Previous Experiences Support Learning of Novel Hierarchical Rules in Childhood

• In Event: Scaffolding the Development of Successful Learning Across Early and Middle Childhood

Abstract

During development children become increasingly able to use previous experiences to guide learning, enabling flexible, goal-directed behavior. To this end, experience with specific task structures and environments may facilitate the learning of novel similarly structured tasks. To test if children utilize previously learned hierarchical rules to master new ones, we investigated if practicing different rues with various demand on switching between them is related to improved performance on unique similarly structured tasks and which neural mechanisms might underlie this process.

8–11-year-olds practiced either mainly single tasking (SI group, N=70, 40 with MRI) or task switching (SW group, N=66, 40 with MRI) on a tablet across nine weeks. We also included a passive control group (MC group, N=37 with MRI). The practice involved tasks that were repeated every other week (15 sessions) along with unique but structurally similar tasks in the weeks in-between (12 sessions). All groups performed one of the repeating practice tasks in the lab before (session A) and after the nine weeks of practice (session D). The training groups performed two additional lab assessments after 3 (session B) and 6 (session C) weeks. Analyses were conducted using multi-level models with subject-specific intercepts and session-specific slopes.

After 3 weeks of training, children in the practice groups showed improved accuracy in the practiced repeated task when performed in the lab (b=0.07, p<.001). The SW group maintained these improvements, while the SI group showed a decline in accuracy at session D specifically on trials where rules were repeated (b=0.07, p=.008) or switched (b=0.08, p<.001), relative to blocks without any switching between rules. On the neural level, the dorsolateral prefrontal cortex and the superior parietal lobe showed practice-related activation reductions in the SW group (Figure 1). These results suggest that repeatedly practicing the same task improved performance due to more efficient task-rule management with experience in task switching.

To investigate whether repeated practice facilitated performance on novel tasks with a similar structure, we tested for overall performance differences between the SI and SW groups in the structurally similar unique tasks they encountered during practice. Additionally, we tested whether performance on unique tasks completed towards the end of the study (after multiple practice sessions with the repeated tasks) was higher than performance on unique tasks at the beginning of the study. Overall, structurally similar unique tasks performed towards the end of the study were associated with higher performance than tasks at the beginning, b=0.03, p=.001. The SW group showed generally higher performance on unique tasks, b=0.12, p=.001. While this advantage of the SW group was more pronounced at the beginning of the study, the SI group caught up towards the study end (b=0.03, p=.01).

Taken together, already a few practice sessions including repeated experience with intensive task switching facilitated performance on novel similarly structured tasks. While children who learned rules mainly separately with a smaller dosage of switching demands could use this experience in the context of a novel task as well, they needed prolonged rule experience to facilitate novel task performance.

Authors

• Sina Schwarze, Max Planck Institute for Human Development
  Presenting Author

• Yana Fandakova, University of Trier