Towards A String Theory of Human Decision Making

Abstract

We introduce a new decision-theoretical model, which allows to analyze human decision making together with its contextual decision architecture using objects called strings. This model extends (i) current understanding of human decision making obtained through traditional decision theories (such as Expected Utility Theory, Cumulative Prospect Theory, Regret Theory, etc.) where choices are typically considered in their discrete form as well as (ii) current advances in cognitive psychology, particularly Quantum Probability Theory, which operationalizes decision making process by using vectors (representing cognitive states) and subspaces (representing different choice possibilities). We consider how strings propagate through decision architecture and interact with each other, allowing to explain important stochastic choice modelling problems (e.g., typical problems when in repeated choice individuals opt for different options for at least a quarter of all of their choices). Model applications as well as mechanisms to employ Machine Learning algorithms for testing the model will be discussed.

Bio

Ganna Pogrebna is a Professor of Business Analytics and Data Science at the University of Sydney Business School and the Lead for Behavioural Data Science at the Alan Turing Institute, UK. Blending behavioral science, computer science, data analytics, engineering, and business model innovation, Ganna helps businesses, charities, cities, and individuals to better understand why they make decisions they make and how they can optimize their behavior to achieve higher profit, better (cyber)security, more desirable social outcomes, as well as flourish and bolster their well-being. She is interested in analyzing individual and group decision-making under risk and uncertainty (ambiguity) using laboratory experiments, field experiments and non-experimental data (specifically, large non-experimental datasets). She studies how decision-makers reveal their preferences, learn, co-ordinate and make trade-offs in static and dynamic environments. Her work aims to develop quantitative models capable of describing and predicting individual and group behavior under risk and uncertainty. Using algorithmic approach, Ganna works on hybrid models at the intersection between decision theory and machine learning (particularly, Anthropomorphic Learning). Her recent projects focus on smart cities, smart technological and social systems, cybersecurity, Artificial Intelligence, human-computer interactions (HCI), human-data interactions (HDI), and business models. Ganna is one of the authors of the Cyber Domain-Specific Risk Taking scale (CyberDoSpeRT), a tool which allows to construct behavioral segmentation in order to design cybersecurity solutions, which received the Organizational Psychology Award from the British Academy of Management in 2018. Her work on risk modeling and understanding human behavior under risk and uncertainty was published in highly reputable peer-refereed academic journals and recognized by the large number of awards including the Leverhulme Fellowship Award as well as the Economic and Social Research Council– the Alan Turing Institute Fellowship Award. In January 2020, she was recognized as one of the 100 top women working in technology in the UK for her contributions to understanding human risk taking behaviour using AI.