A Culture of Generalization

Imagine that you’re part of a primitive society and someone in your tribe sees lightning for the first time. They witness a line of light cutting through the night sky, and several seconds later they hear a loud crash. They tell the rest of the tribe what they saw: “From the Heavens appeared a long line of jagged illumination perpendicular to the Earth, and several moments later was a startling sound.” But then let’s say that the tribe observes lightning and thunder many more times. Before long, everyone in the tribe would know the pattern: “When the jagged lines of light appear, afterwards comes a loud crash.” This marks a transition from talking about particular bolts of lightning and particular rumbles of thunder to making overall generalizations about the categories lightning and thunder.

In modern times, of course everyone know that lightning causes thunder. Unless we’re setting the scene for a story, it’s no longer necessary to mention specific instances. Our knowledge of lightning and thunder has long passed the phase where we’re talking about individual cases of lightning leads to individual cases of thunder. We now understand that all instances of lightning will result in some instances of thunder, or in other words that lightning causes thunder. With improved knowledge of how the world works begets a transition from talking about individual things to establishing general rules.

Now, imagine that you’re walking down the street and you notice a cat. You walk over to the cat, bend down, and try to pet it. But the cat lunges at you and tries to bite your hand. Startled, you yank your hand away and then leave. Later on, you mention to a friend: “Earlier today I tried to pet a cat on the street, but it tried to bite me.” Your friend suggests not trying to pet random stray cats on the street anymore, but you say that you’ve done it a hundred times without issue. Most cats you run into are friendly!

Let’s say that over the following few weeks, you continue your habit of petting cats on the street when you happen to encounter them. But then one day, you nearly get bitten a second time. Now you have two data points to work with. You get a flashback to the original cat who tried to bite you, and you realize that both cats had a similar appearance. When you see your friend, you say: “I almost got bitten trying to pet a cat on the street again. Like the last cat who tried to bite me, it had a strange appearance to its face.”

And then a month later, you’re walking home and you notice a cat with the same appearance. Oh, it’s that type of cat again! Curious about whether this cat will try to bite you as well, you walk over, lean down, and then move your hand slowly toward the cat. Right when the cat shows the very first sign that it’s going to bite, you pull your hand away. Aha! Next time you see your friend, you recount what happened. Your tentative conclusion is: “Cats who have that appearance are aggressive.”

Basically, as you go from less knowledge to more knowledge you go from a particular story, to multiple stories, and then finally to a generalization. First you just mentioned “a cat”. You’re not claiming to know anything about why the cat did what it did. It’s simply you relaying the choice the individual cat made. Next, you mentioned “a cat” again, except with two data points you find a commonality between the two cats who made the same choice. And then finally, you were able to dispense with the mention of any individual cat, instead just producing a general statement about all cats with that appearance. As you gained knowledge, your propositions became less individual and more general. From “a cat did Y” became “two cats, both with appearance X, did Y”. And from that became “cats with appearance X are Z”.

When we’re talking about lightning, thunder, and cats, these ideas are uncontroversial enough. But what happens when we talk about human beings? What if we transition from talking about individual people making individual choices, to categorizing these individuals into groups and then making generalizations about human behavior?

For example, rather than saying that Mike became a computer programmer, Chris became a mechanical engineer, Jessica became a psychologist, and Ashley became a teacher, we could assert that men tend to be more interested in things and women tend to be more interested in people, and that therefore men are more likely to pick STEM professions (where they work with machines and software), while and women are more likely to pick professions which stress people skills. Presumably, this is just a garden-variety transition from less knowledge (where we’re restricted to talking about individual instances) to more knowledge (where we have enough data to generalize our models into overall patterns). But this topic is likely to spark much more controversy than making the same sort of epistemological transition when talking about the weather or animals.

