From February 2017 until March 2018, about 200 children between the age of 6 and 11 years participated in the project. Children were tested during three one-on-one sessions in a mobile lab brought to their schools.
We used four complementary methods to assess strategy use:
1) picture-supported self-reports (asking children to indicate a strategy after each trial of some of the memory tasks)
2) think-aloud protocols (asking children to verbalize all thoughts during some of the memory tasks)
3) self-presentation times of memory items (as certain time patterns should be related to certain strategies)
4) systematic observations of overt strategic behaviour.
Figure 1: Illustrations used to first explain memory strategies and to then ask for strategy self-reports after each memory trial in the self-report condition
[Insert attached picture here]
The main results so far are that analyses of self-reports, think-aloud protocols and self-presentation times lead to convergent conclusions. This means that we can trust the results as reflecting strategy development.
Most children of all ages did use various strategies instead of relying on just one. In the youngest group, just listening to the words, which is not a particularly sophisticated strategy, was the most prevalent approach. Relying on listening decreased with age, while cumulative rehearsal increased. Additionally, between the age of 8 to 11 years, children got better at adapting their strategy choice to task difficulty. Older children were more likely to use sophisticated, but more time-consuming cumulative rehearsal for harder tasks with more words to remember, while using listening or single rehearsal for easier tasks that they could remember with these faster strategies.
Work with adults and cognitive modelling of memory processes done by other labs (not this project) calls into questions whether cumulative rehearsal really improves recall. In contrast, our results indicate that children mostly benefit from choosing cumulative rehearsal.
The convergent results from self-reports, think-aloud protocols and self-presentation times analysed at a group level indicate that many children can report their cognitive strategies in a sufficiently accurate manner. So metacognitive accuracy seems to be good in general. However, these analyses and results cannot tell us how well metacognitive accuracy is developed in each child. We have therefore developed an innovative approach to tackle this problem by designing a cognitive Bayesian model that can integrate the information from picture-supported self-reports, self-presentation times and systematic observations and makes predictions about children’s actual, non-observable strategy use and their metacognitive accuracy. We have to test this model further before reporting results and running further analyses on metacognitive development.
