Amodal completion in visual search: preemption or context effects?