Cat Couplings Revisited

About two years ago I wrote about Cat Couplings. I said:

A cat coupling is a kind of phrasing where it’s unclear whether an attribute is meant as justifiably picking out a subset, or unjustifiably describing the whole, and as a result strengthens the connection between the concept and the attribute.

I’m sorry, what?

It’s easier to explain with some examples. The one that made me take notice in the first place was this statement: “Pessimism has its downsides, but is still preferable to naive optimism.”

It was leveled at me during an argument about progress in society and it got me thinking. What does “naive optimism” refer to, exactly? Is it only picking out the optimism that’s naive — i.e. in this context mistaken — and saying pessimism is better than that? Well, true, but that’s not much of an argument if we don’t know how much of optimism is mistaken.

Is it saying that all optimism is naive simply by virtue of being optimism, and therefore that pessimism is more correct? No, that just assumes its own conclusion.

None of the two interpretations make sense on their own. But, if it’s unclear which one is intended, the weaknesses of each cancel each other out — there’s a broad version that says what you want, and a narrow version that’s easily defensible. The claim looks stronger, and is also harder to counter because you can’t pin down what it’s saying. Furthermore, the end result is to connect the concepts “naive” and “optimism” in the mind of the audience. This might even be the intended (or at least semitended) effect. The name “cat coupling” is a reference to the famous Schrödinger’s Cat thought experiment, where a cat is in a state of both living and dead at the same time until observed, in the same way a cat coupling is in a state of two combined interpretations until specified (which never happens).

My second example was about reducing the standard work day from eight to six hours. A poster on Reddit claimed the only ones who would lose out from it were “rich bosses”. Did they mean that only the “bosses” (I don’t think this person distinguished between managers and owners as “bosses”) that happen to be rich would lose out, or will all bosses lose — and bosses are, as we know, rich?

Neither, and both, I argued. There’s the requisite ambiguity: it makes the argument appear stronger (you don’t know exactly what to pick at ), and the end result is to connect “bosses” and “rich”.

“Naive optimism” was a clear example because it was actually used as an argument. Once I’d noticed the concept I found a lot of other examples that weren’t arguments as such and didn’t put the coupling on such display. That way the ambiguity and its effect could sneak by more easily. I put some in the original post but I found myself unable to stop noticing and collecting them after publishing it. Hence this post, where I’ll discuss some more I’ve found since then.

I’m proud that I’ve managed to get other people to notice cat couplings. The science journalist Tom Chivers pointed out this one on Twitter (from an article in The Telegraph about Brexit):

EU nationals will still be free to work in Britain – though he will make it easier to deport unsavoury foreigners and will also clean up the benefits.

Chivers thought that “unsavory foreigners” was a good example and I’m inclined to agree: “some foreigners that are unsavoury” on the one hand and “some foreigners — who are unsavoury by virtue of being foreigners” on the other. It’s not an argument so it doesn’t become stronger this way; it uses the ambiguity a little differently: to make a nudge-nudge allusion to general hostility to foreigners.

And then there’s Matt Yglesias, who tweeted “Neurotic introverts really having our moment in the sun with this pandemic.” Is that introverts who happen to be neurotic, or just introverts (who are neurotic)? This isn’t the most typical of examples because he’s not aiming to badmouth introverts. Still, on the surface it fits the pattern: he added “neurotic” because just writing “introverts really having our moment in the sun with this pandemic” isn’t particularly funny. He had to make a joke out of it by being self-deprecating.

And this one:

Is this person a contrarian and a twat, or does being contrarian make you a twat? Undecided and irrelevant. The other person is both — and bringing up examples of a person having two attributes and drawing attention to both of them at once reinforces the connection between the two. And you’d probably not say exactly this, and reinforce the connection, if you thought being contrarian was ever a good thing. (I wouldn’t say this, for example).

I recently watched the comedian Sara Pascoe go off on how stupid Pret-a-Manger is when bragging about their food being free from “unnecessary preservatives”. She noted, accurately, that this doesn’t mean anything. Who uses preservatives that are unnecessary? It requires an implied claim to seemingly make sense — that using preservatives at all is unnecessary.

This piece about QAnon includes the sentence:

“As you’d expect, QAnon supporters are much likelier to believe false conspiracy theories than everyone else, whether Q-specific or not.

