That said, I can further observe that I seem to differentially value various nonhuman species.

Simple speciesism is a step in the direction of capturing that, but it ends up with a list of (species, value) ordered pairs, which is a very clunky way of capturing the information and not very useful for predictive purposes.

OTOH, if I analyze that list for attributes that correlate with high value, I may end up with a list of attributes that I seem to value in isolation (then again, I might not). For example, it might turn out that I value fluffy animals, and social ones, and ones with hands, and ones with faces, and various other things.

If I do this analysis well enough, I might be able to predict how much I would value a novel species based on nothing but an evaluation of this species on those terms ("oh, scale-backed lemoriffs are spiny, asocial, lack hands and faces? I probably won't value a scale-backed lemosaur very much."). Then again, I might discover that there were parameters I hadn't taken into consideration in my analysis, and that when faced with the actual species my value judgment might be completely different because of that. ("wait, you didn't tell me that scale-backed lemoriffs are also about as smart as humans and that 10% of Internet users I enjoy interacting with were in fact scale-backed lemoriffs... crap. Now I wish we hadn't eradicated them. I'll add 'intelligent' to the list next time.")