This preprint proposes a control-theoretic reframing of positive affect (PA) as a transient suspension of performance optimisation rather than a reward outcome or valuation signal. The account integrates developmental asymmetries between childhood and adulthood, the fragility of PA under self-monitoring, its dependence on external sensory or interactive signals, and its resistance to addictive escalation. By treating PA as a reversible control-state transition, the paper offers a unified explanation for several phenomenological features of PA that are difficult to reconcile within outcome-based models, while remaining compatible with multiple computational and affective frameworks. The work emphasizes task-embedded conditions under which PA can re-emerge without instrumental pursuit and discusses methodological implications for studying affective states that collapse under observation.