Sequence selectivity of form transformation in visual object recognition