"Drug restored damaged synaptic connections" -> "Drug is good" is a quite seductive argument that bears the danger of being accepted by smart people. The person might focus his fact check whether the claim about restoring damaged synaptic connections is true.

Given the failure of antidepressants in which companies invested a lot of money, it's rational to choose the prior "a new antidepressent isn't likely to create big positive effects" when evaluating a new candidate. Picking the right reference class is valuable.