Pigeon's recognition of occluded objects: differential effect of training experience