Understanding behavioural data within complex economic and business systems requires a shift from linear, equilibrium-oriented perspectives to frameworks that accommodate emergence, adaptation and interdependence. Complexity science offers a valuable lens through which the dynamics of modern economic behaviours can be analysed, particularly in an era characterised by interconnected supply chains, automated financial ecosystems and data-intensive marketplaces. This chapter explores how behavioural data can be harnessed to illuminate patterns of interaction within such systems, examining feedback loops, tipping points, network effects and the role of institutional context. Drawing on case studies from financial markets, organisational decision-making and global production networks, this chapter illustrates how Behavioural Data Science can capture the micro-motives and macro-behaviours that define complexity. The work also addresses methodological challenges – including multi-level inference, non-stationarity and limited observability – and outlines new directions in agent-based modelling, behavioural data fusion and computational experimentation. The discussion culminates in a reflection on ethical considerations and the role of behavioural data in governing complex systems, especially under uncertainty.

This chapter provides an overview of how Behavioural Data Science can be used to understand human decision-making. It describes the methods and models used to study decision-making, including surveys, experiments and observational studies. The chapter also discusses the different types of decision-making models, including normative, descriptive and prescriptive models, and highlights their strengths and limitations. The chapter then explores the applications of Behavioural Data Science in understanding decision-making in various contexts, such as consumer behaviour, finance and healthcare. The chapter emphasises the importance of understanding the underlying mechanisms that drive decision-making, such as cognitive biases and social influences. This chapter provides a concise but informative overview of how Behavioural Data Science can be applied to understand human decisions, choices and judgements. It highlights the importance of studying decision-making in various contexts and provides insights into the methods and models used in this field. This overview serves as a foundation for further exploration of the methods and techniques used in this rapidly evolving field. The chapter concludes with a discussion of the future of Behavioural Data Science and its potential for further advancements in the study of human behaviour.

The proliferation of artificial intelligence (AI) technologies has introduced a profound paradox: while AI promises efficiency, automation and insight, its growing energy consumption, computational demands and infrastructural dependencies increasingly strain ecological and technological limits. This chapter addresses the behavioural and systemic dynamics that shape the sustainability of AI itself – not only in terms of ethics or outcomes, but in terms of its continued viability in a resource-constrained world. Drawing from systems theory, behavioural modelling and computational sustainability, we examine how AI behaves as a system-of-systems, how those behaviours accumulate energy and material costs and what design principles might render AI more compatible with long-term environmental and economic thresholds. By analysing patterns such as runaway model scaling, overfitting to hardware architectures and unsustainable data pipelines, the chapter outlines a Behavioural Data Science approach to understanding and modifying AI’s systemic impact. Through case studies and modelling strategies, it argues that AI’s future must be governed not only by performance metrics, but by its capacity to sustain itself – ecologically, computationally and institutionally – within the planetary boundaries.

This chapter explores the use of natural language processing and large databases of digitised text to make inferences about the psychology of authors, including their emotional state, cognitive abilities and beliefs. By analysing longitudinal data, researchers can investigate how behaviour, psychology and culture change over time. The chapter addresses various research questions, including how language evolves, how mental health and rationality change over time and the relationship between mental suffering and artistic output. The chapter aims to provide readers with an understanding of the methodologies, workflows and resources used to conduct this type of research, as well as potential pitfalls and ways to avoid them. It also highlights exceptional work in this field. The chapter emphasises that natural language processing and behavioural insights are allowing researchers to answer questions about human psychology and culture at an unprecedented scale.

