi and Gambetta, 2013). In this part of the project I will build a computer simulation based on a society of artificial agents which pro-socially play together a simple game. The main aim is to (i) test Origgi and Gambetta’s conclusion over many possible settings and (ii) find out specific external policies to improve societal efficiency.
Theory and method
A dominant portion of research in social psychology is conducted via lab experiments on control groups of actual humans. This research is increasingly complemented by formal investigations, eminently by computer simulations over societies of artificial agents. There are at least three benefits of such an artificial approach. First, artificial agents can be programmed such that only the relevant hypotheses are tested. For example an artificial agent can be programmed such that it only focuses on the reasoning tasks, leaving out all emotions. Although limiting the psychological granularity of agents, this allows to test experimental hypotheses in a more controlled setting. Second, the situation in which an artificial agent is put is always replicable. This allows for computer simulations and makes the study immune to confirmation-biases and other disturbing factors that fuel the actual replicability crisis in social psychology (Simmons et al., 2011). A third positive benefit lies in the possibility of varying more parameters than it is possible in a lab setting, e.g. the group structure, size and time span (Mason et al., 2007). Formal analysis via models may therefore help to test and aid explanations of hypotheses that were provided by psychological experiments.
To determine whether certain group behavior is irrational, a norm of rationality needs to be explicated. Different approaches focus on different features of rational behavior. Bayesian epistemology takes rationality to be a correspondence between the strength of your evidence and the strength of your belief. Logical approaches see rationality as the ability to make correct inferences. Game and decision theory interpret rationality as maximization of utility. This project uses all these notions of rationality, since adopting a single definition of rationality undermines the understanding and analysis of phenomena of different kinds. In this research project, several tools of analysis are employed. Indeed, since the phenomena of interest challenge rationality in different ways they require different approaches of scrutiny, including mixed ones.
|Effective start/end date||2017/01/01 → 2019/12/31|
Related research output
Research output: Contribution to journal › Article
Research output: Chapter in Book/Report/Conference proceeding › Paper in conference proceeding