What is perceptual bias?

Perceptual bias is a concept in personal epistemology: When you evaluate before you finish observing, your brain replaces incoming data with expected data. You stop seeing what is there and start seeing what you already believe.

Example: A senior engineer sees a spike in API latency and immediately says 'it's the new deployment — roll it back.' The team spends two hours reverting and redeploying. The actual cause was an upstream DNS provider throttling requests, visible in the first thirty seconds of the network logs nobody checked. The engineer's snap judgment didn't just waste time — it made the real cause invisible because everyone stopped looking once the first explanation landed.

This concept is part of Phase 5 (Observation Without Judgment) in the How to Think curriculum, which builds the epistemic infrastructure for observation without judgment.