Part of the problem is one of incentives: the incentives to do a study which simply replicates a pre-existing study is low, and many journals won't even publish them.
Yeah that's what I noticed too (my minor was in cognitive psychology - like trying to figure out AI from the opposite direction). When Fleischmann and Pons posted their cold fusion paper, every physics lab in every school grabbed it and tried to replicate it, just because of how cool it would be if it actually worked. I never saw the same zeal to replicate interesting results in psychology. In fact you are banned from trying to replicate some of the most notable experiments because of ethical issues. It always felt to me like "this other guy tried it and this is what happened, so it must be true."