Human Temporal Architecture: Neural Dynamics, Memory, Prediction, and Experience

Human temporal cognition is often compressed into a single internal-clock metaphor. The evidence instead supports a typed architecture in which bodily state, neural dynamics, event segmentation, memory transformation, prediction, control, and task-specific temporal reports occupy different causal positions. This paper develops a self-contained synthesis across neural timescales, interoceptive rhythms, energetic constraints, event-structured memory, retrieval reconstruction, prospective prediction/control, and temporal judgment. Its central methodological result is negative but productive: numerical similarity, correlation, or even local rank closure across levels does not establish mechanistic identity. Measured neural timescales need not equal generated circuit timescales, functional integration windows, represented durations, or reported time. Event boundaries are bundled perturbations rather than unitary causes; remembered duration can depend on stored duration, contextual change, local position, representational change, prediction error, retrieval context, and query construction. Bodily and energetic variables constrain computation but are not thereby clocks. The paper therefore replaces a master-clock architecture with experimentally separable bridges from bodily state to neural dynamics, temporal inference, memory/prediction, and readout, while retaining an explicit boundary between functional temporal organization and phenomenal identity.