Modern cyber ranges are primarily designed for attack simulation, information system security assessment, and cybersecurity training. However, they provide only limited support for decisionmaking related to selecting optimal incident response strategies under the conditions of real network infrastructures and rapidly evolving cyber threats. This paper proposes the concept of a cognitive cyber range that integrates a digital representation of the network infrastructure, semantic knowledge networks, Retrieval-Augmented Generation (RAG), large language models, and multi-agent analysis into a unified decision support platform. The proposed approach is based on the construction of a directed semantic network that integrates knowledge about information system assets, vulnerabilities, attack techniques, security controls, regulatory requirements, personnel competencies, and other domain-specific entities. Unlike conventional scenario discovery methods, a bidirectional subnetwork construction algorithm is proposed that independently performs forward expansion from the current system state and backward expansion from a specified security objective. After the generated subnetworks are aligned and merged, a unified causal model is constructed, from which alternative scenarios for achieving the specified objective are automatically extracted. To evaluate the generated scenarios, a multi-agent framework of virtual experts based on large language models is employed. The framework performs multi-criteria analysis according to technical, organizational, and regulatory criteria, generates explanations for the obtained recommendations, and determines their level of justification. The proposed approach combines structured knowledge, semantic analysis, and generative artificial intelligence within a unified cognitive decision support environment suitable for both cybersecurity education and practical deployment in cyber defense systems. Keywords: Cognitive Cyber Range, Semantic Network, Large Language Models, Retrievalaugmented Generation, Multi-Agent System, Decision Support, Cybersecurity, Response Scenarios, Causal Analysis, Directed Graphs, Semantic Search.