Imagine going through the utter Hell of having your child die -- and then you're blamed for it and even prosecuted for first degree murder?
We won't know why the jury did what it did unless jurors speak out. But an article by law professor Sherry F. Colb suggests that in this sort of case, which is not a "whodunit" but about whether what was done was a crime at all (i.e, whether the death was a murder or accident), jurors sometimes might prefer the happy (happier) ending. See Sherry F. Colb, "Whodunit" Versus "What Was Done": When to Admit Character Evidence in Criminal Cases, 79 North Carolina Law Review 939 (2001). Jurors lean toward preferring that no crime happened at all. Concluding that a mother didn't kill her child is easier to stomach than the horrific idea that a mother killed her own child, a violation of all sorts of "laws of nature."
Acquittal is even more likely in a case built on circumstantial evidence, such as this one. Juries generally undervalue circumstantial evidence, according to studies discussed by law professor Kevin Jon Heller in his article, "The Cognitive Psychology of Circumstantial Evidence," 105 Mich. L. Rev. 241 (2006). Roughly, there's a big difference between a witness telling a jury a story (direct evidence) and the jurors' having to buy a story pieced together from circumstantial evidence.
Perhaps the prosecution should have considered these things -- the sort of thing we address in the growing movement, Applied Legal Storytelling -- and realized that they had a very shaky case?