Federated Learning (FL) enables machine learning models to be trained on decentralized data, preserving data privacy by keeping data local to clients. However, due to data heterogeneity and the distributed aggregation process, models trained using FL may differ internally from models trained in a centralized manner. Explainable Artificial Intelligence (XAI) aims to improve transparency and trust in machine learning models by providing explanations for model predictions. Despite the growing importance of both FL and XAI for trustworthy AI systems, their combined impact is not yet well understood. In particular, it is unclear whether federated training affects the quality, stability, or reliability of model explanations compared to centralized training. This thesis aims to investigate the interaction between Federated Learning and Explainable AI by quantitatively comparing explanations generated by centralized and federated models. The study will analyze how factors such as data heterogeneity, number of clients, and communication rounds influence explanation similarity and divergence. Attribution-based explanation methods such as Saliency, Input×Gradient, and DeepLIFT will be used, and explanation similarity will be evaluated using quantitative metrics such as rank correlation and explanation variance.