Cognitive Sovereignty: A Theory and Initial Validation of Human Autonomy in Algorithmic Environments

Artificial intelligence increasingly mediates human experience, raising questions about cognitive autonomy. Existing constructs capture aspects of this problem but lack integrative theoretical framing and empirical operationalization. We introduce cognitive sovereignty as a multidimensional construct comprising decision autonomy, critical literacy, algorithmic awareness, digital resilience, and trust calibration. Across five studies (total N = 9,829), we developed and validated the Cognitive Sovereignty Scale (CSS). Study 1 used expert review (n = 15) and cognitive interviewing (n = 45) for content validation. Study 2 (n = 650) established factor structure through exploratory and confirmatory factor analysis. Study 3 (n = 712) tested convergent and discriminant validity with established measures. Study 4 (n = 8,427 across 12 countries) examined cross-national measurement invariance. Study 5 (n = 1,200) tested predictive utility in an experimental task with algorithmic recommendations. Confirmatory factor analysis supported the five-factor structure (CFI = 0.94, RMSEA = 0.05). The CSS demonstrated convergent validity with need for cognition (r = 0.46), critical thinking disposition (r = 0.51), and self-efficacy (r = 0.42), and discriminant validity from digital literacy (r = 0.28) and technology acceptance (r = 0.31). Cross-national measurement invariance was supported (configural, metric, partial scalar). CSS scores predicted resistance to algorithmic recommendations (β = 0.34, 95% CI [0.29, 0.39]), with autonomy as the strongest predictor (β = 0.28). CSS predicted incremental variance over need for cognition and critical thinking disposition (ΔR² = 0.08, cumulative R² = 0.31, p < .001). Cognitive sovereignty provides an integrative framework for research on human–AI interaction. The CSS offers a psychometrically sound instrument for measuring this construct across diverse populations. Independent external validation is needed.