If you graph original "scene density" vs print density, it seems like you would want them to be the same (ie., a 45 degree line). Such that a photo of a gray scale should be printed so as to be identical to the original. This works great for reproduction of art work.
Originally Posted by AndreasT
But...if you do the same for a real-life scene, the print has a dull, blah sort of look to it. If you make another print, with slightly more mid-tone contrast, it looks lively and "just right." A graph of this reproduction will be steeper in the middle, demonstrating what Steven has said. (I should mention that I've only done this with studio portrait work; I presume it won't matter when the scene is foreign to the viewer.)
It seems sort of odd that this would be true, yet it seems to be so. I don't fully understand why - it is often explained as due to viewing flare and/or the print not being as bright as the original scene. I'm not fully satisfied with the reasons, but I'm convinced of the results.