Modeling Ambiguity and Expert Hesitancy in Decision Frameworks

34 concepts30 papers1 proposal

When human experts evaluate complex systems—such as the environmental impact of a supply chain or the safety of a railway network—they rarely possess complete information. Instead, their insights are shaped by varying degrees of confidence, ambiguity, and outright hesitancy. Historically, early expert systems struggled to handle this nuance natively. As noted by Reid G. Smith et al. in their case study of The Dipmeter Advisor System - A Case Study in Commercial Expert System Development, early frameworks often lacked sophisticated mechanisms for managing uncertainty, relying instead on simple rule ordering or human intervention to resolve conflicting data.

Traditional Multi-Criteria Decision Making (MCDM) models often assume a world of complete information and well-defined problems, as highlighted by Matthias Ehrgott et al. in 09041 Working Group 4: MCDM and RIMO. When forced into these rigid frameworks, experts are required to provide precise numerical scores. If an expert is uncertain or lacks specific knowledge, forcing a response introduces artificial certainty, corrupting the data and leading to suboptimal decisions. To solve this, a major research direction focuses on mathematically formalizing what we do not know alongside what we do know.

From Binary Logic to Hesitancy-Aware Fuzzy Sets

To capture the true spectrum of human uncertainty, researchers have evolved traditional fuzzy sets into more expressive representations. Standard fuzzy logic maps an opinion to a single membership degree (e.g., how "green" a project is). However, this fails to capture an expert's doubt or lack of information.

The first leap forward came with Intuitionistic Fuzzy Sets (IFS), which introduce both a membership degree and a non-membership degree. Dejian Yu et al. applied interval-valued intuitionistic fuzzy sets in Evaluation about the Performance of E-Government Based on Interval-Valued Intuitionistic Fuzzy Set to better handle the inherent vagueness of expert opinions. Similarly, Xianbin Wu et al. developed induced aggregation operators in Intuitionistic Linguistic Multiple Attribute Decision-Making with Induced Aggregation Operator and Its Application to Low Carbon Supplier Selection to allow experts to express preferences using linguistic variables rather than precise numbers.

Yet, even intuitionistic sets struggle when an expert is completely hesitant or refuses to provide a score due to a total lack of knowledge. To model this explicitly, researchers developed Picture Fuzzy Sets and Spherical Fuzzy Sets, which split the evaluation space into three distinct, independent dimensions: membership, non-membership, and hesitancy.

In TOURISM ENVIRONMENTAL IMPACT ASSESSMENT BASED ON IMPROVED AHP AND PICTURE FUZZY PROMETHEE II METHODS, Chao Tian et al. demonstrated that accounting for human hesitancy prevents data inaccuracies caused by forcing a response. When evaluating highly ambiguous domains like offshore extraction, Xiuli Geng et al. utilized spherical fuzzy sets in A Hybrid Decision Framework for Greenness Evaluation of Offshore Oil and Gas Field Extraction to capture the nuanced degrees of membership, non-membership, and hesitancy simultaneously. Further extending this concept, Interval Neutrosophic Sets (INSs) introduce an explicit indeterminacy degree, which Limin Su et al. leveraged in Project Procurement Method Selection Using a Multi-Criteria Decision-Making Method with Interval Neutrosophic Sets to handle the indeterminate nature of expert judgments without losing valuable information.

Fusing and Propagating Incomplete Evidence

Representing uncertainty at the input level is only half the battle; a robust decision framework must also track how this incomplete information propagates through a system.

One powerful mathematical tool for this is Dempster-Shafer Theory (or evidence theory), which generalizes probability theory by allowing probabilities to be assigned to sets of outcomes rather than single points. This is ideal for handling incomplete or subjective data fusion. For instance, Felix T. S. Chan et al. demonstrated in Evidential Supplier Selection Based on Interval Data Fusion how fusing interval data within a Dempster-Shafer framework preserves a more nuanced representation of expert criteria weights. In safety-critical domains, Jérémie Guiochet et al. combined structured argumentation with Dempster-Shafer theory in Modelling confidence in railway safety case to quantitatively assess confidence levels when relying on imprecise expert judgments.

To understand exactly how these gaps in knowledge affect the final outcome, Junyi Zhu 0008 et al. introduced a partial derivative analysis in Investigations of Symmetrical Incomplete Information Spreading in the Evidential Reasoning Algorithm and the Evidential Reasoning Rule via Partial Derivative Analysis. By calculating the partial derivatives of the final decision output with respect to rule weights and reliability, the authors mapped exactly how incomplete information spreads through the Evidential Reasoning algorithm, providing an analytical window into decision-making under uncertainty.

The Dynamics and Cognitive Roots of Imprecise Beliefs

Why do these mathematical abstractions work so well in practice? Because they mirror both the statistical realities of incomplete data and the cognitive mechanisms of human decision-making.

From a statistical perspective, when an agent operates under deep ambiguity, its beliefs cannot be modeled as a single probability distribution. Instead, they form a set of distributions known as imprecise probabilities. Sayan Mukherjee 0001 et al. provided a critical theoretical foundation for this in Ergodic theorems for dynamic imprecise probability kinematics, proving a strong law of large numbers for dynamic imprecise probability kinematics. This mathematical proof shows that even when starting with highly ambiguous, incomplete information, an agent's uncertainty can systematically converge over time as new data arrives.

Cognitively, humans naturally balance expectations against what they do not know. Romy Frömer et al. explored this in Considering What We Know and What We Don’t Know: Expectations and Confidence Guide Value Integration in Value-Based Decision-Making, proposing a Bayesian sequential sampling model. Their work demonstrates that human attention and confidence are dynamically guided by both our explicit expectations and our awareness of our own uncertainty, validating the need for artificial decision systems to model these components separately.

Integrating Imprecision into Practical MCDM Frameworks

To apply these concepts to real-world engineering and strategic problems, researchers embed these advanced uncertainty models into classic Multi-Criteria Decision Making (MCDM) algorithms like the Analytic Hierarchy Process (AHP), DEMATEL, and TOPSIS.

By replacing crisp numbers with fuzzy or rough intervals, these hybrid frameworks allow complex systems to be evaluated reliably across diverse domains:

Ultimately, whether framed as a fuzzy hierarchy or as a Mixed Constraint Satisfaction Problem—where controllable decisions must map against uncontrollable, unknown world states, as explored in Mixed Constraint Satisfaction: A Framework for Decision Problems under Incomplete Knowledge—the core philosophy remains identical. By treating incomplete information, ambiguity, and hesitancy not as noise to be filtered out, but as core variables to be mathematically preserved, modern decision frameworks achieve the resilience required for real-world deployment.

Key Foundational Work

Method For The Best Solutions Search in Multiobjective Decision Problems (1975) — contributed 1 concept elements to this direction.
The Dipmeter Advisor System - A Case Study in Commercial Expert System Development (1983) — contributed 1 concept elements to this direction.
History & citationVersion 1 · 2026-10-04

Current · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_incomplete_information_d3748bad,
  title = {incomplete information},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/incomplete-information/rdr_d3748bad8b4b4c34af2af109ee6108bf},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-20},
}

This exact version (version 1, frozen text):

@misc{lacuna_incomplete_information_d3748bad_v1,
  title = {incomplete information},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/incomplete-information/rdr_d3748bad8b4b4c34af2af109ee6108bf/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-20},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake