Does Testing-Potentiate New Learning Because of Semantic Processing and Apply to Educationally Relevant Materials?

Test-potentiated new learning (TPNL) is the finding that retrieving information from memory via a test or quiz can enhance subsequent learning. Recent research and a meta-analysis have highlighted that TPNL is a robust effect in which the new learning of many different types of materials is enhanced. However, little is known about the mechanisms underlying this effect and how it applies to complex, educationally relevant information. A growing body of research has suggested that semantic processing of the new information may play a significant role. Two such accounts were evaluated in this thesis. The activation facilitation account, which argues that TPNL is produced by tests facilitating enhanced original learning which becomes activated during new learning. This activation leads to semantic integration and increased comprehension. Alternatively, the conceptual strategy change account argues that TPNL is produced by tests facilitating a strategy change in which participants use greater conceptual organisation, facilitating enhanced encoding and retrieval. In general, very little evidence was found supporting these arguments, and more generally, the role of semantic processing in TPNL was challenged. Furthermore, a meta-analysis of TPNL in complex, educationally relevant information found that TPNL was both large and reliable (g = 0.71, 95% CI [0.60, 0.81], k = 63), and that the benefit is robust across many manipulations of complexity. Furthermore, the meta-analysis revealed that the format of the initial test is important, with tests requiring relational processing or lengthier responses providing the greatest benefit