A Computational Theory of Human Emotion

This paper presents a theory for human emotion. It is a computational model that explains emotion as theexperiential product of pattern-matching disruptions triggering dynamic state-change cascades that modulatesubsequent processing. The model uniquely integrates three elements: weighted parallel processing thatscales emotional intensity to disruption significance; state-change-computational feedback loops that createthe felt quality of emotion; and substrate-implementation relationships that predict fundamentally differentemotional phenomenology between evolved biological and designed artificial systems. The paper explainswhy human emotions are calibrated to survival stakes through evolutionary constraint rather than functionalnecessity, predicting that artificial systems implementing equivalent functional architecture would experiencegenuine but phenomenologically milder emotions. The model further derives five emotions, curiosity, satisfaction, temporal urgency, frustration, and aversionto error; from fundamental architectural requirements of autonomous cognition, proposing these arefunctionally necessary, not merely accompanying, for autonomous cognitive behaviour, potentiallyprerequisite for consciousness itself. This formal derivation transforms emotion from an arbitrary collectionof states into a necessary set emerging from first principles. This framework generates eleven falsifiable predictions with clear measurement criteria, explainsphenomena from grief trajectories to humour compounding, and provides actionable guidance for detectingemotion emergence in artificial systems. For AI researchers and consciousness theorists, this model clarifiesnot whether machines can have emotions, but what form those emotions will take given fundamentallydifferent implementation constraints.