I know that my 300mm f/2.8 II rezzed up to 400mm gives better resolution than my 100-400 or former 400mm f/5.6. I did a "virtual" experiment to test out published data on lenses. slrgear.com has carefully analysed several lenses and plots the data as "blur" units. A blur unit equates to ~1 blur in the Photoshop Filter menu (for more on blur units see: http://www.imatest.com/docs/blur_mtf/
The procedure was have as the original a crop from a very sharp bird photo where the feather structure can just be seen. To mimic 300 vs 400, I reduced the image by 3/4 times, added 1 blur unit to equal the published blur of 1 unit in slrgear.com for the 300mm f/2.8 II, then rezzed it up to 400 by times 4/3. To compare the other lenses, I added blur units 1 at a time. The 400 f/2.8 II has 1 blur unit, the f/5.6 lenses have 3. Of course this test does not take into account other flaws of the lenses etc, and I think the Photoshop blur is too low for the poorer lenses.
But, the results are interesting. The images indicate the loss of high detail resolution for each blur unit. The uprezzed 300mm f/2.8 II is about the same as the 400mm f/2.8 II with one added blur unit, slightly better than the 400 f/5.6.
Those data are for high resolution details. You would not not notice any differences for subjects that lack fine detail but are composed of coarse elements. But, if true, those results show that a phenomenally good (and expensive) 300 is as good as, if not better, than a 400mm that in many quarters is considered as being very sharp.