The Machine Emotional Quotient (MEQ) as a Candidate Framework for Artificial Affective Regulation
Dr.Jagadeesh Selvaraj
Abstract
Whether machines can feel remains unresolved. This article proposes the Machine Emotional Quotient (MEQ) as a candidate framework for studying one narrower target: regulation in an artificial agent whose internal variables are sensed and linked to a valence-like reward. The proposed index combines four provisional dimensions with a causal necessity gate; it is not a validated instrument. The manuscript reports a ten-seed tabular foraging pilot, but the original code and per-seed data were unavailable for this revision, so those values cannot be independently reproduced. A separate independent reimplementation tested two safe starts, 20 training seeds, and checkpoints at 1,200 and 2,000 episodes. Mean survival was often high, but paired checkpoint intervals included zero and seed-level failures remained. These findings establish limited feasibility under the reimplementation’s stated assumptions, not validation of MEQ or reproduction of the original pilot. A descriptive analysis of two unloaded robot-telemetry routines found different motion-current contrasts, but the records expose no controller channel or intervention and do not constitute an MEQ evaluation on a real system. The contribution is therefore a research proposal, a transparent account of preliminary evidence and measurement problems, and a staged plan for validation. None of the results bears directly on subjective experience.