Any subject where you say men tend to do one thing and woman tend to do another is liable to explode into a controversy. When you lump a large variety of individual human beings into a group and then say that belonging to that group has to do with what they choose to do with their life, you can run into trouble. You’re supposed to treat everyone like an individual. Just because someone is female doesn’t mean they’re not going to join the STEM field!

Far more controversial than identifying patterns in how men and women choose professions, of course, is identifying patterns in how women choose their sexual partners. When men write articles on how to seduce women, there’s often major pushback against the idea that they would venture to make such sweeping generalizations about such personal matters. Each woman is unique in her desires. Women have free will. They’re not just helplessly shackled to their group identity as female. They’re not automatons who simply react predictably to stimuli. Only a misogynist would try to reduce roughly half the world’s population to a simplistic model designed to manipulate through formulaic procedures.

But from the point of view of epistemology, there’s a fundamental continuity going through all of these examples. Whether we’re discussing lightning and thunder, cats, men and women choosing their professions, or women choosing their sexual partners, the initial state of ignorance forces full recourse to individual instances, and the eventual blossoming of understanding allows for much more generalization. Surely only a man with little experience with women wouldn’t generalize! What else is not making generalizations other than not having enough data to find reliable patterns? Men who insist upon the unique individuality of each woman is merely a man possessing too little experience to identify any overall trends. And like crabs in a bucket, men with insufficient experience to find useful patterns tell other men who try to find those patterns to stop looking altogether. These groups of men end up trapped in a pit of ignorance. A chorus of losers chant endlessly about the unique individuality of each woman. Being shrouded in darkness about how women think, act, and choose is applauded as virtue. They’re pathetic, but at least they’re not sexist!

Or is there something deeper going on here? Why are certain forms of generalization allowed, while others are prohibited? After witnessing a hundred tornados, you’re allowed to go from talking about the behavior of individual tornados to generalizing about the behavior of all tornados. But no matter how many black people you interact with, you’re still discouraged from making general statements about the tendencies of black people.

From a purely epistemological perspective, both processes are identical, notwithstanding the fact that the former is a much simpler research project and the latter is extremely complex. But culture has a different opinion. The purely epistemological process of building your knowledge is but a small aspect of the whole process. Culture knows that there’s something else of much greater significance to take into account. After all, consider how laughably, deplorably, or lamentably racist it would be to think that you can study black people the way you study tornados! Black people are human beings with free will, not weather patterns unfolding according to mindless meteorological laws.

Basically, there’s no individual sovereignty to protect with tornados and other weather patterns, while reducing black people to their group identity threatens to erase their individual sovereignty. Tornados are natural phenomena which behave according to physical laws, whereas human beings have free will. Foisting the determinism of the natural sciences onto the topic of what various categories of humans do removes the individuality from the equation. There’s a special uniqueness of each human soul that people will fight to the death to protect.

If you say that “a cat” tried to bite your hand, you preserve its autonomy as an acting agent. For all you know, it could have done something different. It could have chosen not to lunge at you. But if you say that “cats with appearance X are aggressive”, then you begin to whittle away at the sense of autonomy. It’s no longer that the cat simply chose by its own free will to try to bite you. It’s now a cat that looks like X, and cats that look like X behave like Y. The cat is less of an agent, and more of an automaton behaving in a predictable way.

In the end, it’s only ignorance that can preserve individual sovereignty. Once you know too much, you crowd out the entity’s free will. Rather than saying the agent made a choice, you’re saying that the thing simply moved helplessly in accordance with the laws of human nature, animal behavior, or natural phenomena. We all need to operate on heuristics at times, but if you generalize too well you start destroying the individuality. There’s a certain preservation of mystery that most people aim for, especially in long-term relationships and marriages. Knowing too much about the mechanics of sexual psychology can start to break that down. If your partner does X simply because they’re the member of a particular subset of women that you’re familiar with, then you’re no longer interacting with a human being. You’re dealing with a machine, and you what know what the machine is programmed to do in what situation.

