Understanding how conscious systems regulate behavior under changing internal and external conditions remains a central challenge in cognitive science. Existing theories have substantially advanced our understanding of information integration, global broadcasting, and predictive processing, yet they provide relatively limited mechanistic accounts of how previously acquired representations become selectively accessible for conscious processing. In particular, the dynamic regulation of representational accessibility has received comparatively little attention as an organizing principle of conscious cognition. Here, we propose the Stratified Interface Architecture of Consciousness (SIA), a constraint-based computational framework in which conscious processing is conceptualized as the dynamic regulation of accessibility across historically organized representational layers. Within this architecture, accumulated experience progressively forms stratified representational interfaces that differ in stability, accessibility, and behavioral relevance. We introduce the Layer Accessibility Hypothesis, proposing that behavioral performance depends primarily on the subset of representational layers that remain functionally accessible under current biological constraints rather than on total representational capacity alone. To explain the mechanisms governing accessibility, the framework introduces Meaning Cost as a regulatory variable reflecting the integrative burden imposed by scenario expansion, contextual complexity, self-referential processing, and temporal projection. Increasing Meaning Cost progressively narrows representational accessibility, after which Alignment organizes the remaining accessible representations into behaviorally coherent configurations. These configurations are subsequently evaluated by a threshold-based Decision Gate, which determines when aligned representations become executable. Within this sequence, Self-Control is reconceptualized as an emergent property arising from the successful regulation of representational accessibility, Alignment, and behavioral execution rather than as an independent executive faculty. The proposed architecture further distinguishes representational capacity from functional accessibility, providing a unified explanation for context-dependent variability in cognitive performance while offering a computational framework applicable to conscious regulation, executive function, and computational psychopathology. Rather than replacing existing theories of consciousness, the proposed model complements them by addressing a distinct explanatory level centered on the dynamic regulation of representational accessibility under biological constraints. Finally, the framework generates experimentally testable predictions that may guide future behavioral, computational, neurophysiological, and clinical investigations into the mechanisms underlying conscious regulation.
Reyhan Karatas (Thu,) studied this question.