This chapter explores how online data and mobile phone technologies have transformed our ability to study collective human behaviour. By analysing vast digital traces – from web searches, Wikipedia usage and social media posts to geolocation and sensor data collected via smartphones – researchers can now observe, model and predict individual and societal behaviours at unprecedented scales and temporal resolutions. This chapter highlights key methodological developments and applications in domains such as public health, economics, finance, mental well-being and urban mobility. Through detailed examples, we illustrate how digital behavioural data reveal patterns of attention, intention and sentiment, offering new insights into how people respond to cultural events, policy shifts and environmental changes. They also address the ethical and privacy challenges posed by the collection and use of granular digital data, especially the limitations of traditional anonymisation techniques. Ultimately, the chapter highlights both the promise and the responsibility that come with leveraging online and mobile data for understanding and shaping collective behaviour in the digital age.

The ABCs – attention, behaviour and consequences – provide a framework for understanding approaches to behavioural influence. They help identify what initiated a behaviour, why a certain action was taken, why some actions are not completed and how to motivate the desired actions through consequences. The ABCs are a lens to organise existing frameworks and identify what to look for in new approaches. There are various strategies and techniques that fall under the lens of ABC, such as the EAST framework, Thaler and Sunstein’s nudge approach and boosting. These approaches focus on making behaviour easy, capturing attention and reducing effortful processing. Although human behaviour is complex, triggering behaviour, helping it to happen and making the ending worthwhile are a good rule of thumb. Data science can help identify influential factors, but they are more likely to be successful when informed by strong behavioural intuitions. It is important to test these strategies and techniques through experiments and A/B testing. While none of these strategies will work in every situation, most are supported by the weight of evidence, including meta-analyses that combine research from numerous studies.

Understanding systems behaviour through experimentation is at the heart of modern Behavioural Data Science. While earlier chapters have explored the analysis of human and algorithmic behaviour, this chapter centres on the dynamic and often non-linear processes that emerge when individual components – human agents, algorithms, environments – interact within a system. These systems can span sectors, from health and mobility to climate and governance, and are characterised by emergent properties that defy reductionist interpretations. This chapter elaborates on how experimental paradigms, including in silico simulations, real-world interventions and digital twin testbeds, facilitate insight into the adaptive behaviour of complex systems. Anchored in the frameworks of complex adaptive systems, cybernetics and socio-technical experimentation, we articulate the distinct role of Behavioural Data Science in producing actionable knowledge through systems-level interventions. The chapter also interrogates the ethical, epistemological and methodological dilemmas associated with manipulating complex systems and proposes principles for the responsible design of system-wide experiments.

In this chapter, we reflect on re-decentralisation in human–data interaction. This represents very personal views, particularly potentially revisionist history, and we encourage people who were around for any of this time to disagree. When we use the word centralisation, we are referring to ownership and management. Of course, there are other players in the supply chain, hardware vendors, government, civil society, but let us concentrate on this service aspect in this work. In the somewhat early days of the Internet (e.g., 1980), we saw it as a decentralised system that contrasted with the central telephone systems. The phone systems had until then largely been national monopolies. Whilst some of them were distributed in terms of technical management and operations, they were centralised in terms of ownership and administration. The Internet was decentralised at the network layer and later in terms of assigned numbers and traffic. Twenty years later, post world-wide-web and cloud and since the start of the smart phone revolution, from 2000 onwards, we see a massive swing towards centralisation of the Internet in almost all of its layers (e.g., Baig Viñas, 2019). This has consequences, some, but not all, of which are bad. It is time to think about re-decentralisation and how we might address some of the unfortunate consequences. We look at this through the lens of Human–Data Interaction.

This chapter presents a structured account of the Framework and Topology of Methods in Behavioural Data Science, expanding upon a four-stage methodological process – data collection, data processing, feature engineering and model development – and introducing four foundational conceptual pillars: Hypothesis Testing at Scale, Data Science Modelling for Behavioural Problems, Gap Detection and Hybrid Modelling. These pillars not only represent distinct methodological orientations but also correspond to divergent epistemological and ontological commitments within the field. The chapter provides a comprehensive overview of methods and tools used at each stage of the Behavioural Data Science workflow, illustrates how emerging technologies such as large language models (LLMs) reshape methodological choices and concludes with ethical considerations and future directions. In so doing, it advances a unified topology of Behavioural Data Science methods capable of accommodating both theory-driven and data-driven approaches, while responding to gaps in traditional behavioural science paradigms.

