2025 Theses Doctoral

How prior knowledge scaffolds memory - a “kind of naturalistic” approach

Huang, Jiawen

In our daily life, we constantly use our structured knowledge built from repeated experiences, often referred to as schemas, to provide a scaffold for perceiving, understanding, and remembering what happens around us. This is a difficult process to study, because these kinds of prior knowledge we use in daily life are typically complex and could take a long time to develop. In this dissertation, I present findings from a few experiments that take a “kind of naturalistic” approach, which allows me to capture some complexity of such knowledge, yet is still feasible in a lab setting.

In Chapter 1-3, I present behavioral and neuroimaging findings from a board game paradigm. In Chapter 1, I show that the development of knowledge in the board game increased predictive eye-movements during sequence encoding, which in turn improves memory. In Chapter 2, I show that there are two potentially distinct processes when people remember something schema-consistent: probability given the context and prediction accuracy. I show that both contribute to better memory, but through different mechanisms. In Chapter 3, I use functional magnetic resonance imaging (fMRI) to show that different default mode network (DMN) regions respond to these two processes.

In addition, I show differences in how the brain makes memory- vs. schema-based predictions. In Chapter 4, I use fMRI to look at people using the method of loci (MoL) to encode a list of words. I found that using MoL creates conjunctive representations in DMN that are more than the sum of its parts. Overall, the dissertation highlights the importance of prediction in the relationship between schemas and memory, and the importance of DMN in this process.