UC San Diego

Environmental cues frequently elicit the retrieval of items stored in long term memory. In many cases, cued retrieval serves a valuable function, providing important reminders and connecting us to our past. At times, however, cues can elicit the retrieval of unwanted memories. A core question in cognitive psychology is how individuals are able to prevent these unwanted memory intrusions. Neuroimaging studies suggest that “canceling” retrieval requires a right prefrontal (PFC) network that is also recruited when we cancel actions. Our ability to study how this network exerts control over retrieval, however, is limited in several ways: 1) current studies use methods with poor temporal resolution that are unable to determine when this network interrupts retrieval, and 2) measuring memory intrusions relies largely on self-report rather than objective measures. This dissertation aims to address these limitations. In Chapter 1, we address the first limitation by validating an EEG marker of the right PFC network. EEG has high temporal resolution, making it a valuable tool for testing the timing of network recruitment. In the two subsequent chapters, we address the second limitation by developing methods to objectively track the emergence of memories during efforts to suppress. We first show that an event related potential, known to track the amount of retrieved information, is reduced when trying to prevent retrieval (Chapter 2). We next used EEG-based decoding and tracked the specific memory rather than the quantity of retrieved information to more accurately determine whether participants were successful in avoiding retrieval (Chapter 3). Although we could still decode the prohibited memory, decoding accuracy was reduced later during retrieval, which may reflect efforts to “suppress” the intruding content. Lastly, by combining these tools, we found that early recruitment of the right PFC network corresponded to reduced decoding and thus, measureable changes in the prohibited memory representation (Chapter 4). Together these data may warrant a two-phase model of control that involves both early and late control processes. Overall, this dissertation contributes necessary tools for gaining a deeper understanding of the right PFC network and our ability to control when past experiences gain access to working memory.