The effect of decentralized behavioral decision making on system-level risk

Area of Science:

• Behavioral economics
• Cognitive psychology
• Risk analysis

Background:

• System-level risks often involve decentralized decision-making by individuals.
• Human decision-making deviates from normative rationality (Expected Utility Theory) and aligns with Prospect Theory (PT), exhibiting heuristics and biases.
• Deception tactics leverage psychological manipulation rather than direct argumentation.

Purpose of the Study:

• To extend Signal Detection Theory (SDT) by incorporating behavioral decision-making principles from Prospect Theory (PT).
• To rederive the optimal cutoff threshold under PT for binary classification tasks.
• To integrate the psychology of deception into SDT's discriminability parameter.

Main Methods:

• Behavioral extension of Signal Detection Theory (SDT) using Prospect Theory (PT).
• Redefinition of the optimal cutoff threshold under PT with a neo-additive probability weighting function.
• Incorporation of deception psychology into SDT's discriminability parameter.

Main Results:

• The PT-based optimal cutoff threshold is unique, computationally tractable, and interpretable.
• The derived threshold is independent of loss aversion and more conservative than the classical SDT threshold.
• Decentralized decision-makers exhibit a bias toward underdetection, increasing system-level risk.

Conclusions:

• Prospect Theory offers a more descriptively accurate model for understanding individual decision-making in risk contexts.
• The conservative bias toward underdetection in behavioral decision-making leads to underestimated system-level risks.
• Integrating behavioral insights into SDT is crucial for accurate risk assessment in decentralized systems.