“False conspiracy theories” is an interesting one. The use of “false” means, despite appearances, that a certain concession is being made. Many would just say “conspiracy theories” because the way that concept is constructed means it overlaps with “false conspiracy theory” almost perfectly. But not quite, apparently. So this one at once reinforces that conspiracy theories are false through the cat coupling, but it also admits that they aren’t automatically so because by using ambiguity you admit that the ambiguity exists.

It reminds me of the phrase “fake internet points” as a reward for winning arguments online. It’s a sorta-cat coupling because either the points are fake because they’re on the internet, or they’re the kind of internet points that are fake. It’s similar to “false conspiracy theories” in the sense that it’s actually the first, “maximalist” interpretation that’s stronger and more likely — the points really are fake because they’re the “internet” kind of points, and theories are fake because they’re the “conspiracy” kind of theory — but the technical ambiguity saves you from having to claim it outright.

Positive cat couplings?

So far I’ve focused on examples when you pin a negative characteristic on something. Could there be positive ones as well? How about this one:

If the government can’t do anything, why aren’t we a libertarian paradise?

On the surface there’s an implication on the one hand that if we lived in a libertarian society it would be paradise, and this is the connection (“libertarian” <-> “paradise”) the cat coupling builds, fully or partially intentionally. That can’t just be assumed, however, so it has to hide behind another, more innocuous interpretation. Note though that this other interpretation isn’t so much referring to a subset of libertarian societies that would be paradisic. That wouldn’t make sense given the statement as a whole. It’s more like “a society libertarians would consider paradisic”. It’s an ambiguous sentence that serves to connect a kind with an attribute by exploiting an ambiguity, yes, but the ambiguity isn’t exactly of the typical cat coupling whole-or-subset-focused kind, so I’d consider it a borderline case.

But there are real, bona fide positive ones. I found this in a discussion of medicine:

Internet research isn’t really a substitute for a professional who’s not only directly keeping up with the latest science in their field, but also has practical experience from seeing people come in and observing directly what works and what doesn’t.

It’s more subtle because it’s longer but it’s actually a perfect structural match for “naive optimism”. The kind is “a professional” (i.e. a doctor) and the attribute is “keeps up with the latest science and has direct experience with what works and doesn’t”. Sure enough, a doctor that does indeed do those things and does them well is a lot better than doing your own internet research[1], but the argument elides that you can’t necessarily rely on all doctors to do that (it’s certainly impossible for a GP to keep up with all the relevant science) by using the cat coupling trick. The result is, as always, to connect the kind and the attribute.

Another group-who-do-certain-things coupling comes to you translated from Swedish Reddit. This one is again negative:

The Left Party has been clear in their criticism of neoliberalism and this has had an effect on the regional government’s handling of the crisis. Generally, they’re the only party who’ve come out and said they want to get rid of venture capitalists who hollow out our public systems, whether in schools or in elderly care.

The kind is “venture capitalists” and the attribute is “hollows out our public systems”.

Types vs. sets

Clearly cat couplings can be quite different while still adhering to the same basic pattern. Some seem intentional and don’t make sense if picked apart because the ambiguity is central to making it work, while sometimes the ambiguity looks incidental and the implication almost a side effect.

They happen a lot even without much intent because they’re so easy to make and so easy to intuitively comprehend. It’s a lot harder to pick them apart explicitly and spell out the shoddy logic. That’s because logical analysis in terms of sets and subsets (“all of”, “most of”, or “some of” etc.) isn’t how language normally works, and that’s because it’s not how thinking works. Remember, logical terminology had to be invented.

Aristotle | Biography, Contributions, & Facts |

We think in types, not sets, and in our heads types get represented by a single instance with the same kind of attributes as an example of the type (this likely has something to do with how neural network-like pattern recognition systems like brains work). Sets, on the other hand, have entirely different attributes than the examples that are in them, such as counts, distributions and correlations. Those attributes aren’t represented well when when we use types to think about what are really sets. The tools with which to think about them have to be acquired through advanced education: statistics, probability, machine learning.

In other words, our intuitive ways to think about categories clash horribly with the newly (relative to the whole of human existence) invented logical terminology and its extension into statistics etc. that enables more exact claims about reality. This mismatch causes an absolutely gigantic amount of confusion and dysfunctional disagreement.

Cat couplings make much more sense and aren’t weirdly ambiguous at all if we understand them in terms of types and not sets. They’re not making logically intelligible claims about physical reality, they’re evoking, constructing and rearticulating types. “Naive optimism” is a type, a concept, a piece of mental machinery, and, if you will, a social construct. So are “rich bosses” and “unsavory foreginers” as well as neurotic introverts and contrarian twats. “Doctors who keep up with the latest science” and “venture capitalists who hollow out our public systems” are arguably types as well, if less succintly described, and some certainly want to rearticulate and reinforce them[2].

