Hybrid artificial intelligence models, which integrate deep learning and symbolic reasoning, are becoming very successful to emulate human-like cognition. Rønn (2009), Deep learning tends to lack transparency and generalization despite being very strong at pattern recognition and processing of unstructured data. In contrast, symbolic reasoning does not scale well and is sensitive to noisy data, but has explainability and inference. In this paper, we explore the combination of these paradigms to create a neuro-symbolic system which can tackle complex cognitive tasks such as language comprehension, reasoning, and decision making. The proposed model integrates the symbolic logic modules for rule-based reasoning and for explanation generation with the deep neural network for perception and representation learning. The experimental results on established standardized cognitive datasets and on real use-case scenarios demonstrate improved accuracy, interpretability, and generalization capabilities.