The quote was in an article titled Evolutionary Psychology and went like this (emphases in original):

Cognitive causes are ontologically distinct from evolutionary causes. They are made out of a different kind of stuff. Cognitive causes are made of neurons. Evolutionary causes are made of ancestors.

The most obvious kind of cognitive cause is deliberate, like an intention to go to the supermarket, or a plan for toasting toast. But an emotion also exists physically in the brain, as a train of neural impulses or a cloud of spreading hormones. Likewise an instinct, or a flash of visualization, or a fleetingly suppressed thought; if you could scan the brain in three dimensions and you understood the code, you would be able to see them...

Evolutionary selection pressures are ontologically distinct from the biological artifacts they create. The evolutionary cause of a bird's wings is millions of ancestor-birds who reproduced more often than other ancestor-birds, with statistical regularity owing to their possession of incrementally improved wings compared to their competitors. We compress this gargantuan historical-statistical macrofact by saying "evolution did it".

...When we're told that "The evolutionary purpose of anger is to increase inclusive genetic fitness," there's a tendency to slide to "The purpose of anger is reproduction" to "The cognitive purpose of anger is reproduction." No! The statistical regularity of ancestral history isn't in the brain, even subconsciously, any more than the designer's intentions of toast are in a toaster!

I frequently visualize things in ways that aren't quite correct. Not to perform precise calculations, but for the sake of my intuition. That might sound just as bad, but I find that intuition works better when it has internal visual cues to support it; your "flow of causality" example is what I'm thinking about here. When I'm thinking about some complicated system, imagining something flowing between the relevant parts to create the cause and effect helps me get what's going on.

So I would say that visualizations that aren't completely correct are not in and of themselves bad, as long as they don't start affecting your predictions directly.

IIRC it was while briefly watching this last year that I realized I was visualizing Europa as a giant ball of water covered in ice, and had been since possibly my single-digit years.

I do a lot of... not "visualization," exactly, as I'm not really a visual thinker, but sensory-conceptualization ("kinesthization" would be the right word, I suppose, were it a word) of systems and of states that don't map to anything I know to exist in the world.

For example, I frequently model the urge to do things as something not unlike a current (aquatic, not electrical) running through my body. I frequently model social interactions similarly, as a flow running between and among people. I frequently model complicated thought-structures -- arguments and software designs and so forth -- as physical networks of strings and beads, like games of cat's-cradle.

When I was doing a lot of cognitive testing after my stroke, and thus being pushed to remember long sequences of numbers and letters in various combinations, I found myself using a lot of kinesthetic markers to augment memory... e.g., I would "hold" items in my hands and order them with my fingers, or "hold" numbers in my right hand and letters in my left, or various things along those lines. It seemed to help, though I never compared results with and without those techniques so it could just as easily have been superstition.

It isn't quite the same thing as what you're describing, I think, but seems related.

I am not generally motivated to replace these imagined metaphorical models with more literally accurate ones, as I find them useful for various purposes. (The ones I don't find useful, I generally discard.)

That said, I'm willing to believe that they are local-maxima that I could profitably replace with other models that map better to the underlying phenomena.