The subject figure (blue) stands at the end of a horseshoe of confederate figures. Each trial, the confederates light up red and lean forward one by one to announce the same wrong answer; the subject then visibly wavers — colour and posture interpolate toward red (conformity) or stay blue (independence) — with the outcome sampled from Solomon Asch's own 1951 published error rates for the current group size. A bank of floating 3D bars behind the room stands in for the published conformity-by-group-size curve, and a marker climbs the bar matching the live group size as trials accumulate.
lean(t) = lerp(0, leanMax, ease(phaseT))
subjectColor = lerp(blue, red, conformed ? 1 : 0)
P(conform | n) = AschCurve(n) // shared with the 2D twin, sampled independently here
- Unanimous confederates — instanced figures added to the horseshoe; more of them means a longer, slower unanimous announcement sequence before the subject must answer.
- Auto-orbit — the camera sweeps around the room so you can see both the group's posture and the subject's reaction without manual dragging; turn it off to drag-orbit freely instead.
- Bar chart — each bar's height is the published conformity rate for that group size; the glowing marker tracks the bar for your current slider setting as a spatial readout instead of a 2D line.
Real-world relevance: physically staging the room — subject last, group unanimous, answer spoken aloud — is what made Asch's effect so much stronger than private, anonymous polling ever produces; the same geometry (who answers when, and how public it is) still predicts conformity pressure in modern group settings.