In a dielectric material, permittivity makes sense, because it's easy enough to interpret the polarization density as the number density of molecules times their induced dipole moment. I'm basing this description off chapter 4 of Griffith's E&M textbook. In this model, the charges are all valence electrons and nuclei. What is the nature of polarization in metals? Clearly light interacts with some kind of miroscopic dipole-oscillators in metals, or else all metals would be silver-colored. It wouldn't make sense for there to be a restoring oscillator force on a conduction band electron, so I assume these dipole-oscillators are dipole moments formed between gold nuclei and valence band electrons. What is the criteria for a metal to start behaving more like a dielectric than a conductor in a given wavelength band? Intuitively, I want to say that metals start looking more like dielectrics than conductors when the restoring force on an atomic dipole is less stiff than the Ohmic drag force on conduction electrons, i.e., that the bound charges are less constrained than the conduction charges. If this is the right idea, what's the formal criteria for the threshold wavelength? Sorry if this is too open-ended. Any pointers, references, or partial answers would be greatly appreciated.