Left visual neglect: is the disengage deficit space- or object-based?