In this paper we outline a holistic approach for understanding and simulating human decision making in knowledge-intensive tasks. To this purpose, we integrate semantic and cognitive models in a hybrid computational architecture. The contribution of the paper is twofold: first we describe a packetcentric ontology to represent network traffic. We show how the ontology is used to describe real-world network traffic and also serve as a basis for higher level ontologies of cyber operation, threat and risk. Second, we demonstrate how the combination of the packet-centric ontology with an adaptive cognitive agent with learning capabilities, can be used to understand the human defender reasoning processes when monitoring network traffic. Through simulation experiments we evaluated the proposed hybrid computational architecture and demonstrate its ability to successfully detect malicious port scanning within legitimate network traffic. We discuss the implications of these findings for improving our understanding of the cognitive processes and knowledge requirements of the cyber defender, as well as the possible use of the hybrid architecture as a cognitively inspired decision support tool.