There are quite a few posts talking about how this isn't natural selection. How it's not evolution. How they should have done this with genetic engineering. How everyone who can write a comment would have done it better than these guys. That's very cute.
It's important to keep in mind that natural selection will effectively span the full probability space of all possible traits as far as offspring go, and only the strongest survive.
What that means is twofold:
1. Given that you're trying to study tradeoffs in nature within the same species, you obviously don't want to engineer any combination of traits that wouldn't occur naturally.
2. What the scientists did was exactly what they should have done. They selected traits that could have occurred naturally, but with a small probability. By enhancing these traits, you can then assess how over time and generations the inheritance would play out.
Obviously, the study's scope wasn't as far-reaching as the ultimate evolutionary end-game of guppies (that would require somehow objectively quantifying the increase in survivability due to larger brain size vs. sheer number of offspring), but they do manage to demonstrate a pretty strong dependence (more than correlation, as this was a directly applied change vs. a control) of the compromise in energy expenditures in developing organs.
Yes, if the results were anything other than "common sense", it would be remarkably exciting news that would warrant further study as to the hiding spot of our brainy fish-overlords. But in science, sometimes it's just as difficult (albeit slightly less rewarding) to isolate a fundamental tenet in its most distilled form.