Two annoying things about Bayesian meme "Perception is evidence": 1 Certain perceptions clearly aren't evidence, such as hallucinations, gestalt effects and so forth 2 Distinguishing belief from perception is way harder than just discarding every assertion that begins "I believe"
2 in particular strongly suggests, to me, that rationality is as much about the capacity to disentangle automatic processes in one's mind and make them deliberate and accountable, as it is about the bayesian intuitions I have so far read about.