PS: One of the best ways to evaluate the non-placebo effects of drugs is to see how they work in animals. I can personally attest to the efficacy of amitriptyline (9th most frequently prescribed antidepressant in the US in 2007) in cats. Many years ago, I had a dear little kitty who had, um, “issues”. The manifestation of these issues which prompted intervention with psychoactive drugs was a habit of peeing on things to get attention — even though he knew full well the attention would be in the form of getting walloped, and to express anger (ditto on the consequences). This cat would pee on the coffee-maker (only when someone was making coffee), pee on the sugar bowl (only when someone was eating at the table upon which the sugar bowl sat), pee on the cable box on top of the TV (only when someone was watching the TV), and would also pee on any object left where it didn’t belong (that one didn’t bother me at all — made for a tidy house, since people quickly learned not to leave backpacks lying on the living room floor, etc). If somebody shut him out of their bedroom at night, he’d leave a giant puddle right in front of their door.
ALL of this behavior stopped immediately when he was put on amitriptyline, which apparently relieved whatever kitty depressions and anxieties were triggering the non-stop pee-assaults. Unless the kitty managed to understand the conversation between me and the vet, and thus understood what the thing getting pushed down his throat every day was *supposed* to do, and also *wanted* to achieve this effect (even less likely than understanding the conversation and the effects the humans were hoping for), then this wasn’t a placebo effect.
However I agree with your point that animals are not very subject to the placebo effect, and I would agree with the potential validity of observations of changes in behavior as a result of an antidepressant.