Confirmatory physiological tests found no significant tool×task interactions for pupil Δ or ln RMSSD, with substantial data loss limiting interpretation¶

Claim · All claims
Evidence · 1 study (2 entries) · causal r2 · q3
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q3 i? No significant tool×task interaction for pupil Δ after Holm correction (pHolm=.308). → Bergh 2026 q3 i? No significant tool×task interaction for ln RMSSD (pHolm=.853); the article says the cardiac result should be read as inconclusive. → Bergh 2026 (2)

Evidence¶

Bergh 2026¶

Bergh, C., Tag, B., Vassar, A., & Renzella, J. (2026). Your Programming Students'Cognition with ChatGPT: Higher Performance, Lower Retention, and Reduced Ownership. https://arxiv.org/abs/2609.21194 doi:10.48550/arxiv.2609.21194

q3 · i? · causal · r2

Confirmatory likelihood-ratio test on task-baseline pupil diameter delta recorded with a Tobii Pro Spectrum; only quality-gated intervals were analysed (79–87% retained per interval). The interaction was not significant; an exploratory Task 3 contrast showed a larger negative pupil Δ for AI-Assisted participants.

"There was no significant tool ×task interaction for pupil Δ (𝜒 2(2)= 4.56,𝑝=. 103,𝑝Holm =. 308) and remains insignificant after Holm correction"

Bergh 2026 (2)¶

Bergh, C., Tag, B., Vassar, A., & Renzella, J. (2026). Your Programming Students'Cognition with ChatGPT: Higher Performance, Lower Retention, and Reduced Ownership. https://arxiv.org/abs/2609.21194. doi:10.48550/arxiv.2609.21194

q3 · i? · causal · r2

Confirmatory test on heart rate variability (ln RMSSD from a Polar Verity Sense); only 25–36% of intervals survived quality gates. The article states the cardiac insignificance should be read as inconclusive rather than as evidence of no group difference.

"There was no significant tool×task interaction for ln RMSSD (𝜒 2(2)= 0.32,𝑝=. 853,𝑝Holm =. 853) and remained insignificant after Holm correction"

Discussion¶

Related Claims¶

• Self-reported mental effort increased less across tasks in the ChatGPT condition while perceived difficulty rose equally in both groups — related
• Wearable outcomes declined significantly across the academic year, including a 5.2% reduction in RMSSD and roughly 680 fewer steps per day by Week 33 — related
• Dictionary-based SEANCE features yielded 21 nominally significant within-person associations with wearable outcomes, but none survived Bonferroni or FDR correction — related
• Physiological measures have limited validity for effort, especially for distinguishing data-driven from goal-driven effort — related