Now cat couplings become pretty straightforward. On the “change my view” subreddit I saw this:

The “slippery slope” argument to anything is a trash rhetorical argument based on fear.

There’s a lot of complex set logic to untangle here, if that’s what we wanted to do. This person clearly puts “trash” next to “rhetorical argument” to make “rhetorical” look bad by invoking the bad side of “rhetoric” as being instrumental and manipulative rather than honest expression of beliefs. On top of that, “based on fear” follows, using a cat coupling style link to suggest that fear-based arguments are “trash rhetorical” ones. Lots of type-building[3].

Then there’s this article with the sentence:

The Senate was always a really fucked-up anti-majoritarian institution.

There are institutions that are anti-majoritarian and fucked-up, and highlighting both these properties suggests but not directly asserts that anti-majoritarian-ness implies fucked-upness. Ambiguity goes further than that, though, because even if we agree that anti-majoritarianess implies fucked-upness it’s still not clear if anti-majoritarianness causes fucked-upness or if it constitutes fucked-upness. Those are different propostitions!

This shows how type- and essence-thinking doesn’t just clash with logic. It doesn’t get along well with causality either. In type-thinking it makes sense to ask whether having an essence or being of a type makes you have a certain characteristic[4] (because the characteristic could be part of the type’s essence and thus by virtue of being of the type you have the characteristic). In terms of physical causality, where physical events cause other physical events over time and there are no eternal essences that make anything be the way it is, it doesn’t make sense at all. Essence like claims get taken apart into degrees of correlation strength and of causation certainty, which are different from essences nd from each other.

Sorry for getting into an anti-essence rant — it’s an itch I have to scratch sometimes. Back to cat couplings. Do you think the person who said “fucked-up anti-majoritarian institution” worked all this philosophy out in advance? Of course not. No need. It’s not a stringent philosophical or scientific claim and if it’s interpreted as one it breaks. It’s just ordinary type-building.

You can try to deliberately create new types with cat couplings, but I think that’s limited to intentional propaganda and not always effective. Most of the time it’s just an expression of people having certain types in their head that they think others will be familiar with and they don’t mind perpetuating. At work recently a client praised a report I’d written for not being “a consultant’s report that’s just packed with jargon to impress and intimidate”. She could have just said that it wasn’t like that and that was good but no, a consultant report like this is a type, she referenced the type, and didn’t mind perpetuating it. I suspect that if we’re about to reinforce a type we don’t want reinforced or rearticulated it gets blocked somewhere in our minds before it even reaches conscious awareness.

I don’t think a poster on Reddit that described themselves as “just an ignorant American” minded reinforcing that type either. Also on Reddit I read a discussion about a scientific study of whether autism severity correlates with certain facial features, and there was this comment:

You’re acting like it was an innocent, neutral hypothesis born out of pure scientific curiosity, rather than a concerted effort to legitimate politically conservative bigotry using official, scientific language.

Specifying politically conservative bigotry in this particular case felt odd to me, because AFAICT autism-based bigotry is among the least likely to have a politically partisan loading (maybe I’m wrong about that). In any case this is a clear effort to reinforce a type through essence-vs-subset ambiguity, and that makes it a cat coupling.

Bad Something

There’s a particular phenomenon of pointing to just the bad examples of a kind in a way that throws shade on the whole kind.

The article Bertrand Russell got Stoicism seriously wrong argues — would you believe it — that Bertrand Russell got stoicism wrong. It says that “if Russell had written the thing today you would suspect that he got his information out of a bad Wikipedia page”.

Not just a Wikipedia page and not a “bad summary”. No, a “bad Wikipedia page”.

On the Star Trek subreddit, which I frequent for relaxation (sometimes unsuccessfully), somebody wrote:

Get over it. Go forward. Come up with something new. Enough with paying people to write bad fanfic about their fav’rit characters from their childhoods.

What else is bad but fanfic? Rom-coms!

People who are unable to let go of infatuations that aren’t going to become relationships happen are a risk to the subjects of that infatuation. Will this person crash my wedding like some sort of bad rom-com?

