Our routines play an important role in a wide range of social challenges such as climate change, disease outbreaks and coordinating the staff and patients of a hospital. Studying these systems via agent-based simulations (ABS) enables researchers to gain insight into complex aspects of these challenges such as human interaction, learning, heterogeneity, feedback loops and emergence. Current agent frameworks do not integrate social and psychological evidence on human routines: humans make habitual decisions, interconnect these habits throughout the day and use these interconnected habits as a blueprint for social interaction. This thesis provides the domain-independent SoPrA (Social Practice Agent) framework that integrates theories on social practices to support the simulation of human routines. Social practice theory is a socio-cognitive theory applicable to model human routines as the theory aims to describe our ‘daily doings and sayings’. The first part of the thesis identifies the aspects of social practice theory that are relevant for agent-based simulation, distils requirements from the literature, reviews current agent models and provides the SoPrA framework that satisfies said requirements. The second part describes applications of SoPrA on the value-alignment problem in AI, identifying social bottlenecks in hospitals and comparing theories on how habits break. This results in an agent framework with a clear relation to current evidence and, due to its modularity and focus on domain-independence, is usable for a wide range of ABS studies that involve human routines. As such, SoPrA is relevant for scientific work in (1) ABS by enabling a new way to know, explore and improve the world, grounded in evidence on human routines; (2) in multi-agent systems by enabling agents that understand and interact with human routines; and (3) in the social sciences by crystallizing theories on human routines and enabling exploration of these theories via simulation. Furthermore, this thesis shows the societal relevance of SoPrA for understanding and improving the role of routines in AI safety, emergency rooms, commuting behaviour and consumption behaviour.@en

Inducing behavioural change requires a good understanding of how habits break. We identified two theories in the psychological literature on this process: the decrease theory and persist theory. Both theories are used to explain behavioural change, but one states the original habit fades out, while the other theory states the habit persists. We use agent-based simulation to show that the two theories lead to different behaviour when the agents are motivated to do multiple alternative actions (e.g., take the bike or take the train), instead of one alternative action (e.g., take the bike). This finding is relevant for the social scientific field, because (1) it shows a scenario where it matters if habits persist and (2) it enables an empirical experiment to discern the two theories.@en

Evidence-driven agent-based modeling plays a useful part in understanding social phenomena. By integrating social-cognitive theories in our agent models, we bear evidence from social and psychological studies on our models for human decision-making. Social practice theory (SPT) provides a socio-cognitive theory that emphasizes three empirically and theoretically grounded aspects of behavior: habituality, sociality, and interconnectivity. Previous work has emphasized the importance of SPT for agents, has made abstract models of SPT, or used SPT to study energy systems. This article provides a set of requirements for integrating SPT in agent models and an evaluation of 11 current agent models with respect to these requirements. We find that current agent models do not fully capture habituality, sociality, or interconnectivity, nor is there a model that aims to integrate all three aspects. For example, current models do not support context-dependent habits, use a comprehensive set of collective concepts, and support hierarchies of activities. Our evaluation allows researchers to pick one of the current agent models depending on their needs regarding habituality, sociality, and interconnectivity. Furthermore, this article shows the usefulness of an agent model that integrates SPT and provides requirements that help modelers to achieve this model.

@en

Autonomous agents (AA) will increasingly be interacting with us in our daily lives. While we want the benefits attached to AAs, it is essential that their behavior is aligned with our values and norms. Hence, an AA will need to estimate the values and norms of the humans it interacts with, which is not a straightforward task when solely observing an agent's behavior. This paper analyses to what extent an AA is able to estimate the values and norms of a simulated human agent (SHA) based on its actions in the ultimatum game. We present two methods to reduce ambiguity in profiling the SHAs: one based on search space exploration and another based on counterfactual analysis. We found that both methods are able to increase the confidence in estimating human values and norms, but differ in their applicability, the latter being more efficient when the number of interactions with the agent is to be minimized. These insights are useful to improve the alignment of AAs with human values and norms.@en

Rumour is a collective emergent phenomenon with a potential for provoking a crisis. Modelling approaches have been deployed since five decades ago; however, the focus was mostly on epidemic behaviour of the rumours which does not take into account the differences between agents. We use social practice theory to model agent decision-making in organizational rumourmongering. Such an approach provides us with an opportunity to model rumourmongering agents with a layer of cognitive realism and study the impacts of various intervention strategies for prevention and control of rumours in organizations.

@en

Social simulations gain strength when agent behaviour can (1) represent human behaviour and (2) be explained in understandable terms. Agents with values and norms lead to simulation results that meet human needs for explanations, but have not been tested on their ability to reproduce human behaviour. This paper compares empirical data on human behaviour to simulated data on agents with values and norms in a psychological experiment on dividing money: the ultimatum game. We find that our agent model with values and norms produces aggregate behaviour that falls within the 95% confidence interval wherein human behaviour lies more often than other tested agent models. A main insight is that values serve as a static component in agent behaviour, whereas norms serve as a dynamic component.

@en