About
Konstantinos' work connects behavioural science with analytics: He employs experiments to understand lay and expert decision making under conditions of radical uncertainty; harvests this understanding to build models of how such decisions should be made; and tests multiple models using machine learning methodologies. This approach is enjoying a rare success—the models are accurate and at the same time transparent.
Konstantinos' group has developed tools for challenging problems such as identifying threats at security checkpoints in Afghanistan while minimizing civilian casualties, regulating UK investment banks without curbing financial innovation, and predicting the incidence of influenza in the US more robustly than big data. Such psychologically inspired quantitative models help set high standards of transparency and effectiveness for approaches such as machine learning algorithms. With the increasing use of artificial intelligence for decision support, understanding decision rationales is vital, especially in sensitive domains, in future epidemics, and other critical events.
Research
Research groups
Research interests
- Human decision making: Prescriptive and descriptive
- AI decision making: Transparency
- Behavioural operations: Modeling and prediction of human behaviour
Current research
Konstantinos works on integrating standard decision theory with the simple rules of thumb people use. He has played a key role in building an interdisciplinary theory of the conditions under which simple models outperform more complex ones, and vice versa. Konstantinos engages with government and other organizations on problems, characterized by complexity and uncertainty, in areas such as health and wealth. He is passionate about projects that enhance well-being, and takes great care to disseminate results to first users. For example, a decision aid for improving civilian safety at checkpoints is included in an Army manual.
Konstantinos' work has been funded by organizations such as the NASA Ames Research Centre, the German Science Foundation, and the European Network for Excellence; published in journals such as Psychological Review, Transportation Research Part B: Methodological, the Journal of Supply Chain Management, AI Magazine, and the European Journal of Operational Research; and covered by media such as the Tageszeitung, Founding Fuel, and Science News. He regularly delivers lectures and short courses to academia and industry--for a lecture on simple heuristics to a general audience see David Blockley Lecture on Systems, for a public debate on whether people think irrationally see Cog Talk, for a plenary conference lecture on behavioural OR see Statistics in Behavioural OR, and for a short video on decoding human behaviour with big see Modeling Human Behaviour.
Konstantinos is the lead author of the book Classification in the Wild: The Science and Art of Transparent Decision Making (MIT Press), with Özgür Şimşek (head of AI group at the University of Bath), Marcus Buckmann (senior data scientist at the Bank of England), and Gerd Gigerenzer (director at the Max Planck Institute for Human Development); see Transparent AI. This short volume aims at presenting to a broad audience of academics and practitioners the overarching themes of the simple-heuristics approach, demonstrate its impact, and show how to apply it. Konstantinos is now working on a single-authored book contracted by Palgrave Macmillan, entitled Cognitive Operations: Models that Open the Black Box and Predict Our Decisions, scheduled for publication in 2023. This book will present and analyse a spectrum of tools for explaining and predicting human behaviour in operational contexts.
Konstantinos is the academic chair of The OR Society's (UK) Special Interest Group on Behavioural OR. He is also very active in the interdisciplinary area of bounded rationality, co-organizing the Annual Winter School in India on the topic. He serves as an associate editor/editorial board member of the Journal of the OR Society, Judgment and Decision Making, the Journal of Mathematical Psychology, and Perspectives on Psychological Science.
Research projects
Active projects
Publications
Teaching
- Module leader for Behavioural Operations, postgraduate module
- Module leader for Management Analysis, core module on introductory undergraduate mathematics and statistics
- Lecturer, Qualitative and Quantitative Methods, core module on postgraduate research methods
- Short course leader, Behavioural OR, National Training Centre in OR
- Lecturer at the Summer Schools on Bounded Rationality (Max Planck Institute for Human Development) and Behavioural OR (rotating – Aalto, Nijmegen, Loughborough)
- Co-founder and co-organizer of the Annual Winter School on Bounded Rationality in India (T. A. Pai Management Institute)
External roles and responsibilities
Biography
Konstantinos studied applied mathematics and cognitive psychology in the Universities of Athens and Freiburg and then obtained a PhD in operations research from the University of Massachusetts Amherst. Before joining Southampton, he was a lecturer at the Department of Mechanical and Industrial Engineering at UMass (2 years and 9 months), visiting assistant professor at the Department of Operations Research of the Naval Postgraduate School (3 months) and the Engineering Systems Division at MIT (2 years); as well as a research fellow/scientist and deputy director at the Centre for Adaptive Behaviour and Cognition of the Max Planck Institute for Human Development (13 years).
Prizes
- Co-Author, Best paper award, International Symposium on Distributed Computing and Artificial Intelligence, Special Session on Multi-Agents and Macroeconomics best paper award, “Davidovic, S., Arinaminpathy, N., Galesic, M., and Katsikopoulos, K. V., Modeling uncertainty in a banking network”, 2014 (2014)
- German Science Foundation Scholarship for Young Researchers (2007)
- Co-Author, Best paper award, ASME Design Theory and Methodology Conference, “Frey, D. D., Herder, P. M., Wijnia, Y., Subrahmanian, E., Katsikopoulos, K. V., and Clausing, D., The Pugh controlled convergence method: Model-based evaluation and implications for design theory”, 2008The Pugh controlled convergence method: Model-based evaluation and implications for design theory (2008)
- Fellow of the US Psychonomics Society (2016)