It’s hard to know and explain when “bad Something” is type building and when it’s not. “Bad wikipedia page”, “bad fanfic” and “bad rom-com” feels like it is, while, say, “bad movie”, “bad novel” or “bad guy” doesn’t. It comes down to preexisting associations — types — and that’s partially subjective and individual.

Sometimes you can figure it out just by context, even if you don’t know the type. Say an alien spaceship appeared in orbit and the captain beamed down to have a beer with you and proceeded to complain that “asteroid mining is hard work and all I’ve got is a crew of pampered Zorblaxians” then you’d know right away that 1) Zorblaxians are considered soft, and 2) the captain doesn’t mind reinforcing that (stereo)type.

Knowing the speaker can sometimes help you resolve the ambiguity of whether something is ambiguous or not, i.e. a cat coupling (yeah, I’m getting tired as well). In this tweet

Another philosophical nothing-burger. Of course scientific theories can only be falsified against other theories, but it makes your computer run, so who cares? Besides, how much theory do you need for most science? A lot of it’s just pattern-observation.

…it’s not clear if “philosophical nothing-burger” is a cat coupling saying that philosophical things are nothing-burgery in essence or if it’s just specifying the domain of nothing-burgeriness, like “scientific misconduct” and “political scandal” do. They’re not cat couplings (political scandals do tarnish politics as a whole but the particular language construct isn’t meant to do so). I do happen to know the tweeter is very critical of philosophy and how it’s done, so I interpret it as a cat coupling.

It’s all connected, man, and language is at the center

Digging into a relatively narrow phenomenon like cat couplings inevitably leads into types and type construction and rearticulation, and how that whole essentialistic way of thinking clashes with the more recent concept innovations like logic and physical causation. That further leads to interpreting intent, preexisting associations, and just how much work we do, often without realizing it, filling in the blanks (positively or negatively) in what other people say.

That in turn inevitably leads us toward contemplating just how much farther removed from concrete, physical reality and stringent logic our language constructs really are, compared to how we feel they are when we’re not staring directly at them. Effortless and seemingly straightforward claims quickly turn into complicated messes if we try to pin them down in exact terms — and even if we do we lose the actual intended meaning, because it’s denominated in type-based mentalese currency and not Hard Reality Coins. Despite attempts, no one’s figured out a reliable conversion process.

• • •


The phrase “internet research” as opposed to just “research” (which is conducted largely on the internet in 2021 anyway), is also worth looking into.

What types we have access to in our heads, and which ones we have in common with others and thus can use to communicate with, matters a lot for how easy it is to say certain things and be understood. I won’t repeat the whole thesis of Wordy Weapons of Is-Ought Alloy here, but do read that one if you find this compelling.

I’d object that fear can be correct, rational and prudent and not necessarily wrong as a basis for argument, but that’s another story. It says something about how safe our societies are if we habitually think of fear as some kind of irrational hang-up.

We get into very old and messy philosophical questions here about essential and incidental properties.

9 thoughts on “Cat Couplings Revisited

  1. Time has moved weirdly over the past year and a half — when I saw the title of this new post I could have sworn that your original “cat couplings” post was from just a few months ago, although I couldn’t remember what the term referred to.

    This is very clearly a particular type/variant of the motte-and-bailey rhetorical device, which I’m surprised you didn’t place emphasis on (although I did just check and see that you mentioned it in a footnote in your earlier post). I think it’s subtle enough variant, though, that (as I recall) back when motte-and-bailey was exploding into the rationalist discourse nobody was really isolating it.

    We need to critically analyze Twitter rhetoric so as not to get confused by cat-couply motte-and-bailey fallacies.
    A meta- cat coupling, possibly? 😛

    Liked by 1 person

    1. Yeah it feels very recent to me too but it was almost two years ago. Time flies, as they say, an expression that grows more and more accurate as you get older.


  2. The concept of cat couplings reminds me of a grammatical feature I remember from taking some Spanish classes. There is a distinction between _adjetivos especificativos y explicativos_, the former type of adjectives restricting the referent (e.g. «the white shirt») and the latter reinforcing an intrincic attribute (e.g. «the white snow»). The difference lies in whether the adjective follows or precedes the noun, so the two examples would be «la camisa blanca» or «la blanca nieve», respectively.

    It would seem, then, that cat couplings would be harder to create in Spanish because the position of the adjective would imply either a restrictive or reinforcing meaning. But I’m not qualified to make that judgement, and I really doubt it’s true – language is pretty flexible.

    Liked by 1 person

    1. Harder seems pretty undeniale in that case, no? I had someone say something similar about French on Twitter and it just delights me how varied languages are.