More importantly, a culture of generalization can lead to a system of self-fulfilling prophecies. If girls are taught from a young age that they’re not well-suited for X, then most of them won’t pursue X. And those who try to pursue X anyway will be met with pushback. Family will tell them to choose a path they’re better suited for, friends will find their behavior strange, and employers will reject them out of concern that they won’t be a good fit. Even if women are no less willing and capable than men in a particular field, a culture that insists otherwise can prevent most women from ever getting off the ground in that field.

To be precise, there are two types of generalizations. If you say that those individuals whose biological sex is male choose a STEM profession more often than those individuals whose biological sex is female, you’re making the first kind of generalization. It’s purely a matter of statistics. You’re not telling women not to pursue careers in the STEM field. You’re merely reporting a fact, and there are exceptions. Not every man and women follows the trend. But now imagine that you belong to a Christian church. You tell one of your Church friends a story about something you did. They look at you disapprovingly and say: “Christians don’t do that sort of thing!” This is the second type of generalization. Your friend isn’t making a statistical claim about those people who would self-identity as Christian. Rather, your friend is defining a Christian as someone who doesn’t do X. If you do X, you’re no longer a true Christian.

The first kind of generalization is always probabilistic when applied to individuals, while the second type needn’t be anything less than watertight. Even if most humans born male make a certain decision and most humans born female make a different decision, that doesn’t mean that any particular individual’s destiny is set in stone. They could swim against the current. But when you define a “man” as a person who possesses a certain attribute, 100% of men will have that attribute and 100% of people without that attribute aren’t men.

Sweeping generalizations of the first type are overly simplistic pseudoscientific nonsense, while sweeping generalizations of the latter type are true by definition, but perhaps harmful in their effects on society. In the former sense of generalization, you can’t say that every woman is X or Y. But in the latter sense, you can. What’s left to consider is just what the psychological and sociological impact is of proliferating such a definition in society. What happens when you say that men are pilots and women are stewardesses? That men are the breadwinners and women cook, clean, and take care of the children? That an unemployed man unable to provide for his family isn’t a real man? That an infertile woman without any children isn’t a proper woman?

Obviously such propositions aren’t attempts to give a dry description of statistical reality. They’re social-pressure-leaden cultural prescriptions. They’re not statistically incorrect, as they don’t ask to be taken literally. No matter how strong the gender roles are in a society, there are surely going to be some people who go against the norm. But you can still say that men do X and women do Y. Exceptions to the norm simply aren’t counted, as such people cease to be taken seriously as men or women.

English renders both epistemic description and cultural prescription with the same grammar. But they’re very different modes, and as such their propositions should be evaluated according to very different standards. English isn’t transparent about marking which is which. This leads to ambiguity, equivocation, and confusion.

Imagine that we say: “Women are attracted to confident men.” That’s an analytical description of the cause-and-effect reality of attraction. Although overly simplistic, it’s a useful model at times. To falsify it, you simply need to find a woman who’s not attracted to confident men. It’s an empirical observation whose truth value is at the mercy of empirical reality.

By contrast, consider the statement: “Men don’t cry.” This isn’t an epistemic description of reality, but a cultural prescription. If you encounter an adult male and then witness him crying, you haven’t do anything to falsify the statement. Saying that men don’t cry isn’t an analytical proposition about whether adult males perform a certain action. It’s a prescription for how adult makes should behave in society. You can disagree with it, sure. It may well be a silly relic of the from past generations. But you can’t falsify it with a contrary observation. If you find an adult male who’s crying, you don’t update your belief. You merely revoke his status as a man.

Although the former statement is an epistemic description and the latter a cultural prescription, there’s no indication in their structure that makes that clear. The two propositions are entirely different in function, yet they’re jammed into the same linguistic form.

