Listeners are timed on how long they can stay silent while their partner speaks — then rated on recall accuracy.