Presentation Profiles in Global Broadcast: Support Structure Beyond Content

Authors: Mark Walsh
Cited by: 0

Abstract

Two internal states can carry the same content yet differ in subjective quality and control, such as vivid versus faint, confident versus doubtful, or urgent versus indifferent. Many information-theoretic and predictive-processing accounts treat these differences as local precision or confidence attached to a content estimate. We propose that the broadcast state can preserve compact support structure alongside content, and that this globally available support constitutes a presentation profile. We introduce a minimal bookkeeping that distinguishes latent content variables (X) from evidence features (E) and vehicle variables (F) that summarize how evidence was acquired and encoded. Architectures fall along a spectrum from content-dominated broadcast, where support variables influence processing only locally, to content-plus-presentation broadcast, where compact support summaries (E,F) are preserved in global state and reused. We operationalize this distinction with the conditional mutual information I(Wt;Et,Ft|Xt), which measures how much broadcast activity carries about support structure beyond content. A key functional addition is an auditor: a meta-controller that arbitrates across subsystems using presentation profiles to produce system-level confidence, resolve cross-stream conflict, and accumulate an audit trail that shapes future routing and policy selection. The framework predicts dissociations in which reported quality and metacognitive calibration track support preservation even when content discrimination is approximately fixed, and in which disrupting arbitration separates local module confidence from system-level assessment. We illustrate these failure modes with blindsight, anosognosia, and split-brain syndromes, and discuss implications for artificial systems.

Fields of Study & Subjects

Philosophy and Theoretical Science Embodied and Extended Cognition Cognitive Science and Education Research

Publication details

Journal / source
PhilSci-Archive (University of Pittsburgh)
Publication date
1 January 2026
Language
EN
Document type
Preprint