To address the limitations of existing swarm intelligence optimization algorithms that generally rely on "individual position updates" and "instant fitness feedback," this paper proposes a novel continuous optimization paradigm based on information diffusion dynamics. Inspired by the hunting and perception behaviors of the Indian cobra, this method no longer models the optimization process as the direct movement of solution points in the search space. Instead, it treats each candidate solution as an information source, achieving global search and local convergence through the diffusion, superposition, interaction, and feedback of multi-scale information fields in the solution space. This paradigm changes the modeling assumptions of traditional optimization methods at the object level, extending "point optimization" to "field optimization." By introducing mechanisms such as multi-scale diffusion, information resonance and cancellation, toxic backlash, and time-delayed feedback, it effectively alleviates problems such as premature convergence, noise interference, and decreased efficiency in high-dimensional searches. This paper systematically presents the mathematical modeling, dynamic mechanism, and complete algorithm flow of this optimization paradigm, and analyzes its stability and search characteristics from a theoretical perspective.