Visual statistical learning and relearning: A cohort study with children age 5 to 7 years and adults

Abstract

Tracking sequential information allows children to quickly acquire environmental statistics which is important for, e.g., language learning. Thus, it has been suggested that children should excel in implicit statistical learning and outperform adults therein. Older children and adults seem to rely more on explicit learning mechanisms. Findings on how the retention and the generalization of sequential regularities changes across development have been inconsistent, however. The employed research designs did not allow drawing conclusions, neither about long term retention, nor about transfer of sequential knowledge. The present dissertation aimed at closing this gap by investigating three child groups (5-year-olds, 6-year-olds, 7-year-olds) and two independent groups of adults, who learned visual sequences which were defined by complex rules as defined by an Artificial Grammar (AG). This age range was selected due to well-known extensive changes in cognitive functions, such as memory and cognitive control. All groups completed three learning sessions on separate days over the course of one week (Year 1). After one year, remaining AG knowledge was tested in three “relearning” sessions with the original item set (Year 2). In a subsequent session, transfer to a new visual stimulus set was tested which implemented the same AG. Seven-year-olds and one adult group served as controls for the other groups after the delay, so their study design differed slightly. They both had an additional transfer session at the end of Year 1, to test rule generalization before the delay. Their learning trajectories were discussed separately as Project 1, while learning outcomes of 5-year-olds, 6-year-olds and the second adult group were compared in Project 2. Project 1 and Project 2 both included assessments of explicit sequence knowledge and memory and German grammar skills in children and adults, which were related to AG learning for all participating groups in Project 3. We hypothesized to find an increase in learning performance across sessions in all child groups and adults for the first stimulus set in Year 1. We expected to observe preserved AG knowledge as well as transfer effects in all groups. Younger age groups were predicted to feature higher retention over one year and quicker relearning of the AG with the first stimulus set, and to show larger transfer to a second stimulus set in Year 2. Additionally, we assumed that children predominantly rely on implicit knowledge, while adults acquire more explicit knowledge about the underlying AG. We predicted that higher capacities in working memory and declarative memory retrieval, and stronger German grammar skills are associated with better AG task performance. Children from 6 years onwards successfully learned the visual AG, showing an adult-like increase in learning across three sessions while being continuously outperformed by adults. All age groups used their acquired AG knowledge after a 12-month delay for quicker relearning of the same input that enabled them to reach higher final performance levels, as compared to Year 1. Furthermore, all age groups transferred their AG knowledge to new surface features. However, relearning results did not confirm that younger children outperform older children and adults in the long run (Project 1 & Project 2). When controlling for maturational effects after the delay in 5-year-olds and 6-year-olds, both groups showed quicker re-acquisition of AG rules in the first session compared to age-matched naïve controls. The group who started at 5 years of age showed gains from prior learning when they were relearning at age 6 years, even though they had not demonstrated successful learning of the AG in the first year (Project 2). Levels of explicit AG knowledge did not differ between any of the investigated age groups (Project 1 & Project 2). Exploratory evidence was provided that memory skills (working memory & declarative memory encoding/retrieval), and to a lower degree German grammar skills, were associated with multi-session AG learning outcomes (Project 3). The present findings corroborate the idea that repeated exposure to sequential regularities results in long-lasting memory traces and the ability to generalize these regularities to a situation with new visual input. However, the current study does not provide support for superior retention of sequential regularities over a longer time period early in development, neither for younger children as compared to older children, nor for children as compared to adults.