To see how this leads to ambiguity, consider the proposition: “Chicks like to be told what to do.” Is this an epistemic description or a cultural prescription? Are we arguing that women are generally attracted to a strong man who takes leadership in the interaction, or are we implying that women who want to take control themselves aren’t to be properly considered “chicks” anymore, as they’re overly masculine and not worthy of being counted within the data set? If we observe a woman who wants to take the leadership role herself, should we update our original proposition? Or do we just kick her out of the data set for being a weirdo who’s not really a woman anyway? Do we dispassionately update our generalization to account for her in our model of reality, or is our generalization merely a thinly veiled arm of misogynist social pressure designed to get women to submit to traditional gender roles?

To make the distinction clearer, imagine that a customer at a restaurant looks at the menu and says to their friend: “This restaurant doesn’t serve steak tartar.” This is likely just an epistemic description. If the customer turns the page and sees steak tartar listed, then he would presumably change his mind. “Oh look, they do have it!”

But then imagine that one of the chefs at a restaurant that doesn’t have steak tartar on the menu decides to make it anyway with ingredients usually used for other purposes, in order to serve it to a friend who dropped by. The owner might notice and then react with anger: “We do not serve raw meat here!” Maybe he doesn’t have a license for handling raw meat in that way, which would make it a liability to serve steak tartar. Despite having the same linguistic form as the previous example, here the statement doesn’t function as a dry description of reality but as a prescription for social action. The fact that the chef did in fact serve raw meat doesn’t falsify the owner’s proposition, as it’s not an empirical claim. It’s a social command.

But why do we use the same generalization-like phrasing not only to describe what reality has historically been, but also to prescribe what we would like reality to conform to? Why are statements like “women are X” taken not only as a description of what is, but also as a prescription of what ought to be? How can a dry description of reality turn sexist upon colloquial interpretation? Why does English jam both descriptive and prescriptive meanings into a single type of linguistic form?

The reason, I believe, is that long-standing regularities in what people do end up baked into the social fabric deeply enough that continuing to do those things becomes a social obligation. A generalization about what people have done in the past easily shifts into a generalization about what people should do in the future, since an abrupt change in behavior can break the assumptions that the established social system relies on to work properly.

For example, let’s say that you go to a party on a Friday night. A friend of yours named Tom shows up and he brings a certain video game console and a few games. Everyone has a great time playing multiplayer games. The next Friday night, a party is held at the same house. The same people show up, and Tom brings the video game console again. People have a great time playing again, and the tradition begins to solidify. After a while, everyone simply makes the assumption: “Tom brings the video game console.” Others start bringing games for that console knowing that Tom will bring the console itself. That is, people start building their choices around the tacit assumption that Tom will continue doing what he was doing before. Tom will start to feel a social obligation to continue to repeat this action, as abruptly stopping will throw a wrench in the social framework.

Or consider the norm in Japan of taking off your shoes as you enter a person’s house. In Japan, people make the generalization: “Japanese people take off their shoes upon entering a house.” Of course there are Japanese people who don’t follow this norm, but the generalization is as prescriptive as it is descriptive. It’s a blanket generalization about behavior because it’s a blanket request for a certain type of behavior.

When you have a long-standing tradition of people not wearing shoes in the house, the structure of the houses may reflect that. The builders might use materials for the flooring that have certain advantages, but would be hard to clean if you walk on them with shoes you wore outside. The longer a group of people do a certain thing, the more them doing that thing will be baked into every aspect of the society as a foundational assumption. If people always wore shoes in the house, maybe people could now without being disruptive. But people haven’t worn shoes in the house for a very long time, so now everything is built around that. And thus the generalization that Japanese people don’t wear shoes in the house becomes the tacit request to not wear shoes in the house.

Ultimately, it’s conservative to equivocate between what people have done and what people should do. A conservative use of English takes empirical observations about what people generally do, puts it into the phrasing of a sweeping generalization, and then expects you to interpret it as a suggestion to fall into line with that trend. Decades ago, it was mainstream to say: “Men are the breadwinners and women are the homemakers.”

