It sounds like we're just throwing things at the brain and hoping something sticks, and as long as it appears to work whether we know how or not, and it doesn't cause any serious side effects, it's packaged and sold?
Unfortunately, this is the best medicine can manage at times. When we don't have a clear understanding of the underlying causal nature of a disease like depression sometimes the best we can really do is the "brute force" method of trying various compounds until you get ones that produce acceptable effects without too much in the way of side effects.
There are still drugs prescribed like Lamictical and stuff that we know works (as an anti-seizure med for instance) but we are left in the dark grasping at straws as to exactly why. Same for some drugs that prevent cancer development. At one time people didn't even know how barbiturates worked. Honestly though, the FDA does not see this as a problem, because the alternative is leaving the condition totally untreated in a lot of cases. If the side effect profile is well-established before the drug is released, even if the mechanims is unknown, then doctors can prescribe with a modicum of safety.
The current paradigm of antidepressant development has brought us this far, but with current research, we are learning that maybe we weren't right all along. It's just going to take a while before researchers can put all the puzzle pieces together and come up with mechanistic explanations that would be more satisfying than the current ones. Like EA said, we are kind of running on fumes in the whole antidepressant department right now, and the best we can manage is something like "most antidepressants exert nonspecific changes in neural growth and plasticity". Wheether or not this effect is from activation of select serotonin receptors, activity at a secondary target, or from some combination of factors, remains to be established.
I think the rat/mouse model is the best we have
I think honestly, murine models are pretty shitty and you might as well be using C. elegans or something. It is the best we really have, though, unless you considere "live fire testing" on real depressed people a good ethical choice. The rat depression tests, to me, do not seem to correlate to humans very well; I don't know many people who are forced to swim around to find platforms in submerged tanks, for instance. There is also evidence that other systems models beased on rats/mice, like the model of obesity, do not exactly translate very well to humans (remove gene XYZ in rats, they never become obese, remove XYZ gene in humans and they don't live past 3 months)