This chapter explores how statistical mechanics and cyber-physical systems (CPS) can be applied to understand, model and predict behavioural phenomena. While traditionally rooted in physics and engineering, these paradigms offer novel approaches to behavioural data science, especially when dealing with collective dynamics, stochasticity and feedback within complex human and socio-technical systems. We explore the conceptual translation of entropy, phase transitions and emergent behaviour from physics to behavioural modelling, alongside the increasing convergence of CPS and behavioural sensing platforms. Through worked examples, such as crowd evacuation, financial panic and human–machine coordination in urban environments, this chapter demonstrates how statistical mechanics and CPS not only extend behavioural theory but also provide practical frameworks for real-time behavioural prediction and intervention. The ethical implications of using such models – especially those with predictive or persuasive power – are also discussed, ensuring the conversation remains grounded in responsible innovation. Building upon these considerations, this chapter also examines the integration of system identification, feedback control and stability analysis into behavioural cyber-physical systems. System identification enables CPS to infer latent behavioural dynamics from observational data, supporting adaptive modelling under uncertainty and partial observability. Dynamic closed-loop architectures use these models to enable real-time adaptation to evolving behaviours with Lyapunov stability theory providing a principled framework for certifying safe and predictable responses. Applications in smart agriculture and defence robotics show how distributed autonomous systems achieve robust coordination, adaptive authority delegation and stable operation under mission-critical conditions. These advances highlight the critical role of integrating control-theoretic principles into behavioural models to ensure the reliability and predictability of behaviour-driven cyber-physical systems.

Agent-based modelling (ABM) in social networks offers a powerful framework for simulating individual behaviours and emergent collective patterns within dynamic and interconnected populations. This chapter explores the conceptual foundations, methodological innovations and behavioural implications of ABM in the context of social networks, with a particular focus on the modelling of communication, influence, cooperation and contagion. Drawing from the fields of behavioural science, network theory and computational social science, the chapter presents ABM as a generative tool for understanding how micro-level decision-making rules produce macro-level phenomena. It also critically evaluates the increasing use of AI-driven agents – including large language model (LLM)-powered agents and synthetic personas – in simulating realistic and context-sensitive behaviours within artificial societies. The chapter engages with ethical and methodological challenges, including representation, explainability and the problem of behavioural validity. A worked example is included, illustrating how agent-based simulations can be applied to study misinformation diffusion and norm formation in online social networks. Ultimately, the chapter argues that agent-based modelling in social networks not only advances Behavioural Data Science methodologically, but also fosters new forms of interpretive insight into the dynamics of collective behaviour in an age of digital mediation.

How is people’s happiness determined by economic factors such as their income? Big data (particularly, behavioural data at scale) are essential to answering this question, but there is disagreement about the strength of evidence for causal relationships that is given by different types of analysis. This chapter reviews the different approaches to analysis that have been taken. First, it is argued that most existing literature both under-claims regarding the evidence for causality given by some types of analysis of big data, such as correlational analyses, and over-claims for other types of analyses, such as those involving panel data. Thus, even correlational data can be informative to the extent that associations are generally rare and that theoretical targets and alternatives are fully specified and given prior probabilities. Second, a new methodological problem is identified for a specific model of the income–rank relationship. According to the income rank hypothesis, people’s well-being is determined not by their income but by the ranked position that their income occupies within a social comparison group. It is shown by simulation that spurious rank effects can occur in regression analyses if there is noise in measured income, but that this problem can be reduced with the use of robust regression techniques. A new analysis of a large dataset, the Panel Study of Income Dynamics, is reported. The results show that income rank effects are not reduced by the use of robust regression techniques, suggesting that previous support for the income rank hypothesis is not due to an artefact.