This preprint introduces the Human Cognition Measurement System (HCMS), a cognition-aware assessment framework designed to measure human understanding beyond correctness-based evaluation. Traditional assessment systems treat accuracy as a proxy for mastery, ignoring metacognitive alignment and reasoning stability. HCMS operationalizes understanding as a multi-dimensional construct by integrating accuracy, confidence calibration, repeated-trial consistency, and robustness under controlled perturbation. Through controlled experiments, this work demonstrates that learners with similar accuracy profiles can exhibit substantially different cognitive stability and confidence–accuracy alignment. In particular, confidence miscalibration is shown to predict degradation in reasoning consistency under perturbation—patterns that static test scores fail to detect. HCMS is presented as a diagnostic measurement instrument rather than a predictive model, emphasizing interpretability, reproducibility, and cognitive validity. This work contributes empirical evidence that assessment systems must move beyond static scores to faithfully represent how humans understand, not merely what they answer.
Muhammad Rayan Shahid (Fri,) studied this question.