Performance improvements during early motor skill learning occur predominantly during short rest periods (micro-offline gains), replicated in a large crowdsourced sample and in a laboratory study¶

Claim · All claims
Evidence · 2 studies · 2 causal r2 · q3 · i3 large
Currency · newest study 2026 · checked against its sources 2026-10-26 (automated, 2 of 2 entries) · not reviewed by a person
Design limits · Marlene Bönstrup 2020: no comparison group (read from the study by two models, 2026-10-26)

Subclaims¶

q3 i? In early learning (trials 1–5), total early learning was accounted for by performance improvements during rest periods, while micro-online changes were nil, in both crowdsourced (N = 389) and in-lab (N = 27) groups. → Marlene Bönstrup 2020

q3 i3 In a no-cue motor sequence task, cumulative and micro-offline learning gains were significantly positive with large effect sizes, while micro-online learning did not contribute positively. → Kawasoe 2026

Evidence¶

Marlene Bönstrup 2020¶

Marlene Bönstrup, Iñaki Iturrate, Martin N. Hebart, Nitzan Censor and Leonardo G. Cohen. (2020). Mechanisms of offline motor learning at a microscale of seconds in large-scale crowdsourced data. npj Science of Learning. https://doi.org/10.1038/s41539-020-0066-9

q3 · i? · causal · r2

Crowdsourced replication (N = 389) of the sequential finger tapping task on Amazon Mechanical Turk, 36 trials of 10 s practice and 10 s rest. Micro-offline gains were "substantial" (crowdsourced 1.24 ± 0.12 keypresses/s, P < 0.001; in-lab 1.68 ± 0.32, P < 0.001) while micro-online changes were nil (P = 0.72; P = 0.48).

"This early learning was accounted for by performance improvements during rest periods rather than during practice periods. On average, micro-online changes were nil in both groups"

Kawasoe 2026¶

Kawasoe, R., Matsumura, K., Shinohara, T., Arima, K., Takeo, Y., Ikeda, T. & Sugata, H. (2026). Wakeful targeted memory reactivation during short rest periods modulates early motor learning. npj Science of Learning. https://doi.org/10.1038/s41539-026-00407-9

q3 · i3 · causal · r2

One-sample t-tests in the TMRno group of a laboratory experiment (17 of the 69 participants randomised to four groups), run to reproduce prior micro-offline learning findings with a 10-s practice/10-s rest key-press task. The article reports "Cohen’s d = 2.63" for cumulative and "Cohen’s d = 1.75" for micro-offline learning; micro-online gains were negative.

"A one-sample t-test showed significant learning gains for cumulative learning and micro-offline learning (cumulative learning: t = 10.85, p < 0.001, Cohen’s d = 2.63; micro-offline learning: t = 7.22, p < 0.001, Cohen’s d = 1.75), but not for micro-online learning"

Discussion¶

Learner Variables¶

• Long-Term Memory — outcome: instruction changes it

Related Claims¶

• Random termination of practice periods does not reduce micro-offline gains, ruling out preemptive motor slowing — related
• Halving practice-period duration from 10 s to 5 s leaves micro-offline gains unchanged, arguing against recovery from motor fatigue — related
• Micro-offline gains are inconsistent with a dominant contribution of latent motor slowing, supporting consolidation as their driver — related
• Lower resting GABAB-mediated inhibition in M1 (measured as LICI) is strongly negatively associated with micro-offline motor consolidation gains during early learning — related
• Retroactive interference immediately after practice reduces the learning rate relative to interference after 10 s of rest, indicating within-seconds memory stabilization — related
• Auditory TMR cues replayed 1.3x faster than the preceding practice speed yield greater early-phase motor learning gains than same-speed cues — related