Shift from offline to online learning definitively reveals age-related changes in sequential learning.

Implicit learning plays a crucial role in the effortless execution of many everyday activities. One of the most common methods used to assess implicit learning in the literature is the Serial Reaction Time Task (SRTT), which involves learning a sequence of movements. However, the developmental trajectory of sequence learning (SL) remains a topic of debate (Zwart et al., 2019). Recent studies suggest that adults primarily rely on online learning mechanisms, whereas 6-year-old children tend to show learning gains during offline periods (i.e., rest intervals). Interestingly, 10-year-olds exhibit a mixed pattern, combining both online and offline learning processes (Du et al., 2017). These findings seem to reflect age-related shifts in learning strategies, as partially supported by Du et al. (2016). Yet, some of these results were obtained using atypical versions of the SRTT, such as a foot-stepping version (Du et al., 2017) or tasks involving bimanual responses (Van Roy et al., 2024) or explicit instructions (Voisin et al., 2024). This raises the question of whether the observed developmental changes reflect genuine age-related shifts in learning strategies, or instead, age-specific sensitivities to task design. To address this issue, we replicated the study by Du et al. (2017) using a classical, touch screen SRTT specifically adapted to minimize age-related biases (Experiment 1). We also conducted a partial replication of Du et al. (2016) (Experiment 2) by manipulating instructions (explicit instead of implicit in Experiment 1). Our findings confirm age-related differences in learning dynamics, in line with previous studies (Du et al., 2017). Notably, although explicit knowledge was associated with online learning indices, providing explicit instructions did not appear to enhance online learning in children. We discussed which individual and contextual parameters could influence strategy changes reflected by online and offline learning during sequence learning.