To sum up the article, the effect of lots of bicyclists leading to careful driving is greater than the protective effect of wearing a helmet while biking. Because helmet laws do cause a drop in ridership, it's counterintuitively a net negative for cyclist safety. That is the claim of the article.
To refute this, one must show that the "critical mass theory" is not significant and that driver attentiveness to cyclists is not a function of the number of cyclists.
The claim is not refuted by giving studies showing that when cyclists crash and hit their head, then helmets are (incredibly) significant for reducing head injuries. Nor is the claim refuted by saying cyclists should not be in the road, or saying (without evidence) that there is no reduction in ridership from helmet laws (the article cites examples demonstrating that reduction).
So again, the car centric here need to focus on whether the critical mass effect is actually a real thing, or need to find the evidence that ridership is not a function of helmet laws (the data, generally and from this article, indicate that both are the case).