The Dynamics of Emotion as Constrained Selection

Most theories of emotion assume that emotional experiences arise from the interaction between bodily states, cognitive processes, and the surrounding environment. Despite this convergence, a central empirical problem remains insufficiently explained: similar physiological states can give rise to markedly different emotional experiences depending on context.Existing approaches provide partial accounts of this variability. Biological models characterise the role of bodily signals, predictive and constructivist frameworks emphasise inference and conceptual knowledge, and socio-cultural theories highlight the influence of norms and shared meaning. However, these perspectives often fail to distinguish between two levels of top-down organisation: conceptual knowledge, which provides the categories used to interpret affective states, and socio-cultural constraints, which regulate which interpretations become plausible and stable in a given context.In this article, we propose a tri-directional framework in which emotion emerges from the ongoing interaction between bodily signals, predictive processes, and socio-cultural constraints. Within this perspective, emotional experience is conceptualised as a process of constrained selection under uncertainty: bodily signals generate ambiguous affective input, predictive processes organise candidate interpretations, and socio-cultural constraints bias their stabilisation.A central implication of this framework concerns the role of stress. Rather than producing a uniform increase or decrease in emotional responding, stress is conceptualised as a constraint on regulatory dynamics that reduces the range of accessible interpretations and amplifies the system’s dominant mode of stabilisation. This leads to the prediction that, under stress, emotional responses will diverge rather than converge depending on contextual and socio-cultural factors.By integrating biological, inferential, and socio-cultural perspectives within a unified framework, this approach provides a more precise account of emotional variability and generates testable predictions regarding the dynamics of emotion under conditions of uncertainty.