Yes I agree with your first statement.But then IMO one doesn'nt need to propose a mechanism for a phenomena in order to prove that the phenomena exists. Just statistically significant positive results balanced against a control group is enough.