The universal trait hypothesis
Like many other psychology students, I was taught that “the various human traits and characters can be captured by five basic traits: conscientiousness, extroversion, emotional stability, openness-to-experience, agreeableness.” As someone who is not, by nature, very reflective, I never questioned what “can be captured” meant or how this caption was done. I later learned, when I started reading more on intelligence and early psychometrics, that the Big Five was also “captured” through factor analysis (which I have explained a little here.) But how intelligence started was that people had vague ideas about what it was supposed to be and constructed tests for it. They then have something concrete — test scores — to conduct factor analysis with. How do you have vague ideas about what personality is supposed to be? In any case, isn’t the resulting Big Five supposed to be those ideas? (So how do you have those ideas before deriving the Big Five so that the derivation isn’t circular?)
Since I’m writing a paper on traits, I’ve been reading on some of the histories. It turns out that modern personality was built on this thing that is called “the fundamental lexical hypothesis”. One statement of it is:
“the most important individual differences in human transactions will come to be encoded as single terms in some or all of the worlds’ languages”
– Goldberg (An Alternative “Description of Personality”: the big-five factor structure, 1990)
That seems reasonable… until the “most” is secretly dropped. 20th-century psychologists, notably Allport, Odbert, and Cattell, started gathering dictionary words that they took to describe “individual differences”. I haven’t read any primary literature on this, but it sounds like some researchers (especially earlier ones like Allport) categorized words by having coders (possibly themselves) group them by similarity. Later researchers, such as Cattell, have participants imagine a concrete person they know and rate this person with these words. Cattell then performed factor analysis on these ratings and came up with the 16PF. The 16PF are 16 factors that are not orthogonal (meaning that they correlate with each other), but Cattell preferred them for interpretability. If you require the factors to be orthogonal, you end up with 5 factors — the Big Five.
There was quite a bit of discussion early on about whether the Big Five is “universal”.
Whether such dimensions exist is an empirical question. There is no guarantee that thoughts, feelings, and actions must covary in the same way in all different groups. For example, it is unlikely that the same personality factors found in adults would also be found in infants, and it would surprise few social psychologists if the factors found in modern industrial societies differed from those found in hunting and gathering cultures.
– Costa & McCrae (Four Ways Five Factors are Basic, 1992)
It is an interesting thought when you think about it: the question isn’t about whether culture X is systematically less agreeable than culture Y; the question is whether it makes sense to describe culture X as agreeable or not-agreeable at all. One thing we can do is to interpret agreeableness on an interval scale where the mid-point means “not applicable”, kinda like how subjective temperature is colloquially understood, where the mid-point is “comfortable” (or neither hot nor cold). We can basically force a scale of this kind onto people and everyone will get a rating. But that’s not what these psychologists wanted, or else they wouldn’t be worried about universality.
What they wanted is that people who “are sort of in the middle in terms of agreeableness” are different from people “to whom the concept of agreeableness does not apply”, and they want to know if everyone is in fact not the latter. It’s “an empirical question”, after all. But how do you test it? You can give questionnaires to people, and they will give an answer. How do you know if they aren’t supposed to be able to give an answer?
What does it mean for a trait to not be universal?
I first learned about the taste umami in a psych class too (because somehow taste is important to the psyche). I was told that it differed from savory and there was no English word for it. I just checked on Wikipedia and it seems that the word “savory” is appropriated to describe umami now. Coincidentally, there is also no Chinese word (at least in my dialect, which happens to be Beijing-mandarin) for “savory” that isn’t identical to “salty”. I looked online and found that they appropriated the word “xian (鲜)” for this purpose. In my understanding, this word is pretty much exclusively used to describe seafood and, in particular, the quality of the ingredient rather than how a dish is prepared. So, pretty different from “savory”.
In any case, I was always very confused by this. Apparently, umami is one of the primary basic tastes in that we have special receptors for it. And it’s not like English or Chinese speakers don’t have these receptors. This seems to be a counterexample to the fundamental lexical hypothesis, no?
It’s possible that, while other people (non-Japanese) do experience the taste, it’s not that important to them. So, this is one possibility: maybe being agreeable or not just isn’t that important in some societies. It’s not one of the things that are really relevant when they try to describe each other.
However, even in this case, there is a sense in which the description applies; it’s just not very important. What does it mean for something to not even apply?
Here’s an example. There is a Chinese term used to describe people that’s invented fairly recently (20-30 years old), “xiaozi (小资)”. It seems to come from the French “petite bourgeoisie”, but the connotation is probably very different. Basically, it describes someone who is not super wealthy, but wealthy enough to care about quality of life. They are very particular about what they want and they tend to want things that are more distinctly Western. They watch French films, wear lesser-known designer brands, drink coffee, buy flowers on weekends. Like hipsters, they despise “mainstream” cultures. Unlike hipsters, however, these people tend to accept capitalistic ideals and value systems.
Linguistically, most Chinese people will have a stereotype associated with this word, and they more or less agree on whether a person is more or less “xiaozi” than another person. For all intents and purposes, this is a personality description. If I develop a test for it, then some people would score high while others score low.
Here’s the question: think about our hunter-gathering ancestors. Does it make sense to say that they also have a measure along this dimension, even though they probably didn’t have a chance of expressing it? In fact, think about people who lived a few hundred years ago in places without a lot of cross-cultural traffic. Does it make sense to say that they also had something inside them similar to what is measured by “xiaozi”?
It may be tempting to say yes in a rather trivial sense: if I gave my test to them, they would score something on it. If the fact of the matter is that they cannot express any behaviour that would prove or disprove the existence of this concept, isn’t it a meaningless dispute whether they actually (like, metaphysically?) have it? If the dispute is meaningless, then isn’t it easier to just assume that they do have it for the sake of theoretical consistency?
This is all well and fine, and I’m not someone who advocates metaphysical debates that aren’t empirically differentiable. However, recall that the hypothesis we were testing was precisely whether personality is the sort of thing that everyone has the same dimensions of. We wanted to know whether it made sense to talk about personality features like they were innate and universal. If we are unwilling to accept the apparent implausibility of applying some personality descriptions to some population, then we are engaging in blatant circularity.
What’s at stake?
In some sense, the question about whether everyone has the same personality dimensions really does not matter. If I’m never put in a situation where I can express agreeableness, there really isn’t any harm in assigning me a random number for agreeableness. However, the thesis that everyone has the same structure in X very often results in the automatic belief that X is biological/evolutionary. It is natural to reason: if everyone scores somewhere on trait agreeableness, then we can find out certain facts, such as “agreeableness increases with old age”; then, it makes sense to develop biological/evolutionary theories to explain why this is true. We see this inference happening in personality research where people worried about universality and then stopped worrying about it. We also see this inference happening in intelligence research, where people didn’t worry about universality at all. This inference may very well not be valid, but it certainly is not sound if the assumption that “X is universal” turns out to be false.
- It might happen after all - May 14, 2023
- Another job market data point - December 17, 2022
- Our place in the fediverse - November 30, 2022