Rail complexity affects retention and transfer in opposite directions: retention gains increase with complexity while transfer gains decrease, with no inverted-U¶

Claim · All claims
Evidence · 1 study (3 entries) · causal r1 · q2 · i3 large
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i3 Transfer performance showed a significant main effect of complexity level (η²partial = 0.300), strongest at the lowest complexity. → Yinqi Huang 2026 q2 i3 Retention improvement showed a significant main effect of complexity level (η²partial = 0.372), largest at the highest complexity. → Yinqi Huang 2026 (2) q2 i? The hypothesized inverted U-shaped complexity-learning relationship was not observed. → Yinqi Huang 2026 (3)

Evidence¶

Yinqi Huang 2026¶

Yinqi Huang, Yilin Cai, Mengyao Li, Yue Chen & Robert C. Wilson. (2026). Human learning dynamics during teleoperation of surgical robots. npj Science of Learning. https://doi.org/10.1038/s41539-026-00442-6

q2 · i3 · causal · r1

LME on absolute transfer PCA scores across three complexity levels with modest per-condition samples (n = 4 per condition as printed); the printed effect size was η²partial = 0.300 (f = 0.655), which the article calls "a large effect sizedespite the modest sample size".

"The LME revealed a significant main effect of complexity level, Fð2; 18Þ¼ 3:858; p ¼ 0:040; η2 partial ¼ 0:300; f ¼ 0:655;"

Yinqi Huang 2026 (2)¶

Yinqi Huang, Yilin Cai, Mengyao Li, Yue Chen & Robert C. Wilson. (2026). Human learning dynamics during teleoperation of surgical robots. npj Science of Learning. https://doi.org/10.1038/s41539-026-00442-6

q2 · i3 · causal · r1

LME on retention PCA improvements across complexity levels; the printed effect size was η²partial = 0.372 (f = 0.770), described as "a large effect size". Retention improvement increased with complexity (Level 3 M = 11.74 vs Level 1 M = 3.32).

"Similarly for retention, we also observed a signi ficant main effect of Complexity Level, Fð2; 18Þ¼ 5:331; p ¼ 0:015; η2 partial ¼ 0:372; f ¼ 0:770:"

Yinqi Huang 2026 (3)¶

Yinqi Huang, Yilin Cai, Mengyao Li, Yue Chen & Robert C. Wilson. (2026). Human learning dynamics during teleoperation of surgical robots. npj Science of Learning. https://doi.org/10.1038/s41539-026-00442-6

q2 · i? · causal · r1

Authors' statement of the retained-versus-transferred dissociation: transfer decreased with complexity while retention increased, contradicting the Challenge Point-based inverted-U hypothesis.

"Together, these results do not support Hypothesis 1 as originally stated. Specifically, we did not observe an inverted U relationship between task complexity and post-training performance."

Discussion¶

Learner Variables¶

• Long-Term Memory — outcome: instruction changes it

Related Claims¶

• In one thermodynamics study, retention and transfer scores tended to be lower on topics of higher element interactivity, a descriptive pattern with several sessions that did not fit it — related
• Delayed post-learning assessments of ICL, ECL, and GCL do not differ between increasing and decreasing complexity sequences — related
• Human trainees show no clear asymmetric transfer of motion scaling ratio, and scaling ratio effects are small relative to complexity effects — related
• In LQG transfer simulations, complexity affects all metrics while scaling-ratio effects appear only in efficiency metrics, with a rotational-path-length Complexity × Ratio interaction — related