I think taking a macroscopic view like this points at interesting types of analysis. Evolution is analogous to a hill climbing algorithm navigating towards local optima in a probability field. Macroscopic evolutionary trends give a low resolution peek into the underlying probability distributions.

That’s not what this theory is suggesting.

Frogs, birds, etc have very distinct spherical visual cortexes. Humans and squirrels don’t have that kind of spherical structure. So, other than naming where vision is processed the same thing the overlap isn’t particularly strong even at the most basic structural level.

It’s perfectly reasonable to extrapolate from studied organisms to species that share recent common ancestors. But reptiles don’t represent the probability distribution, they represent different results from that distribution under different evolutionary pressures. Extrapolation backwards and our common ancestor was very different from all decedents.