Maybe anything that has a complex-enough internal model to be able to successfully interact with us will have the equivalent of an inner narrative. Maybe the way we humans do it is actually the simplest way.

Maybe. Eric S. Raymond argued a couple months ago that there's no hard problem at all, which I guess was making a similar point, though he didn't elaborate that I can find. To me, it doesn't seem like the subjective inner narrative should be strictly necessary, but that's just my own intuition.

I don't think that's right. The reason for something arising originally doesn't have to be the reason it continues existing.

You may be right. Still, some new reason for altruistic behavior that benefits the meme at the expense of the host would probably be needed, along with a deciding function for when it should be used.

Thanks for your feedback. You raised multiple interesting points and questions that I hadn't considered.