Neurodegenerative diseases such as Alzheimer’s disease, specifically accompanied by posterior cortical atrophy, progressively degrade visual cognitive function. However, studying the disease mechanisms and efficacy of potential interventions in these biological systems remains challenging. This thesis bridges computational and clinical neuroscience by leveraging convolutional neural networks as in-silico models of neurodegeneration in the visual system. Through four interconnected studies, it demonstrates how deep learning frameworks can simulate disease progression, neuroplasticity, and cognitive rehabilitation, offering insights into both biological processes and artificial intelligence (AI) resilience. First, I establish that neural injury in convolutional neural networks—via iterative synaptic ablation, neuronal ablation, and weight decay—induces progressive declines in object recognition accuracy and representational structure, mirroring cognitive trajectories of posterior cortical atrophy. This degradation disproportionately impacts later network layers, analogous to the vulnerability of higher-order visual cortices in humans. Next, I improve biological plausibility by simulating neuroplasticity through retraining, which slows cognitive decline and better preserves internal activation patterns. Finally, I develop a framework to evaluate cognitive intervention strategies, showing that accuracy-guided retraining preserves task performance and representational geometry during intermediate stages of model degeneration. Geometric analyses link successful rehabilitation to efficient compression in deeper layers, proposing a mechanistic basis for cognitive maintenance. By aligning computational insights with clinical observations (e.g., layer-specific degradation parallels preserved basic vision in posterior cortical atrophy), this thesis underscores the potential of in-silico models to accelerate therapeutic discovery. Future work integrating patient-specific biomarkers and counterfactual simulations could advance precision medicine for dementia, offering a pathway to translate AI-derived insights into targeted clinical strategies.