Solomon Asch's 1951 line-judgment experiment: a real participant sits in a room with confederates who unanimously give an obviously wrong answer on a majority of trials. Across Asch's original studies, subjects conformed to the wrong majority on about 37% of critical trials (0% when alone with the task, rising sharply as the fake majority grows).
This simulator drives the virtual subject with a logistic conformity model fitted to Asch's published data:
P_conform(n) = P_max · (1 − e^(−n / n0))
P_max = 0.37, n0 = 1.6, n = number of confederates
if a "true partner" dissents from the majority:
P_conform(n) → P_conform(n) × 0.22 (Asch: conformity fell to ~5.5%)
subject's own baseline perceptual error (control condition, n=0):
P_error0 = 0.02 + 0.25 · difficulty
- Confederates slider — group size n. Asch found near-zero conformity with 1 confederate, ~13% with 2, and a plateau around 32–37% from 3 confederates upward — reproduced here by the saturating exponential.
- Task difficulty slider — how close the two closest comparison lines are; raises the subject's baseline chance of a genuine perceptual mistake independent of social pressure.
- True Partner toggle — one confederate breaks from the group and gives the correct answer, which historically collapses conformity by roughly six-fold — the model applies that same multiplier.
- Each trial: the board shows a standard line and three comparison lines: A, B, C — exactly one length-matches the standard. Confederates answer first (unanimously wrong unless the True Partner is on), then the highlighted subject answers last, sampled from the model above.
Real-world relevance: this same conformity pressure shows up in jury deliberations, workplace groupthink, and social-media pile-ons — minority influence research (Moscovici) shows a small, consistent dissenting voice can reverse it, which is exactly what the True Partner toggle demonstrates.