What is the Friggin' problem with the imagination?

I have been going back over my review of Philosophy of Science in Practice: Nancy Cartwright and the Nature of Scientific Reasoning, and want to note an interesting problem, or perhaps pseudo-problem, that I did not have space to bring up in the review.

Roman Frigg and James Nguyen, in their essay in the reviewed work, have a very enlightening discussion of how scientific models "represent" some phenomenon. (Roman was one of my lecturers when I studied the philosophy of science at the LSE, and an excellent lecturer at that, so I hope he will forgive my pun on his name in the post title! Also, I am writing this post without the book in front of me, so I beg forgiveness if my rendition of the authors' terminology is not exact.)

Frigg and Nguyen's primary criterion for how a scientific model represents is that it is "declared" to represent some class of events: for example, a histogram of adult heights in the United States represents those heights because it has been declared to do so. (This simple declaration does not make a model a good model: to be a good model, it must further exemplify some salient aspects of the thing being modeled.)

But, per this definition of a model, the authors feel embarrassed by entities like "a map of Middle Earth" or "a drawing of a unicorn": since there just is no such thing as Middle Earth or unicorns, what do such models "represent"? Their solution seems to me to be unnecessarily convoluted: they claim that such models, while being "representations," do not "represent" anything: they are "representations" only because they resemble other things that truly do represent (e.g., a map of France, or a drawing of a horse). (And here is where I may have screwed up the exact terminology, since I can't locate the book at present, but nevertheless I think I have their sense correct.)

I can't see why they choose this solution, rather than going for what to me is the more straightforward answer: a map of Middle Earth represents the way J.R.R. Tolkien envisioned Middle Earth in his imagination, and a drawing of a unicorn represents what people imagine unicorns to look like. Of course, imaginary things are not real in the same sense that the furniture of the physical universe is real, but so what? Surely, the creatures of our imagination exist in some sense, even if in a more shadowy sense than do real animals and real countries.

And I think that my solution to this problem sidesteps a serious difficulty facing the solution offered by Frigg and Nguyen: let us imagine that no histograms had ever been created before Tolkien, but that Tolkien himself invented the histogram to "model" the distribution of hobbit heights. Per Frigg and Nguyen, this would not have been any sort of representation at all since, at the time Tolkien created it, it would have borne no resemblance to any representation of a "real" phenomenon: hobbits "do not exist." (I use the scare quotes because my contention is that they did exist at the time of the model creation, in Tolkien's imagination.) However, later on, if other people picked up on Tolkien's idea, and began to represent the distribution of the values of "real" entities with histograms, Tolkien's histogram would mysteriously, post-facto, turn into a representation, since it would now resemble models that actually do represent.

And that post-facto transformation of non-models into models, I think, renders the Frigg and Nguyen model of models less plausible than the one I offer here.


  1. Sounds like they "represent" fatuity.

  2. Those who prefer not base their judgement textual evidence rather than hazy memories can find the chapter here:


    Terminology matters because it expresses concepts that matter. The crucial concept is not declaration but denotation. Models are representations of a target if they *denote* the target. Since something that doesn’t exist (like a unicorn) cannot be denoted it cannot be represented. Following Nelson Goodman we solve this problem by distinguishing between a representation of a Z (e.g. a representation of unicorn) and Z-representation (a unicorn representation). The drawing on the wall in which you see a unicorn is a unicorn-representation but not a representation of a unicorn. Resemblance plays no role anywhere in our account. The connection between properties of representations and properties of their targets is established by translation keys (not similarity). A strip of litmus paper turned red represents the solution as acidic, but red is not in any meaningful way similar to acidity.

    1. And, for those reading, I followed the link, and it turns out my memories were not "hazy" at all: I got all of Frigg's terminology exactly correct.

  3. "Models are representations of a target if they *denote* the target. Since something that doesn’t exist (like a unicorn)..."

    Well, this just completely ignores my point that unicorns *do* exist, in our imaginations, and that drawings of unicorns denote those imaginings.

    And resemblance certainly DOES play a role in your account: in declaring why a map of Middle Earth is a representation. (Not because it "resembles" Middle Earth, but because it resembles other maps!)

  4. "unicorns *do* exist, in our imaginations"
    I think you misstate a wee bit here, overstating what you said above: conceptions of unicorns do exist in our imaginations. And you are right drawings of unicorns denote them. Otherwise we wouldn't all agree that a hornless drawing, or a drawing of a pig with a horn don't match. Just as we would reject a street map of Bellingham Washington as a map of Middle-Earth.

  5. "conceptions of unicorns do exist in our imaginations"

    Yes, that's what I meant. There are not flesh-and-blood unicorns prancing around in our minds.

  6. So I have a memory of what my dead cat looks like. I make a sketch, and my neighbour says "why that looks just like Pyewacket".

    I don't see how Frigg answers these simple questions.

    Does my drawing represent of denote Pyewacket? There is no Pyewacket, she's dead. She dies before the sketch.

    Why can the sketch not denote my memory or imagining of the late Pyewacket? This alleged impossibility is essential for Frigg's argument, or Gene wins by concession.

    My neighbour is not reacting to *resemblance*? How did she know to use the word "Pyewacket"

    And what if actually there never was a Pyewacket, only previous sketches I made and called Pyewacket?


Post a Comment

Popular posts from this blog

Central Planning Works!

Fair's fair!

More college diversity and tolerance