Processing multidimensional objects under different perceptual loads: The priority of bottom-up perceptual saliency