So if I understand what you're suggesting, mice might have inherited a self-gene-modification facility that for smells passing certain criteria (highly associated with a threat somehow), can splice into the genome a representation of a receptor for that smell directly engineered from the molecular structure that "is" that smell. By modifying the genome, it seems we must mean modifying the genome in some or all sperm cells in males, and some or all egg cells in females.

Alternatively, mice sperm or egg cells might contain a previously unknown organelle into which the organism somehow routes samples of really bad smells that would play an active role in structuring the olfactory area of the embryo mouse brain predetermining their reaction to that molecule. This might predict a "washing out" effect over some number of generations.

If such a special case of Lamarkian modification is remotely plausible, it seems impossible to generalize to any sorts of trait other than smell perception.

I vaguely remember some other cases where internal stem cell structure other than the DNA played some role given as an example in Miriam Solomon's Social Epistemology as examples of challenges to Darwinism -- but they also seemed to come down to freakish ungeneralizable phenomena.