This is wrong; you are right. The sensor can't tell which part of the exposure goes "on top" of the other. You don't have to take my word for it, either; here's a quote from the Flash basics guide on Scantips:

Some people imagine that the delayed result of rear curtain sync causes a sharp stopped image superimposed on top of the blurred image (so is sharper), which may appear true of the leading edge, but the opposite is true of the trailing edge. So while there definitely are two separate exposures, and the flash does freeze the action when it triggers (and the continuous light continues blurring it), the effect is not "on top" of anything. Each pixel can only contain the one total accumulated pixel exposure value, regardless of when.

It is, sadly, pretty easy to find other photographers on the Internet getting this wrong. I probably should avoid linking to them, but I can't resist this one, where the author gives the explanation of first-curtain shutter "blurring over your freeze", and then posts an example of how second-curtain is better clearly showing blur which appears "on top" of the frozen subject in addition to apparently in the background. Ah well — sometimes what we believe overrides what we see!