This article begins with the puzzle
Given three words — “trip,” “house” and “goal,” for example — find a fourth that will complete a compound word with each.
As they speculate how the brain solves such puzzles it seems to me that there is an easy story assuming Kanerva’s model of memory.
Still it requires elaboration of that model.
The article presumes a brain mode, probably temporary and global, where many matches, instead of just a few, are wanted.
This is the mode where ‘insights’ are more likely.
It seems clear that a ‘focus’ parameter which is global to at least some memory module (localized or not) can provide this variation between narrow and wide searches.
Upon wide search there is the problem of which of many responses to attend to.
Monte Carlo may often do the trick.
For the above word problem we must wonder which one-way associations occupy one’s head.
Perhaps poets and mathematicians differ in this regard.
For me, more nearly a mathematician, upon hearing “side-trip” I retrieve a set of notions and connotations and such notions often lead back to “side-trip”.
“Trip”, however, does not lead very directly to “side-trip”.
I think that I just try common words until one fits well with “trip”.
Computer science or computer design may have something to learn here.