You are thinking of the null group. Placebo works, that is why you have to controll for it so you don't wrongly think the improvements is due to your treatment. Although, this really isn't really relevant to the conversation about how making sad people less sad is a good thing.