A Unified Theory of "Wanting": Multi-timescale, Mechanistic, and Internally-Integrated Drives

This theoretical preprint develops a unified, cross-disciplinary framework for human "wanting" — the set of motivational states that steer behavior toward future states — integrating short-term neuromodulatory reward mechanisms with deeply internalized, developmentally acquired social-belief systems that become unconsciously trusted long-term goals. We propose a three-component architecture: (1) Reward-Driven Wanting, dominated by phasic dopaminergic prediction-error mechanisms and fast reinforcement learning; (2) Internally-Internalized Social Wanting, arising from developmental exposure to social signals and consolidated into persistent goal-belief states that are largely unconscious and resistant to explicit doubt; (3) Reflective/Metacognitive Wanting, involving explicit valuation, narrative coherence, and possible revision of the other two. We formalize these components with a compact dynamical model, propose candidate neural mechanisms rooted in neuromodulation, hierarchical predictive coding, memory consolidation, and cortical–subcortical interactions, and offer testable empirical predictions and experimental paradigms. Finally, we discuss implications for psychology, psychiatry, social dynamics, and artificial agents.