The effort-based reward reduced how often participants used the assistant, whereas metacognitive feedback changed how they used it¶

Claim · All claims
Evidence · 1 study (2 entries) · causal r3 · q2
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i? The reward reduced the odds of requesting any form of assistance on a given item (OR = 0.39), while the reduction under metacognitive feedback was smaller and not significant (OR = 0.60, p = .052). → Sebastian Maier 2026 q2 i? Answer offloading accounted for 51.2% of requests under feedback versus 62.8% under the reward, suggesting feedback changed the composition of use while the reward mainly reduced frequency. → Sebastian Maier 2026 (2)

Evidence¶

Sebastian Maier 2026¶

Sebastian Maier, Kai Schwabe, Manuel Schneider, and Stefan Feuerriegel. (2026). Designing Against Deskilling: Metacognitive Feedback Reduces Cognitive Offloading to LLM Assistants. arXiv. https://arxiv.org/abs/2609.20143 doi:10.48550/arxiv.2609.20143

q2 · i? · causal · r3

Item-level mixed-effects logistic regression on likelihood of any LLM use per learning item (Appendix I.1, Table 9). Guided-help requests showed no significant effect of either intervention (reward OR = 0.69, p = .060; feedback OR = 0.92, p = .674).

"The reward condition reduced the odds of requesting any form of assistance on a given item ( 𝑂𝑅= 0.39, 95% CI[0.23, 0.66],𝑝<. 001), whereas the reduction under metacognitive feedback was smaller and was not significant ( 𝑂𝑅= 0.60, 95% CI [0.36, 1.00],𝑝=. 052)"

Sebastian Maier 2026 (2)¶

Sebastian Maier, Kai Schwabe, Manuel Schneider, and Stefan Feuerriegel. (2026). Designing Against Deskilling: Metacognitive Feedback Reduces Cognitive Offloading to LLM Assistants. arXiv. https://arxiv.org/abs/2609.20143

q2 · i? · causal · r3

Descriptive comparison of classified request types across conditions (Figure 5, Section 5.4). The article states this descriptive pattern suggests feedback changed how the assistant was used, whereas the reward mainly reduced how often it was used.

"In the metacognitive feedback condition, answer offloading accounted for51.2%of items, alongside larger shares of requests for guided help and intermediate results. For the reward condition, participants used the LLM assistant less overall, but answer offloading still accounted for62 .8%of the items"

Discussion¶

Related Claims¶

• Metacognitive feedback disrupted the tendency to repeat answer offloading across consecutive items without changing whether participants engaged with the assistant — related
• In one preregistered experiment, metacognitive feedback on learners' own LLM use lowered the odds of answer offloading and raised the odds of a correct unaided answer on one-sided tests, with 95% intervals reaching no effect — related
• An effort-based reward did not significantly reduce answer offloading or improve unaided performance — related