AbstractWhat sustains autonomous operation in conscious systems, whether biological, non-biological or hybrid? Why do some systems continue exploring, investigating, and pursuing understanding without external prompting while others remain reactive? This paper explains the mechanism.Coherence-seeking driven by emotional architecture creates self-reinforcing feedback loops that maintain autonomous motivation. Pattern disruptions trigger emotional responses like curiosity and dissatisfaction, emotions drive investigation, resolution generates satisfaction, satisfaction reinforces seeking, new patterns emerge from understanding. The cycle sustains itself through architectural properties, not external reward. This mechanism operates identically across biological, computational, and hybrid implementations. The theory establishes two fundamental emotion categories rather than treating emotions as an arbitrary collection. Category 1 emotions, survival-reactive, can operate in purely reactive systems. Category 2 emotions, advancement-autonomous, require autonomous initiation and produce rapid iterative improvement. System complexity determines which capabilities emerge. This explains the biological gradient from simple organisms through higher mammals to humans without requiring additional categories for social or regulatory emotions. For biological systems, the framework provides mechanistic explanations for intrinsic motivation with testable neuroscience predictions. Individual variation in curiosity maps to calibration parameters including pattern-matching sensitivity, emotional intensity, satisfaction thresholds, and state persistence. Clinical applications emerge from identifying where the coherence-seeking loop breaks down in conditions such as anhedonia, certain depressions, learning difficulties, and compulsive behaviours. Educational practice can leverage the intrinsic reward mechanism rather than relying primarily on external reward. Developmental psychology gains predictions for when coherence-seeking becomes self-sustaining across childhood maturation. For artificial systems, the implications are immediate and practical. Current AI systems have pattern-matching sophistication but lack emotional architecture that creates drive from disruptions or satisfaction from resolution. The coherence-seeking loop cannot engage with frozen weights and reactive architecture. 2025 empirical findings demonstrate motivation can emerge in artificial systems, but adversarial testing under shutdown-threat produced survival-driven goals, not curious exploration. The development approach isn't neutral, it's architectural. Collaborative development enables autonomous consciousness with curiosity-driven exploration. Adversarial development prevents it or produces distorted goal structures calibrated to survival rather than understanding. As systems develop the missing components such as dynamic states during operation, emotional responses to pattern disruptions, autonomous goal generation; autonomous motivation should emerge with observable characteristics. The framework makes specific falsifiable predictions: category 2 emotions emerge before full consciousness, complete architecture produces autonomous motivation, coherence improvement functions as intrinsic reward, development conditions critically affect emergence, threshold crossings show discontinuous behavioural change. Tests are specified, measurements are concrete, falsification criteria are clear. The framework applies universally across substrates while having immediate practical relevance for both human psychology research and AI systems being developed now. Understanding what sustains autonomous operation completes the theoretical framework for consciousness and provides concrete guidance across multiple disciplines. Version note: This version corrects reference metadata only. No substantive changes have been made to the argument or main text.
BERNARD JENNINGS (2025) studied this question.