Transcendental Certainty?

I've read an argument I don't think I'm comfortable with, but I'm not quite sure why. Any suggestions? I'm reading Janet Broughton's Descartes's Method of Doubt for Ernie Sosa's epistemology seminar. In the beginning of Part II of her book, she attributes a very interesting argument to Descartes in Meditation II. She doesn't believe that he's actually employing the cogito at the point everyone thinks he does. Very briefly, here's the traditional interpretation that Broughton does not adopt (she calls it the "Cogito First reading"):

I am certain that I think. Anything that thinks must exist. Therefore I can be certain that I exist.

Broughton thinks that Descartes is employing a different argument to reach certainty of his own existence -- one which does not depend on a prior certainty. Here is the argument:

1. The only way I could doubt that I exist is to invoke a skeptical scenario. 2. The invocation of any skeptical scenario implies my existence. 3. Therefore, all skeptical scenarios about my existence are incoherent. 4. Therefore, I cannot entertain rational doubt about "I exist". 5. Therefore, I can be absolutely certain that "I exist" is true.

I'm sure that something about this argument makes me uncomfortable, but I'm not sure what it is exactly. It definitely has to do with the "certainty" status magically appearing in (5). Is (5) really a valid inference from (4)? It seems plausible that (5) could be a valid inference from "There is no rational grounds for doubt that I exist". But (4) just says that I can't entertain any. Isn't there too much psychology here for the inference to go through? Maybe there are possible grounds which I'm psychologically unable to entertain? I'm just thinking this through for the first time, so any suggestions/pointers/criticisms/confirmations would be appreciated.