An Inference-Based Architecture for Intent and Affordance Saturation in Decision-Making

2025-12-28

AI Summary

Traditional choice models fail to account for decision paralysis—the inability to act despite being informed and motivated—posing a fundamental challenge to rational decision-making theories. Method: We propose a hierarchical reasoning architecture that decouples *intention selection* (goal formation) from *affordance selection* (action execution), and introduce a novel bidirectional KL-divergence variational inference framework—combining reverse and forward KL objectives—to formalize intention saturation and affordance saturation as distinct modes of convergence failure under a unified divergence-minimization objective. We embed autism-related decision rigidity within a generalizable reasoning continuum and integrate static and drift-diffusion dynamic modeling for multi-alternative response-time simulation. Results: Our model successfully reproduces key empirical phenomena—including decision inertia, systemic response stalling, and heavy-tailed reaction time distributions—while clearly distinguishing the two saturation failure modes. This provides a novel theoretical framework and computational foundation for understanding non-pathological decisional slowness.

Technology Category

• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning
• Reasoning under Uncertainty: Sequential Decision Making
• Knowledge Representation and Reasoning: Preferences

Application Category

• User Modeling, Personalization and Recommendation: Psychology-informed user models and recommender systems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Decision paralysis, i.e. hesitation, freezing, or failure to act despite full knowledge and motivation, poses a challenge for choice models that assume options are already specified and readily comparable. Drawing on qualitative reports in autism research that are especially salient, we propose a computational account in which paralysis arises from convergence failure in a hierarchical decision process. We separate intent selection (what to pursue) from affordance selection (how to pursue the goal) and formalize commitment as inference under a mixture of reverse- and forward-Kullback-Leibler (KL) objectives. Reverse KL is mode-seeking and promotes rapid commitment, whereas forward KL is mode-covering and preserves multiple plausible goals or actions. In static and dynamic (drift-diffusion) models, forward-KL-biased inference yields slow, heavy-tailed response times and two distinct failure modes, intent saturation and affordance saturation, when values are similar. Simulations in multi-option tasks reproduce key features of decision inertia and shutdown, treating autism as an extreme regime of a general, inference-based, decision-making continuum.

Problem

• Modeling decision paralysis from intent and affordance saturation
• Separating intent selection from affordance selection in hierarchical decisions
• Simulating autism as extreme inference-based decision-making continuum

Innovation

• Hierarchical decision process separates intent and affordance selection
• Formalizes commitment using reverse and forward KL divergence objectives
• Simulations reproduce decision inertia via forward-KL-biased inference