Decoding Affect: A Computational Model of Affective Adaptation and Construction

Affective construction and adaptation are complementary processes through which affective value is generated and evolves over time. Although prior research has identified key determinants of these processes, including reference points, stimulus inherent value, attention, negativity bias, adaptation rate, and time, a unified framework that quantifies how these factors interact and jointly shape affect remains lacking, leaving important nonlinear affective phenomena insufficiently explained. We propose the Attention-Reference Interaction Adaptation (ARIA) framework, which formalizes affective valuation as an attention-gated dual-reference integration process that evaluates current sensory inputs relative to both prior experience and allostatic set-points. By specifying how value salience and negativity bias jointly govern attention allocation and adaptation rate, ARIA suggests that relativity and asymmetry (two foundational properties of affective valuation) arise from a shared attentional mechanism. Consequently, these properties are contingent, dynamically coupled, and conditionally reducible, rather than fixed, independent, and irreducible axioms as assumed in classical behavioral economic theories. Furthermore, we map ARIA's computational components onto reward and interoceptive prediction errors, capturing how external and internal discrepancies jointly drive the construction of affective experience. Through numerical simulations and an initial empirical test, we show that ARIA reproduces classic affective phenomena, such as exponential decay, affective after-reactions, and alliesthesia, while also accounting for anomalous patterns, including cases in which environmental improvements fail to elicit pleasure and conditions under which inherently positive stimuli adapt more slowly than negative ones. Ultimately, ARIA offers a tractable computational framework for explaining affective dynamics and generating testable predictions about well-being and environmental experience.