It reminded me of another wildlife documentary I saw a year or so ago which was about limpets and starfish. Starfish love to eat limpets - but limpets are able to defend themselves by raising their shells up and down slightly to squeeze the starfish's skin against the rock they are attached to, almost like a somebody stamping on your foot repeatedly as you try to approach them. Often, this is enough to annoy the starfish enough that it goes elsewhere and leaves the limpet in peace.
It would never occur to me that limpets, which seem like such simple animals, would have developed this defence method for dealing with starfish, but of course they have - they've been evolving in parallel with starfish as their chief predators for millions of years.
As humans we tend not to get predated on by other animals, but imagine if there were human-hunting beasties out there, as there must be in a typical D&D world; wouldn't it be the case that such predators would have developed highly specialised ways of killing humans, and wouldn't we have developed highly specialised responses?
Perhaps it is a species blind-spot, but it is quite hard to imagine what such a predator would be like. If it had been engaged with us in an evolutionary arms race for millions of years, it would probably be extremely resilient against weapons, extremely good at hiding (because we are good at organising ourselves into large groups), and also possibly very good at playing on our weaknesses - it might be able to perfectly mimic the sound of a child in distress, say. Maybe it would hunt us by tempting us away where we might be alone, and then hitting us with overwhelming force because of our pathetic physical weakness in comparison to most animals.
Or maybe it would be very like a mimic. Animals grow to resemble all kinds of things - leaves and sticks, most notably (although there are mantids and spiders which can perfectly disguise themselves as flowers to trick bees and other nectar-gatherers) - in order to escape detection. Why shouldn't mimics do the same to fool us?
I suppose the standard argument against the evolution of mimics would be: humans are rational, so unless a mimic was near-perfect in its ability to mimic it wouldn't really fool us. And so it couldn't really evolve - how would nondescript monster x incrementally evolve into something that could disguise itself as a door or chest or chair, if all of the intermediate stages wouldn't fool anybody?
The response might be that the mimic could evolve in exactly the same way the mammal eye evolved. Having a tiny bit of an eye that does nothing but tell you the direction in which light is coming is better than no eye at all. If there is one blobby ancestral species without even a rudimentary eye, and then one day a mutation provides a member of that species with a little nub of an eye that allows it to detect light, then its genes will proliferate because that will give it a big advantage over the others. So imagine a miscellaneous blobby human-hunting creature without any special ability to mimic anything humans would recognise: if one day one of these blobby things was born with a mutation which meant that, in a very dim light, at a certain angle, at a certain time of day, it looked sort of like a chest to a very short-sighted human, then that would be an advantage over the rest of the blobby things. That singular mutant blobby thing would have a slight edge on the others of its species and be more likely to breed. Its genes would gradually proliferate. And natural selection would take the mutation and run with it. Within a hundred generations, you might have a whole species of blob things who are really pretty decent at looking like chests.
This is the most rambling entry I think I've ever written: I'm tired and I've been drinking hoegaarden all night. I don't have the will to make it coherent. But you get my drift.