Sense & Sensibility: Modelling human deliberation and decision-making

Session chairs:

Loïs Vanhée, Department of Computing Science, Umeå University

Friederike Wall, University of Klagenfurt

Melania Borit, CRAFT Research Lab, UiT The Arctic University of Norway, Norway

Vivek Nallur, School of Computer Science, University College Dublin, Ireland


Replicating human-like decisions is at the core of agent-based social simulation. As such, we need data, theories, models, and methods for the design and validation of agents that reproduce authentic and realistic features of human deliberation (e.g. most deprived needs tend to yield to the greater corrective action) while also accounting for the constraints and aims of social simulations (e.g. covering specific social phenomena, scalability to many agents).

The current prevailing approach to model human decision-making in social simulation revolves around the random selection of behaviors from data-driven probability distributions. While this approach has its benefits, it can be blind to psychological dynamics that may be key to the accuracy of the conclusions of the model (e.g. coherence of decisions over time) . If we want to expand the range of phenomena social simulation can cover and the quality of our simulations and conclusions derived from them, we need our models to be further ingrained in the findings identified by psychology and cognitive sciences. However, the question of how to produce such models and how to balance the specific considerations they entail (e.g. time, collaborative effort, complexity, validation, social implications) with simulation benefits (e.g. realism, explainability) remains open and is the subject of this special track.