This seems like a poorly designed experiment to me. People providing partial information about a word that's "on the tip of their tongue," would be expected to make wrong guesses. If they could make the right guesses, then there are good odds that they would be able to reconstruct/recollect the word. You could just as easily interpret this as the word having been "misfiled" so that its recall is blocked by incorrect information associated with it.
I get TOT far more often than I would like, but there's almost always a word that I get to eventually.
@Detry 10...9...
I was asked a couple years ago what's keeping us from having walking, talking robots that can at least pass as human to a drunk guy, this is the tech that I named. I figured we'd have good language and voice models by then (and we pretty much do), but the movement of biological systems is really hard to emulate with servos or other actuators. Muscles have the quality where they get thicker as they get shorter and can do so continuously. That goes a long way to making a face move in a realistic way.
We've had artificial muscles for some time, but they either require high energy, need to get hot to contract (and are therefore slow to cycle), or require dangerously high voltages. This one solves the high voltages problem by going miniature. You can get the high fields you need in much smaller voltages that way. It's still early days, but this could be a game changer.