When people speak like this, it limits the range of thought in an Orwellian fashion. They don’t state: “Statistically speaking, most men take on the role of being the breadwinners, while most women take on the role of being the homemakers. But of course there are plenty of exceptions, and perhaps this trend will change in the future.” Instead they just decree that all men are X and all women are Y, effectively erasing the exceptions from the realm of consideration. The exceptions become nonpersons.

Conversely, it’s liberal to carefully separate description of the past from prescription for the future. A liberal use of English constantly assures the listener that there are exceptions, that it’s okay not to follow the trends, and that each person is unique. If someone argues that “men are X” and “women are Y”, the pushback will always be that it’s not okay to make sweeping generalizations.

With all of that said, I’d like to advance the proposition that conservative cultures encourage generalization, while liberal cultures discourage it. Modern American culture pulls out all the stops to prohibit generalization, at least about certain groups. If you produce a generalization about women or black people, you’ll be accused of countless sins. But within modern Japanese culture, generalization remains rather normal. Leftism has swept America, while Japan has largely maintained its right-wing ethnonationalist status.

A richly evolved culture of generalization supports an orderly rightist system where society prescribes each person a role according to various long-standing traditions related to what kind of people have a comparative advantage in what kind of action. Sure, it’s more than possible for a woman to be an excellent CEO of a major company, and it’s no less possible for a man to fulfill the traditional role of a housewife. But just because someone excels at something doesn’t make it their comparative advantage. On average, men are probably better suited to workaholic leadership roles, while women are probably better suited to a balanced life at home with the kids.

In such a culture, people don’t need to reinvent the wheel every generation. They don’t need to spend decades figuring out what they want out of life. Society simply takes as data their gender, race, and upbringing, and then leads them down a ready-made path based on what their comparative advantage is expected to be. Rather than embarking on an endless journey to find themselves, they need only follow their destiny. They need only follow God’s chosen path for them in life. As a result, they’re settled into a good career, married to a good man or woman, and competent within an important sphere of life relatively early.

While suffocating at times, it’s important to note that generalizations about men, women, whites, blacks, and so on ensure that most people find something that’s more or less their comparative advantage without much wasted time. It prevents people from getting lost on the winding road of soul-searching which seems to consume so many people in modern Western society.

On the other hand, a tolerant culture of anti-generalization supports an open-minded leftist system where the individual has more freedom to pick their own path in life. There’s less pressure to conform to any stereotypical trajectory, and thus there’s more room for people to get creative in finding their niche. People aren’t aggressively pigeonholed into simplistic categories based on outdated stereotypes. People refuse to be sorted into neat little boxes. Encouraged to break from the rigid lockstep of old-fashioned prejudice about who’s fit for what role, people are able to be more innovative about how their life is structured.

When traditional gender roles turn anachronistic, a leftist system that discourages generalization can help people break free from the shackles of old-fashioned thinking. It decalcifies old societal patterns. With less rigidity in how people see each other’s roles, society has the opportunity to shuffle around the pieces more freely. And with this freedom comes the chance for society to discover a better paradigm.

And while a rightist pro-generalization system can fall to suffocating rigidity, a leftist anti-generalization system can fall to a sort of drifting looseness. In the modern Western world, there are countless women who chose a workaholic career over a balanced family life because they thought it would make them more fulfilled in life. They thought they were breaking free of the patriarchal shackles of misogynistic culture. But now many of them have found themselves anxiety-ridden, depressed, and drifting through a meaningless life. They grasp at straws looking for true satisfaction. They find themselves roleplaying motherhood with their family of cats. For such women, it would have been better if they just heeded tradition. All of the rumination would have been unnecessary. They would have experienced a deep sense of meaning right out of the gates.

But then again, who can blame such women for having tried? The leftist drive is to look for novel possibilities in culture. Often the rightists will be able to say: “I told you so!” But that doesn’t mean the venture wasn’t without merit. You need to embark on a journey in order to know where it leads. Failure is a part of life, and there’s calculated risk involved in sailing into unchartered territory.