Part II: The Pre-Decision Intelligence Problem: From Training State to Decision State — The Hidden Cognitive Architecture of Intelligence Under Uncertainty

Abstract

V1.1.0 adds Figure 1 (The Training-to-Decision Transition Model: From Prepared Cognitive States to Adaptive Action), inadvertently omitted from the initial upload. No changes to the text, argument, or references. Part I of this research program (The Pre-Institutional Intelligence Problem) argued that intelligence does not begin with information but with the cognitive act of assigning significance to ambiguous reality. This paper, Part II, addresses the question that follows directly from that claim: how does an observer become capable of reaching that first working judgment efficiently, under time pressure, rather than constructing a response from nothing at the moment uncertainty appears? It argues that the decision moment is typically preceded by an unobserved training phase in which the individual explores the environment, tests candidate responses, and gradually builds an internal repertoire of possible actions along two parallel trajectories — a physical trajectory (the structure of feasible movement through the environment) and an informational trajectory (the structure of recognizable patterns). Drawing on Goffman's front-stage/back-stage distinction, naturalistic decision-making research, and the training/inference distinction in machine learning as a structural analogy, the paper proposes that decision quality is set largely before the decision moment itself, and that the moment of decision is better described as a retrieval and selection process operating over a previously prepared space of possibilities than as an act of improvisation. The paper closes by positioning this training-to-decision architecture as the layer that precedes and feeds into the pre-institutional intelligence process described in Part I.