Proposed extension, drawing on Ryan and Deci’s 2020 review, Buçinca’s 2021 experiment and Anthropic’s 2026 report: retain the existing protocol and independently assign added supports by chance with comparable information and AI access. Measure task enjoyment and endorsed purpose separately; record felt autonomy, competence and connection. Test knowledge through unaided explanations and checks against cited evidence. Add AI-reliance tasks scoring acceptance of expert-checked correct versus incorrect advice, distinct from opposing-preference shifts. Record need for cognition separately from task motives. Confidence, engagement and agreement alone are insufficient. Count participation and total preparation, review, maintenance and learning effort. Motivation and judgment improving together would not prove that motivation caused better judgment; clearer instruction could help directly.