Dyad and Agent Modelling
Mr. Martin Schoemann
Prof. Stefan Scherbaum
Interpersonal psychological factors are - in addition to intrapersonal psychological factors - crucial for the development of mental health problems. Testing underlying factors of mental health problems for causality is difficult because they can hardly be manipulated experimentally. As a possible solution, we propose to use one major, in other research fields already approved cognitive modeling approach, called agent-based modeling (ABM). To illustrate the application of this approach in psychological research, especially clinical psychology, we develop a model based on the underlying factors of grandiose narcissism as personality trait and implement it as ABM. This way, we examine the characteristics that are crucial for the development of grandiose narcissism, focusing on the role of social interactions. To validate our findings, we evaluate the simulated data using a Social-Network-Analysis (SNA) approach and compare it with results of SNA from real data. After examining underlying factors, it is possible to derive possible interventions that can be tested using ABM. Our approach provides a promising example for applying ABM in psychological research, especially when examining interpersonal factors of mental health problems.
This is an in-person presentation on July 21, 2023 (09:00 ~ 09:20 UTC).
Prof. Michel Grabisch
This paper investigates the impact of different mental models on the formation of belief clusters and the spread of opinions within social networks, using an agent-based model. Every behaviour can be grouped into two categories: fundamental and emergent behaviours. A fundamental behaviour is a property of an agent that is independent of other properties or agents, while emergent behaviours are caused by the interactions of fundamental or other emergent behaviours. This paper studies fundamental behaviours to understand which ones are better to explain the emergent behaviours that are empirically observed. In our simulations, agents interact within a randomly generated network and observe signals about whether climate change is real or is a conspiracy. Then, they use various mental models to update their beliefs, where mental models are defined as functions which agents employ to update their beliefs in response to new information. We create three mental models: a Bayesian updating rule which rational agents use, a heuristic-based updating rule which boundedly rational agents use, and a Dirac-Von Neumann rule which agents who can be irrational use. Furthermore, we use state-of-the-art random networks to understand how community structures respond to these mental models, how clusters are formed to represent conspiracy theory groups, and in which models conspiracy theory groups can grow by convincing rational agents. We aim to reveal which belief update methodologies tend to create more realistic clusters in which polarizations arise due to agents being resistant to novel information. Moreover, we explore if rational agents can end up believing in conspiracy theories under any community structure. This paper is the first one to study multiple belief update methodologies in social networks using agent-based modelling, allowing us to reveal novel behaviours of belief diffusion.
This is an in-person presentation on July 21, 2023 (09:20 ~ 09:40 UTC).
Dr. Sven Banisch
Capturing the complexity of interpersonal dynamics – emerging from the approach and avoidance motives of two individuals in dyadic interplay that unfolds simultaneously on multiple time scales in order to satisfy their psychological needs – remains a scientific challenge. In line with calls for embracing complexity in psychological research using formal modeling, the purpose of this mathematical study is to investigate the underlying mechanisms of the formation and maintenance of affiliative interpersonal relationships using evolutionary game theory. After formalizing interpersonal situations based on the affiliative motives of their interactants, a relational state space is constructed that reflects the ways of relating available to the interactants in the momentary state of their interpersonal relationship. This allows for modeling the evolution of an interpersonal relationship as a trajectory – driven by positive and negative reinforcement – in the relational space. Depending on the motives of both interactants, three qualitatively different interpersonal dynamics emerge: (1) global stability with only one relational attractor (e.g., an interpersonal relationship of pure friendliness in the long run), (2) bistability with two mutually exclusive relational attractors (e.g., either pure friendliness or pure distance), and (3) cyclicality with periodic orbits in the relational space (e.g., oscillation between friendliness and distance). Grounded in empirically supported psychological constructs, the formal model generates the well-known pattern of interpersonal complementarity. Over and above, novel interpersonal patterns emerged that might point to some underlying mechanisms of the interpersonal maintenance of psychopathology. The model limitations as well as avenues for empirical tests and further development are discussed.
This is an in-person presentation on July 21, 2023 (09:40 ~ 10:00 UTC).