When I hear statistics about different countries, I sometimes think “yeah, that makes sense,” though it’s hard to know to what extent I already knew that or believed in a high probability of that and to what extent I merely convinced myself I already knew that just so I wouldn’t have a new thing to remember.

Consider false etymologies: you are learning a foreign language and you encounter a word that sounds like some other word you already know. So you think, "that makes sense. Arigato must have come from obrigado, after all Portuguese missionaries had been to Japan etc." Or rather, you realize that just because they sound similar it doesn't mean they are related, so instead you think, "I wonder if it is so...let's pretend this is correct so it becomes easier to remember." So whenever you hear obrigado you think, "oh, yeah obrigado, I remember that. It's like arigato." But as long as you keep in mind that this is just a mnemonic, you still have to remember that you made it up. It's arbitrary. What you might start to do instead as a shortcut is to pretend that there is nothing to remember and the connection is natural and obvious. You end up sacrificing a bit of the sanity of your world model and using it to encode the meaning of the word. You warp your rationality so that a new contingent fact appears to follow from what you already knew. "Of course! Obrigado means arigato. I should have known that," as if a reasonable person should be able to infer the meaning of the word even when hearing it from the first time just from the fact that it sounds like arigato, though of course it sounds dumb when you come out and say it like that.

I guess I can’t prove that it’s never worth it. Maybe sometimes it’s better to cultivate this bias towards a true fact and then to try to duct tape around it afterwards whenever it gives you trouble. I mentioned it “sounds dumb when you come out and say it like that.” So maybe just realize this and snap out of it somehow. Though this seems like trying to have your cake and eat it too. One might wonder whether this would corrupt your sanity in less obvious ways. And besides I have no idea how you’d go about figuring out if this is worth it, so I guess I’d rather try to keep in mind that the mnemonic is just that even if in theory it’s not always optimal. So I feel inclined to say that these things that I call corrosive mnemonics are generally a trap and a bad idea. 

I also speculate that children are really bad at (avoiding) this, that is, they are bad at making a distinction between what you should have known and what you couldn't have known, though my only basis for this is the one time when I was a very young child doing a test where I had to draw a _ girl, where I didn’t know the word that the test used to describe her, and I remember drawing whatever came to mind without even noticing that I didn’t really know that word.