On the assumptions underlying KS-like contradictions