There is an interesting RadioLab podcast on inheritance that is worth listening to. It discusses the differences between Darwinian and Lemarckian mechanisms of inheritance and more.
They discuss the case of the midwife toad and the work of Paul Kammerer. One part of this involved work that suggests the role of epigenetic inheritance (changes that do not involve mutations in the DNA, but rather hereditable changes in gene expression.)
The whole idea of the "Central Dogma" associated with molecular biology (namely that information moves from DNA to RNA to protein, but not backward from protein to nucleic acid, seems pretty clear. BUT, it is also true that a behavior can influence selection. This is the topic of the book and the video lecture: Evolution in four dimensions. by Eva Jablonka and Marion J. Lamb.
Sadly, there are some errors in the Radiolab: Inheritance pod cast. For example that methyl groups are "sticky", and that the transcription factors "knock off" methyl groups is weird (and wrong, at the molecular level). Both processes are mediated by enzymes (histone methyltransferases and demethylases). In fact, there are a number of modifications of chromatin, all of which are enzymatically mediated.
Also interesting is the RadioLab episode on Stochasticity. The section that starts around 43 minutes in is relevant to gene expression.