robot could certainly be programmed to describe specific processes like “seeing yellow” when it detects certain wavelengths of light, or even to talk about “feeling angry” under defined circumstances, without actually consciously seeing or feeling anything. But it seems impossible for a system to make a distinction between a conscious and unconscious experience in general without having an actual conscious experience as a reference point.
