Conceptual metaphor is widely recognized as a fundamental cognitive mechanism through which humans organize and construct meaning based on embodied and cultural experiences. With the rapid advancement of generative artificial intelligence, Large Language Models (LLMs) have demonstrated remarkable capabilities in processing natural language, including the interpretation of metaphorical expressions. However, few studies have systematically compared metaphor comprehension between LLMs and learners of Vietnamese from the perspective of Cognitive Linguistics. This study investigates and compares how LLMs and learners of Vietnamese as a foreign language interpret common Vietnamese conceptual metaphors. A mixed-methods approach is employed, combining quantitative analyses of interpretation accuracy with qualitative analyses of cognitive reasoning strategies. The dataset consists of conceptual metaphorical expressions collected from Vietnamese literary works, journalistic texts, and everyday communication. Responses generated by LLMs are compared with those produced by learners at different proficiency levels. The expected findings suggest that while LLMs are generally capable of identifying the generalized meanings of many metaphorical expressions, they remain limited in representing culturally grounded knowledge and embodied experiences. In contrast, human learners rely more heavily on contextual understanding, linguistic experience, and cultural cognition in constructing metaphorical meaning. Based on these findings, the study proposes a comparative cognitive model that distinguishes the probabilistic reasoning mechanisms of LLMs from the embodied cognitive processes of human learners in interpreting Vietnamese conceptual metaphors. The research contributes to the intersection of Cognitive Linguistics, Artificial Intelligence, and Vietnamese language education by providing new insights into metaphor comprehension across human and artificial cognitive systems.