Cross-Substrate Pain Representation: A Unified Formalism for Human, Artificial, and Biological Nociceptive Systems

Pain is conventionally studied within the boundaries of a single substrate — human self-report, animal nociceptive behavior, or engineered damage-avoidance in robotics — with little cross-talk between the resulting formalisms. We previously introduced PainVector, a multidimensional computational framework that represents human pain as a state vector combining self-reported and physiologically inferred intensity with auxiliary dimensions of predictability, controllability, location, memory, and confidence. In this paper we show that the same core computational structure, retained without alteration, extends to substrates that lack self-report entirely, while the operational mapping of individual dimensions — and, for the machine substrate, a small set of substrate-specific extensions — is adapted to the measurable properties of each substrate. We instantiate the framework on two additional substrates: an artificial nociceptive system (the Machine Pain Vector, MPV) and the ASH polymodal nociceptive sensory neuron of Caenorhabditis elegans. The extension rests on a single structural mechanism: substrates without self-report set the self-report weighting term to zero, collapsing the final intensity estimate to the inferred term without any change to the intensity equations. We construct explicit variable-mapping tables for both non-human substrates, grounded in the artificial-nociception and C. elegans literatures, and argue that the formalism's value lies in this demonstrated transportability of form rather than in any claim of shared subjective experience across substrates. We close by discussing a natural extension toward genetically driven inter-individual variability in nociceptive sensitivity, using the C. elegans npr-1 polymorphism as an illustrative case. A supplementary numerical script instantiates the formalism on synthetic, illustrative data for all three substrates to confirm the equations are computable and produce differentiated outputs; this is a worked illustration, not empirical validation, and the variable mappings themselves remain grounded in published literature rather than in jointly validated shared data.