Toward a Multi-Layer Decision Architecture for Predictive Formulation, Expert Validation, and Consumer Acceptance in Sensory-Driven Product Development

This paper presents a conceptual and methodological architecture for linking predictive formulation engines, trained sensory panel validation, and consumer hedonic feedback into a unified decision-support framework for industrial product development. The approach bridges the gap between analytical coherence and market acceptance by formally distinguishing structural conformity—whether a product behaves as intended—from hedonic truth—whether it is liked by consumers—and integrating both dimensions through a five-layer synthesis architecture. A weighted, normalized conformity score (C*) is computed from expert panel data and modulated by inter-rater dispersion, then cross-referenced with consumer liking scores (L) and preference segment topology, including a quadratic external preference mapping (EPM) model that accounts for non-linear 'bliss point' dynamics. The model is intentionally staged: an operational heuristic layer for rapid deployment, followed by a data-enriched adaptive clustering framework. The framework is presented as a conceptual and methodological contribution rather than an empirically validated system; its purpose is to formalize the decision logic that currently remains implicit in R&D practice and to define a structured agenda for empirical validation. The resulting architecture is especially suited to food and beverage R&D contexts in which expert panel truth, consumer heterogeneity, and iterative formulation must coexist within a single, computationally grounded environment.