A Multi-Attribute Extension of MDFT

From everyday consumer technologies and business processes to high-stakes domains such as military operations and critical care, autonomous systems now permeate nearly every facet of modern life, making it essential to model and predict human decision-making under complex and uncertain conditions. This paper introduces an extension of Multi Alternative Decision Field Theory (MDFT) that enables decision modeling over 3 or more attributes while preserving the mechanisms that provide its explanatory and predictive strength. Building on the generalized psychological distance (GPD) presented by Hotaling, the proposed approach decomposes each decision by considering pairs of attributes and options. Pairwise distances are computed using the GPD and aggregated to form a higher-dimensional representation of multi-attribute preferences. We show that the extended model captures similarity, compromise, and attraction effects in settings with increased dimensionality of the decision space in terms of attributes and that it is capable of capturing real-world choice distributions. By extending MDFT beyond 2 attributes, this work contributes to bridging the gap between simplified laboratory-based decision modeling and the complexity of real-world decision-making. The resulting framework provides a foundation for modeling dynamic, context-dependent, preferences in domains where decisions involve numerous competing priorities. This extension expands the applicability of MDFT to higher fidelity decision environments and supports future efforts in human-autonomy teaming (HAT) and context-aware decision prediction.