In Milgram's baseline study, the wall separating the participant from the learner acted as a physical buffer; when this was removed in the 'touch proximity' variation, obedience fell to 30%, showing how buffers facilitate obedience.