Fingerprint presentation attack detection (FPAD) has been a pivotal part of biometric security systems that aim to distinguish genuine fingerprints from spoofed ones. Existing FPADs have achieved reasonable performance; however, they exhibit limited interpretability and generalization, especially when encountering unseen attacks. To address these limitations, we propose an explainable fingerprint presentation attack detector framework that incorporates a MobileNet-inspired feature extractor and a customized classification head along with explainable analysis. Efficient feature extraction is ensured using the MobileNet feature extractor, thereby reducing computational cost. The classification head refines the feature representation, thus improving the accuracy and explainability. A feature extractor built upon multiple convolutional layers, batch normalization, and ReLU, followed by fully connected layers, including dropout and dense layers, for final classification. Incorporating explainability into the proposed method offers transparency in decision-making and enables insight into the most responsible features for detecting spoofing attempts. Extensive evaluations on the standard and diverse LivDet 2011 benchmark dataset show that the proposed method achieves high accuracy while providing intuitive explanations of its predictions.