> conscious perceptions are guided entirely by evolution, emphasizing those aspects of reality needed to keep us alive and filtering out those we can safely ignore.
It's also known that it's often enough evolutionary advantageous having the model of the surroundings which is more accurate than that of your competition (There's some paper I've read about that, I just don't have the time to find it now. Maybe somebody has some more ready).
Therefore the successful products of evolution correctly reflect "outside world" in their models, and even have the "safety mechanisms" and "error correction" facilities (based on the feedback, of course). There are even experiments with people: if you'd get the glasses that invert the picture you see, you'd for a while see the world top down, but if you wear them long enough, the internal adjustment of the model would happen and you'd see again up as up, even if the "signal" is provably reversed compared to what you received for your whole life.
On the opposite side (that the "accurate enough" is not "always correct") we also already know the examples where the "accuracy" breaks in humans: that's the cause of people ascribing to the agency of gods the phenomena with purely natural causes. There wasn't evolutionary need for an intrinsic development of more accurate model of these phenomena (e.g. what makes the Sun move across the sky or what the stars are). The scientific method, luckily, allowed humans as a group to overcome these limitations, e.g. Aristarchus, some 2200 years ago, that is, at least 200 years BCE: https://en.wikipedia.org/wiki/On_the_Sizes_and_Distances_(Ar... (note, at that time the term "science" still didn't exist).