Cognitive bias doesn't exist in perception, it's interpretation, and attributions are even thoretization. Difficulty of classification is not a problem of perception, it's a problem of classification. If we can't name the color, it doesn't mean we don't see the color as it is.

Interpretation informs perception and perception isn't some constant in the human experience. A difficulty of classification can absolutely be a problem of perception because none of these things exist in a vacuum. Literally every human interpretation of reality is at best an approximate summary of what we experience of reality because it's impossible to remove bias and, believe it or not, we still don't know everything.

We don't know everything, but we discern reality just fine. Those are two different claims.

Sure, we discern reality, but that doesn't mean our perception isn't biased or correct.

Seemingly you talk about interpretation and theoretization, not perception. We have problem with building far reaching theories based on perception, but an error in a theory is not a bias in perception, see the color example above.