Define "complex memory". Do you mean episodic memory?

fMRI only shows correlation, not causation; any neuroscientist knows this. To prove a causative relationship, you need to use other methods (lesions or TCMS, for example). Show me an example of what you're talking about.