The Neural Bases of Object Memory: Neuromodulation and Hemodynamic Modeling in Mild Cognitive Impairment

Abstract

Successful memory for object images is shaped by a variety of factors, including the semantic features of those stimuli and the fidelity with which the features are represented in the brain. This dissertation investigates the role of stimulus representations in object memory across the adult lifespan, combining behavioral, neuroimaging, and neuromodulation approaches. It advances our understanding of how semantic representations support object memory, how neuromodulation can leverage these representations to offset episodic memory decline, and how representational modeling methods must be adapted for healthy aging and neurodegeneration. Chapter 2 demonstrates how the memorability of natural object images is supported by semantic dimensions derived from image features, and that this relationship is reflected in neural activity across key mnemonic regions including hippocampus and rhinal cortex in younger adults. Chapter 3 turns to older adults with amnestic Mild Cognitive Impairment (MCI) to explore the mechanisms by which parietal Transcranial Magnetic Stimulation (TMS) improves episodic memory. Representational Similarity Analysis (RSA) with large language model-derived semantic embeddings provides evidence that TMS enhances semantic representational strength in inferior parietal lobule while promoting representational differentiation in hippocampus, supporting memory success. Chapter 4 presents evidence that data quality issues arise due to the insufficiency of certain neuroimaging methods for modeling hemodynamics in older adults, and due to sensitivity to RSA modeling parameters. We then take steps to improve the reliability of RSA by offering a more physiologically-grounded framework that pays greater attention to model validation. Overall, these findings make a novel contribution to the understanding of memory representations and to the modeling techniques that can be used to characterize these dynamics in aging and MCI populations.