[m-photonic]

This is true, but it's too easy to go from that to assuming that ALL our cognitive biases were selected for. Some of them just emerge because it's hard to design a brain that doesn't work like that.

But some really basic cognitive biases, like the tendency for irrational confidence, are clearly evolutionarily adaptive.

[api]

True -- if you take this too far you get into the flimsier parts of evolutionary psychology.

On irrational confidence I've long wondered about whether Dunning-Krueger type effects are adaptive. If we had a rational image of our level of expertise, would we give up in despair in the face of undertakings like trying to understand theoretical physics?

[leppr]

I don't think the over evaluation part of the Dunning-Krueger effect has much to do with evolution. Basically it is a fundamental physical limitation: if you've never even seen an obstacle you can't say whether you could jump over it or not. After that it's just the natural optimism that's required by all species on earth to do anything non-trivial, nothing special. The under evaluation part is more interesting though, why not just keep going?

[deciplex]

Since we can understand theoretical physics, giving up trying to understand it would not be rational, at least not in every case. You'd need to somehow derive the utility you expect to gain from understanding a thing and then compare that with the resources required to understand it. Admittedly, both of these things are hard to estimate when you have no idea what you're talking about in the first place.