Its not just caching. This stuff seems to survive generations too by some as yet unknown magic. Google "inherited PTSD".
>Perhaps the most intriguing animal paper so far is a study published online in the winter of 2013 that appeared to show that fear memories can be inherited. Researchers from Emory University trained male mice to associate an odor with an electrical shock, so that they would get startled simply by smelling the odor by itself. Surprisingly, the scientists found that the smell also startled the next two generations of mice. https://www.vox.com/2014/8/18/5927269/epigenetics-definition...
At any rate, RNA definitely doesn’t survive generations. It’s a short-lived molecule. Stable epigenetic marks modify the DNA, they don’t involve RNA (except as an intermediary).
What little evidence there exists shows a very narrow range of effect in model animals. I’ve co-authored one such study that specifically shows the heritability of stress in mice. I can confidently say that this only works in very specific, artificially constrained instances, the evidence is tenuous, and mechanisms are elusive.
To clarify, this isn’t to say that trans-generational epigenetic inheritance doesn’t exist. Rather, we know that it does in specific circumstances (in particular in invertebrate models). But the claims by Rosner and in some of the more sensational (and flawed) papers require a big leap of faith that most scientists in the field aren’t willing to take, lacking evidence and a mechanistic explanation.
What’s more, none of the scenarios described by Rosner requires trans-generational epigenetic inheritance. It’s much easier explained by social/cultural transmission. Invoking epigenetics in this context is pseudoscience.
The first (i.e. epigenetic inheritance, particularly via DNA methylation) is a ubiquitous biological process by which a state is encoded in the cell and can be passed down to daughter cells upon cell division. Hence the daughter cells inherit the mother cell’s state. At its simplest, it just acts like a “bookmark” in the genome which informs the cell what parts to read, and what parts to skip over.
But things are different when talking about multi-cellular organisms that pass heritable information through the germ line. This is what trans-generational refers to. For the longest time it was assumed that no epigenetic information could be transmitted through the germ line due to a concept known as the “Weismann barrier” (think of it as the librarian: upon return, she removes all the bookmarks from the books). We now know that, under specific circumstances, this barrier can be “violated”, and “leaks”. However, unlike normal (genetic) inheritance, this leak does not allow a structured transmission of information according to our current understanding (that is, a bit of information encoding a phenotype “A” wouldn’t necessarily cause the same phenotype in the offspring; it would simply lead a perturbation). There are more specific exceptions to this rule which, for instance in nematode worms, allow a more structured information transmission (though it’s an open question whether DNA methylation is causally implicated).
In a word: yes. Even within the field this causes confusion (researchers of course know the distinction in principle, but the term “inheritance” is suggestive, and has led more than one person down the garden path). To make matters worse, “epigenetics” also can refer to several different things.