Developmental Plasticity Mediation: A Causal-Explanatory Framework for Functional Affect in Artificial Systems

We propose Developmental Plasticity Mediation (DPM) as a candidate causal-organizational framework for studying functional affect in artificial systems. DPM concerns a history-conditioned mediator whose intervention changes how later evidence updates a learned significance state, rather than merely biasing current inference or action. We define an intervention-based evidence-response criterion, distinguish it from persistence, historical acquisition, regulation, and cross-context transfer, and state an identifiability limit: any fully specified DPM implementation can be represented by an augmented-state adaptive controller using the same transition and policy structure. Consequently, the mediator does not establish a unique computational mechanism or, by itself, identify an emotional process. A higher-level DPM description would be justified only if it provided stable held-out interventional predictions, transfer, or explanatory compression beyond strong matched adaptive-control and meta-learning alternatives. We give formal propositions and counterexamples, clarify common errors involving boundedness and prediction-error equivalence, and outline a proposed model-comparison study. The manuscript makes no claim that DPM is a novel low-level mechanism, that it is necessary or sufficient for emotion, or that any system has subjective feeling or phenomenal valence.