<p>A central aim of affective science has been to reduce the complexity of human emotional experience to a small set of continuous latent dimensions. Among the most influential is the Valence–Arousal–Dominance (VAD) framework. Traditionally, these dimensions are extracted as orthogonal factors using standard linear factor-analytic techniques. In this paper, we propose a different perspective: reframing the same components as forces within a dynamical system, with valence serving as the primary control variable. We show that arousal and dominance, while conceptually distinct, are nonlinearly related to valence through a single-peaked function. In this framework, arousal surges and control efforts rise with stress up to a point, but decline under sustained negative feedback, eventually leading to depression. This representation draws on established models of stress, in particular Selye’s General Adaptation Syndrome and Lazarus and Folkman’s Transactional Model of Stress. We argue that for such single-peaked relationships, standard factor-analytic methods yield informative but overparameterized models. Instead, unfolding models provide a more appropriate and innovative psychometric approach, enabling a unidimensional representation of emotion. We illustrate this with data from an emotion-induction experiment, showing that responses on arousal and dominance items can be recovered as an inverse-U function of an underlying valence dimension.</p>