Working memory, a cornerstone of human cognition, allows us to temporarily hold and manipulate information. The delayed response experiment is a fundamental tool for understanding working memory. By inserting a delay between the presentation of stimuli and the behavioral decision, researchers can probe the neural mechanisms underlying information maintenance and manipulation during this delay period. This review delves into the delayed response paradigm and introduces attractor dynamics as a potential mechanism for working memory function. We then systematically explore two common methods for analyzing working memory networks: trajectory tracking, and energy landscape and flux theory. Finally, we summarize the key features of these two dynamical approaches and provide an outlook for future directions of working memory research.