Metamemory or just Memory? Searching for the Neural Correlates of Judgments of Learning

Abstract

Judgments of Learning (JOLs) are judgments of the likelihood of remembering recently studied material on a future test. Although JOLs have been extensively studied, particularly due to their important applications in education, relatively little is known about the cognitive and neural processes supporting JOLs and how these processes relate to actual memory processing. Direct access theories describe JOLs as outputs following direct readings of memory traces and hence predict that JOLs cannot be distinguished from objective memory encoding operations. Inferential theories, by contrast, claim JOLs are products of the evaluation of a number of cues, perceived by learners to carry predictive value. This alternative account argues that JOLs are made on the basis of multiple underlying processes, which do not necessarily overlap with memory encoding. In this thesis, the neural and cognitive bases of JOLs were examined in a series of four ERP experiments.

Across experiments the study phase ERP data showed that JOLs produce neural activity that is partly overlapping with, but also partly distinct from, the activity that predicts successful memory encoding. Furthermore, the neural correlates of successful memory encoding appear sensitive to the requirements to make a JOL, emphasising the close interaction between subjective and objective measures of memory encoding. Finally, the neural correlates of both JOLs and successful memory encoding were found to vary depending on the nature of the stimulus materials, suggesting that both phenomena are supported by multiple cognitive and neural systems.

Although the primary focus was on the study phase ERP data, the thesis also contains two additional chapters reporting the ERP data acquired during the test phases of three of the original experiments. These data, which examined the relative engagements of retrieval processes for low and high JOL items, suggest that encoding processes specifically resulting in later recollection (as opposed to familiarity) form one reliable basis for making JOLs.

Overall, the evidence collected in this series of ERP experiments suggests that JOLs are not pure products of objective memory processes, as suggested by direct access theories, but are supported by neural systems that are at least partly distinct from those supporting successful memory encoding. These observations are compatible with inferential theories claiming that JOLs are supported by multiple processes that can be differentially engaged across stimulus contents.