Offline Consolidation of Implicit Motor Sequence Learning: Constraints of Initial Learning Effects and Sleep-Dependent Improvement

From fundamental activities like dressing and walking to more complex skills such as painting and dancing, many daily human behaviors rely extensively on motor sequence knowledge, which is largely acquired through implicit learning. Therefore, enhancing the implicit learning effect of motor sequences is of great importance. This involves not only how to improve initial learning performance, but also, crucially, whether individuals can achieve further consolidation and enhancement during the offline period after the initial learning effect approaches its upper limit. However, the relationship between the initial implicit learning effect of motor sequences and offline consolidation remains unclear: Does a higher initial learning effect lead to a better offline consolidation effect? Or, due to the constraints of learning saturation, does a higher initial learning effect instead result in diminished offline consolidation gains? If the latter is true, can the offline consolidation advantage conferred by sleep help individuals overcome the limitations of learning saturation, allowing them to benefit from offline consolidation even when the initial learning effect is high? To address these questions, the present study employed a modified serial reaction time task, eliminating interference from perceptual information, and conducted three experiments to investigate these issues, aiming to clarify the relationship between the learning effect and offline consolidation.This study conducted three experiments using a completely between-subjects design. All participants visited the laboratory twice: on the first visit, they completed the first phase of a serial reaction time task; on the second visit, they completed the second phase of the serial reaction time task and a prediction task assessing conscious awareness of the acquired knowledge. Experiment 1a manipulated learning duration to induce high and low initial learning effects, examining offline consolidation performance after 12 hours of waking intervals. Specifically, we investigated whether offline gains would emerge under low learning conditions and whether offline consolidation benefits would diminish under high learning conditions due to learning saturation. Experiment 1b further incorporated intermediate intervals of 4 and 8 hours for the high learning effect group to exclude the possibility that offline gains were obscured by shorter offline windows, while also validating the robustness of Experiment 1a findings. Experiment 2 introduced sleep as a variable under high learning effect conditions, investigating whether sleep-dependent consolidation advantages could overcome learning saturation constraints, enabling individuals with high initial learning effects to still benefit from offline consolidation. We also examined whether nocturnal sleep and daytime napping produced comparable effects.Results revealed that offline consolidation of pure motor sequence learning was significantly modulated by initial learning effect levels: the low initial learning effect group exhibited offline gains following 12 hours of waking intervals, whereas the high initial learning effect group showed only offline stability without gains (Experiment 1a). When offline intervals were shortened to 4 and 8 hours, the high initial learning effect group maintained the same pattern, ruling out the hypothesis that their offline improvements might be concealed within a shorter offline time window (Experiment 1b). Upon introducing sleep during the offline period, individuals with high initial learning effects overcame the learning saturation ceiling and demonstrated sleep-dependent offline gains, with nocturnal sleep and napping producing similar effects (Experiment 2).In summary, this study reveals the modulatory role of initial learning effects on offline consolidation, clarifies the unique function of sleep in transcending learning saturation constraints, and provides empirical support for synergistic optimization of motor sequence learning performance from both the “initial learning” and “post-learning offline consolidation” dimensions.