Retrieval-Induced Facilitation in Real-World Scenarios

点击翻译按钮获取中文摘要

Project Summary Episodic memory, or the retrieval of information from events, is a complex process that is dynamic in nature. When retrieving this information, there are changes that occur. These changes can either enhance (i.e. retrieval-induced facilitation [RIFA]) or inhibit (i.e. retrieval-induced forgetting) memories for other, related information, and the mechanism of this enhancement/inhibition is unclear. In previous research, these concepts have been explored in studies using static images, word pairs, or literary passages to test these phenomena. In real life, we receive a constant, continuous stream of information that we segment into events with distinct boundaries. There is good reason to believe that retrieving one part of an event might lead to RIFA for other parts of the same event, but that event boundaries might restrict the benefits of retrieval practice. I propose to use an innovative behavioral paradigm in combination with computational modeling to address how retrieval impacts memory for naturalistic events. I have developed a task in which participants are exposed to videos depicting relatable events (i.e. going grocery shopping) that are separated by event boundaries. Aim 1 is to understand the relationship between RIFA and event segmentation. Based on prior research indicating a connection between memory retrieval and event segmentation, it is proposed that retrieval practice will facilitate retention of other information from the same event, but RIFA will not occur between events. In Aim 2, using computational modeling, we propose a possible mechanism of RIFA. Based on prior models, such as the Complementary Learning Systems (CLS) model, this mechanism likely involves both error-driven and Hebbian learning, with emphasis on the connection between the entorhinal cortex, the hippocampus, and the neocortex. In the proposed study, we will expand upon previous models on RIFA to better characterize the relationship between these regions in event-based, realistic scenarios. We expect that with retrieval practice the hippocampus will become more efficient at driving learning in the neocortex, which will ultimately lead to better accuracy for information within the same event as the retrieval practiced information due to a shared event context. Starting with a simple context change to mimic a change in event, we will increasingly make the model more complex by training the model to use prediction errors to mark event boundaries, then ultimately, inputting naturalistic videos to best mimic a real-life scenario. This project builds on my previous work and addressing these aims will give me new training in the most advanced methods in cognitive neuroscience, specifically computational modeling. More broadly, we have shown that memory retrieval practice can be a clinically valuable approach in patients with schizophrenia (Ragland et al, 2020), and these studies will give us the information we need to optimize this method to enhance everyday episodic memory in psychiatric populations.