Eliciting confidence ratings impairs rule-based category learning compared to both no-rating and judgment-of-learning conditions on ambiguous transfer items¶

Claim · All claims
Evidence · 1 study · causal r3 · q3
Currency · newest study 2025 · checked against its sources 2026-10-20 (automated, 1 of 1 entry) · not reviewed by a person

Subclaims¶

q3 i? In a pre-registered experiment (N = 710), the confidence rating group performed significantly worse on ambiguous transfer items than both the JOL group and the no-rating group. → Kit S. Double 2025

Evidence¶

Kit S. Double 2025¶

Kit S. Double, Micah B. Goldwater, Damian P. Birney. (2025). Reactivity to confidence ratings: evidence of impaired rule-learning. Metacognition and Learning. https://doi.org/10.1007/s11409-025-09413-5

q3 · i? · causal · r3

Pre-registered mixed ANOVA on the full sample (N = 710; 235 confidence rating, 251 no-rating, 224 JOL) of a 3 (condition) × 3 (item type) design. The interaction was significant, and "Pairwise comparisons indicated that the only significant differences were for the ambiguous transfer items, where the confidence rating group performed significantly worse than both the JOL group and the no-rating group."

"There was also a main effect of condition, F(2,707) = 3.41, p =.034, ηp 2 = 0.010 and a significant interaction between condi- tion and item type, F(4,1414) = 2.63, p =.033, ηp 2 = 0.007, see Table 3 for group differences and pairwise comparisons."

Discussion¶

Related Claims¶

• High levels of induced social presence during non-interactive teaching can impair learning relative to restudy (a double-edged sword) — related
• Rule learners report higher confidence in their described rules than non-learners, but confidence does not differ within flexibility groups — related
• The confidence-rating impairment on ambiguous transfer items persists among participants who met the learning criterion — a narrower finding that bears on this claim
• Confidence ratings are unlikely to be an effective educational tool when knowledge transfer is the goal — a broader claim this one bears on
• Reactivity differs by rating scope: item-focused confidence ratings impaired rule learning while task-focused judgments of learning did not — related
• Confidence ratings do not impede category learning per se: no condition differences in training accuracy or on unambiguous and trained transfer items — related
• Response-time effects are limited to slower responding in the JOL condition during the first two training blocks — related