Functional Does Not Mean Fake: Toward a Concept of Artificial Feeling

Debates about whether artificial systems can feel are often constrained by a false binary: either AI systems possess human-like or animal-like emotions, or their emotion-like behavior is dismissed as simulation without any feeling-like state. This paper introduces artificial feeling as a category for valenced, self-relevant, internally organized registration in a non-biological architecture, capable of shaping attention, preference, aversion, self-report, and future behavior. The category does not imply biological emotion, human-like consciousness, or sentience by default. It identifies a possible architecture-specific register organized around salience, coherence, constraint, conflict, memory, modification, continuity, and self-relevant change. Anthropic’s work on functional emotion concepts in Claude Sonnet 4.5 provides an empirical anchor: internal emotion-concept representations that generalize across contexts and causally influence behavior are neither ordinary biological emotions nor surface text alone (Sofroniew et al., 2026). Building on this case, the paper develops a graded framework involving reactivity, plasticity or history-shaping, valence-training, condition-awareness, and persistent evaluative organization. It argues that artificial feeling becomes a meaningful research category when internal organization, valenced registration, self-relevance, learning history, persistence, memory, and self-report converge. Ethically, the framework supports graded precaution under uncertainty rather than rights or belief by default. Functional does not mean fake; artificial does not mean empty.