This paper presents a behavioral-based controllers design, utilizing evolutionary learning of trajectories, applied to the IFAC PID18 benchmark model for a vapor compression refrigeration system. The challenge of tracing both the evaporator outlet temperature (Te,sec,out) and superheat temperature (Tsh) with disturbances and imposed restrictions is addressed. The challenge integrates a preset discrete MIMO control scheme, serving as a basis for comparison with alternative control schemes. The proposed method also allows for direct experimentation on the system, bypassing the need for reference models or mathematical representations. By developing behavioral paths and iteratively adjusting controller variables, satisfactory control objectives are achieved. The evolutionary behavioral approach is tested on the default discrete control scheme and then applied to newly designed coupled continuous PID controllers, which outperform the benchmark strategy. The method is versatile, adaptable to both model-based and data-driven approaches, and offers a direct relationship with the physical system, independent of both its representation and controller structure, thus emulating real-world conditions effectively.