The Constructivist Computational Model of Emotions: Computational Preparation and Interpretive Commitment

Computational models of emotion usually treat emotion as the output of appraisal. I present the Constructivist Computational Model of Emotions (CCME), a small, hypothetical, executable specification that draws the line elsewhere. In the CCME, computation prepares and stops. Predictive simulation, allostatic regulation of the body budget, retrieval of category schemas, and weighting under prediction error narrow the field of possible interpretations, form a situated referent, and rank candidate concepts. Emotion begins only when an agent interprets that referent against a concept's identity conditions under a governing goal, commits to the concept, and marks the commitment with an identifier. I call the transition the meaning boundary. The model separates learned category schemas, which predict and rank, from emotion concepts, which fix admissibility; it separates a default hedonic-regulatory goal, under which a system attributes, from a goal an agent adopts for the act, under which the agent interprets; and it locates error in the commitment, so that a feeling can remain accurate while the interpretation bound to it fails. Two C# programs check these distinctions: one prepares a full trace and constructs no emotion, and one constructs fear or interest from the same snake under different goals. What is new is a hard boundary between computational preparation and interpretive commitment. Working paper 4 of 8 in the series on the frameworks of How to Design an Emotion: The Deconstruction and Computation of Emotional Meaning from First Principles (Universal Moral Concepts III, New Intellectual LLC, 2026; ISBN 979-8-9898135-9-9 paperback, 979-8-9898135-3-7 Kindle). The book is the full treatment. Book page: https://newintellectual.com/how-to-design-an-emotion; related: Part IV. Buy on Amazon.