Is your definition actually accurate enough to capture instances of intuitive judgements that "this is a conscious being" without being overly broad to include "obviously non-conscious" things?

By calling it a spectrum, the already fuzzy definition becomes so broad as to be almost useless. Your spectrum and "knowing" model admits conscious GPS, albeit "low consciousness." What about spiders or GPT-3 or suitibly chosen digits out of an RNG? At best these all fail the intuitive answer test for most people.

Perhaps you have a mental model of Yudkowsian self-optimizers? Or maybe you are thinking more classically along the lines of Kant?

The whole problem I am trying to communicate is that your are playing and refining a model without defining exactly what your are modeling.