Seems to me that a viable explanation for the first experiment is that heavy cognitive processing trips some circuitry in the brain that says "We got a lot of work to do. Get me some glucose."
On the other hand, I believe there is also a body of more recent research indicating that the "cognitive fuel tank" model is too simplistic to give an accurate description of how it works.
It went through peer review process and got praised for being revolutionary.
However the author was a MD.
Peer-reviewed scientific literature means nothing on its own.
However, the very common exception is when someone with an agenda abuses the term "peer-reviewed scientific literature" to describe a single article or a string of articles from a single or a few closely-affiliated sources in order to support their agenda. Of course, this is taking advantage of the true meaning of the term. See also "clinically proven".
It was still absurd and should not have happened.