  3. Some of these examples remind me of the Tversky and Kahneman question, you know it:

    “Linda is 31 years old, single, outspoken, and very bright. She majored in philosophy. As a student, she was deeply concerned with issues of discrimination and social justice, and also participated in anti-nuclear demonstrations. Which is more probable?
    Linda is a bank teller.
    Linda is a bank teller and is active in the feminist movement.”

    Somehow adding the extra detail makes it seem more probable, though obviously it is not.

    Liked by 1 person

    1. Yeah. I’ve never been fond of the Linda problem really. It’s supposed to show a flaw in our reasoning but it really just shows how ordinary use of language works differently from formal logical statements. Obviously the implication here is that the first option includes “and isn’t active in the feminist movement”. We take context into account when interpreting things, and there are always unspoken assumptions we also include. This is just a more sophisticated version of:

      – Which months have 28 days?
      – Just February.
      – No, all of them do! Hahaaaahahaaaaaaa!



  4. This kind of reminds me of the old writing advice to kill your adjectives. If you ever find yourself reaching for an adjective, use a better noun, or just set things up so you don’t need an adjective at all – e.g. don’t write “She smiled beautifully” when you could write “She smiles, and it’s the smile of my memories.” or something like that. Wherever you do use adjectives, use them sparingly, as with “The air is cold. The ice is real. All else is memory.”

    If I had to sum it up, it’s to never tell anything you can’t show. You can’t always show instead of telling, but you should never tell what you can’t back up. I think something similar is at play here, with people trying to tell things without showing them, smuggle them in without backing them up. Like the article said, it’s a matter of sneaking in associations, but it helps to dissect how exactly that’s being done. If the writers of those Cat Couplings had to delete those adjectives, they’d be forced to show readers the point rather than telling them it. With only nouns, to look specific they’d have to be specific, instead of hiding within the Motte and Bailey* of adjectives – a noun is an noun, but an adjective doesn’t always specify how much of the noun it’s modifying, as you’ve pointed out.

    All this further reinforces my belief that good writing forces good thinking. You can’t be vague and hide from the implications of your own words if your own words are clear and straightforward. This goes doubly so when you remember that good writing is a matter of good editing – going over your words again and again is the best way to find their mistakes, grammatical and otherwise. There’s no substitute for putting in the work, and writing forces you to put in the work. To write is to think with accountability. Cat Couplings are how you write without accountability.

    (*Also, on an unrelated note I maintain that the Motte and Bailey fallacy should have been called the ‘Fort and Field fallacy’ right from the start. It’s so much clearer that way!)

    Liked by 1 person

    1. I’ve been thinking of writing about it for years but I’m quite skeptical of some interpretations of the “show, don’t tell” thesis. It works primarily for narratives (fiction and journalism) that’s about evoking feelings and images, and less well for declarative writing that’s about ideas.

      Even in fiction, the “be concrete and sensory, and avoid adjectives and abstraction” is more a particular style than the definition of good writing, and it hasn’t necessarily been the norm in all times and places. There’s a good version of it, but it’s more about making the reader think what you want them to think through clever prompting rather than directly telling them — but that’s different, I feel.

      About the rest, sure. And I like Fort and Field as well but the ship has sailed…


      1. When it comes to Cat Couplings, it probably is a separate matter from “Show, don’t tell”, but it’s kind of analogous in terms of it being about how people can dodge proving things with evidence (e.g. whether optimism is naïve) by asserting them without tying yourself down to your own claim (e.g. “Oh, I was just talking about the kinds of optimism that are naïve.”). Perhaps a Cat Coupling is what happens when you realize that even telling is too clear and would reveal too much of the unsupported nature of your argument, so you have to hide it behind the ‘plausible deniability telling’ of a Motte and Bailey/Fort and Field. The answer to that would perhaps be something like “State, don’t insinuate” to match “Show, don’t tell” – a simple reminder to both authors and readers of what not to do/not to trust, something short and memorable.

        (Also, is ‘Plausible Deniability Telling’ a better or worse name than ‘Cat Coupling’? It’s probably worse, in terms of being nothing but ugly functionality…)

        Also, maybe there’s still a chance for Fort and Field? Just write Motte and Bailey/Fort and Field whenever you would write ‘Motte and Bailey’, and hopefully that sets up a transition period where people get used to both, before you retire the first one like it’s a piece of legacy code.


Leave a comment

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s