Double blind experiments do show causation (ignoring metaphysical discussions).
There are 3 possibilities: Chemical -> Behavior, Behavior -> Chemical, or ??? -> (Chemical and behavior).
A randomized trial eliminates the latter two possibilities (since chemical is correlated only to a coin flip). Double blind eliminates experimenter bias.
Then, it could be that the chemical somehow causes happiness. Or it could be that the chemical causes memories of one's children, and everyone in the study has happy memories of their children.
So, the causation is still in doubt. Many explanations are consistent with the observed data.
---
Yes I am proposing that chemicals correlate with "aggression" in humans and rats for different reasons. But I don't accept that the thing called "aggression" is the same thing in humans, and in rats. To help illustrate why not, consider aggressive behavior by video game characters. Same word, but not the same phenomenon -- in that case, no emotion is present, and the cause is some C code.
---
Double blind experiments (with controls, randomization, etc) do not show causation, they show correlation. You need to have an explanation of what the cause is, or that kind of study doesn't get you anywhere. Because without an explanation, you don't even know what you should be looking for, or have any idea why that correlation exists.
BTW, if you know of a reputable, scientifically-oriented source which believes that double blind experiments demonstrate causation in the absence of an explanation, not just demonstrate correlation, I'd like to see it.