Temperament and Big Five: What Science Kept of the Four Types

A corporate workshop ends with eighteen people reading out their results to each other. Two of them came out as melancholics. One is an accountant who has her budgets finished three days before the deadline and color-coded blocks in her calendar. The other is a designer who submits the night before the cutoff, forgets an invoice twice a year and comes back to his work only when something grabs him.

Both got the same label and they have almost nothing in common. And this is not a fault in the test. The ancient typology works with two of the five traits today's psychology measures and leaves the other three completely open.

Eysenck's two lines

In the 1950s and 1960s Hans Eysenck did something with the four temperaments that had never occurred to Galen. He was not interested in whether bodily fluids stood behind character. He was interested in how many independent traits we need in order to describe the differences between people, and his analyses produced two.

The first is extraversion. It describes where your attention goes and where you take your energy from: whether company recharges you or drains you, how quickly you move into action and how much stimulation you need around you to feel good.

The second is neuroticism, or emotional reactivity. It does not say whether you are sad. It says how strongly and how long an unpleasant thing knocks you off balance. At one end of the scale is the stable person for whom criticism in a meeting is over by lunchtime, at the other the one who is still replaying the same sentence on Friday night.

Cross those two lines and you get four quadrants, and in them sit the four classic temperaments. The sanguine is a stable extravert, the choleric an unstable extravert, the phlegmatic a stable introvert and the melancholic an unstable introvert. The word unstable here is a technical term for higher reactivity, not a diagnosis and not something that can be read as a defect.

Eysenck acknowledged his debt to the ancient Greeks himself and drew the four temperaments right into a circle around the two axes in his books. That was not deference to tradition. He was led by the idea that something bodily lies behind the difference between an introvert and an extravert, specifically a different level of arousal in the nervous system: whoever has a higher baseline needs less stimulation from outside in order to have enough. Later research rewrote the details of that hypothesis, but the two dimensions themselves stayed.

The history of those four names from Hippocrates (around 400 BC) to Galen, who named them six centuries later, is traced in the overview of the four temperaments. Eysenck's contribution was that instead of four boxes he offered two continuous scales. A person can stand anywhere along them, including exactly in the middle.

From two dimensions to five

The next generation of researchers took a different route. Lewis Goldberg (1990) built on the lexical hypothesis: whatever matters in the differences between people has been given a name by language, so it is enough to collect all the adjectives that describe people and see which of them travel together. Paul Costa and Robert McCrae took care of the measurement side in the 1980s.

Those analyses repeatedly produced five dimensions instead of two. Extraversion and neuroticism stayed, and openness to experience, conscientiousness and agreeableness joined them. This five-factor model is used today as a shared language for describing personality, and its chief advantage is boring but important: it comes out the same way again and again across different languages and cultures.

What do the three new dimensions measure? Openness to experience describes an appetite for the new and the unusual, from travel to abstract ideas. Conscientiousness concerns order, planning and seeing things through. And agreeableness says whether in a conflict you instinctively look for an arrangement or go after what you want even at somebody else's expense.

Let us go back to the accountant and the designer from the workshop. Both can be unstable introverts, so both land in the melancholic corner. They differ in conscientiousness, and that simply does not exist in the ancient system. What makes this interesting is that conscientiousness is, according to the meta-analysis by Barrick and Mount (1991), among the best personality predictors of job performance across occupations. The four temperaments say nothing about it at all.

The map: temperament and its Big Five profile

The table shows how the four temperaments look translated into five-factor language. It is an approximate conversion, not an equation.

Notice the empty spaces. Openness to experience does not appear in the table even once, and neither does conscientiousness, because neither follows from Eysenck's two axes. The notes about agreeableness in the choleric and the phlegmatic already go beyond the original model and come more from observing how those types are described in practice.

That a measurable relationship exists between the classic temperaments and the five-factor model was also confirmed by work published by Strelau and Zawadzki (1995). The links come out mainly for extraversion and emotional reactivity, which matches what the ancient physicians were noticing. On the remaining three dimensions the ancient typology simply has nothing to offer.

What the typology leaves out

The first simplification is that dimensions are continuous whereas a type is a box. Dividing people into introverts and extraverts creates the impression of two groups. In reality most people cluster around the middle of both scales, so the largest group is the rather uninteresting one that is a bit of both.

The second problem follows from that. For somebody near the middle, the final label is decided by a difference of a few points, and those points can shift depending on whether you take the test after a good weekend or on a Tuesday after an argument. The further from the middle, the more stable the result. This, incidentally, is why people with a pronounced profile find the description of their type so accurate while everyone else shakes their head.

The third thing is the probabilistic nature of the whole description. The sentence "melancholics are meticulous" means, in the data, roughly that you will find somewhat more meticulous people in the melancholic corner. It does not mean your melancholic friend has ever handed anything in on time.

What does that look like in practice? Two people join a development team and both come out as phlegmatics. The first stays calm because nothing rattles him, and he also keeps a detailed task list, so he can be relied on to the letter. The second seems equally calm, except that his composure comes from not finding deadlines important. A manager who runs them the same way will be in for a surprise with the second one within a month. The difference between them does not lie in temperament but in conscientiousness, which none of the four ancient descriptions covers.

Now try to picture the ten people you know best. How many of them would you place, without reservation, into one of the four boxes? Most people arrive at about half, and that is an honest estimate of the resolving power of the whole system.

Why the types survived anyway

If usefulness were decided by accuracy alone, the four temperaments would have disappeared sometime in the 1980s. They did not, and the reason is a quality psychometrics does not measure: comprehensibility.

Say "he's a typical choleric" and everyone at the table immediately pictures specific behavior. Try the same with "high extraversion, higher neuroticism, lower agreeableness" and half the room disengages at the second word. A type works as a name you can use in an ordinary conversation, whereas five numbers in percentiles are a tool for somebody who knows how to handle them.

The second advantage is that a typology describes the whole person at once. The five-factor model gives you five independent readings and you have to assemble the combination yourself. The four temperaments carry that combination inside them already, and when you add a streak to the primary component you get sixteen profiles that cover contradictions within one person too. How two components play off each other is shown in the overview of temperament combinations.

And third: a type is memorable. You will forget five numbers from a questionnaire within a month, but the fact that you are a phlegmatic with a choleric streak will come back to you a year later, when somebody in a meeting asks you for a fast decision again.

How to read the two together

The most sensible approach is to treat temperament as the front door and the five-factor model as the detailed map. Our temperament test has 24 questions and takes about four minutes; the result is a primary temperament, a streak and a note about the component that came out lowest. It can then be read as a name for what you have long been observing in yourself.

When you want to go into detail, the Big Five test shows your position on five independent scales, including the three the ancient typology says nothing about. That is exactly where the difference between the accountant and the designer from the opening gets decided, because both can have the same temperament and yet a different relationship to deadlines, to rules and to new things.

Both comparisons are approximate. Neither of those tests is a diagnostic tool and neither says what you are suited for and what you are not. They help mostly where you need to name why a certain kind of work or a certain kind of people sits better with you than another. The concrete method of observation is covered in the piece on how to find your temperament.

One thing about the ancient typology deserves notice even two thousand years on. Neither Hippocrates nor Galen had statistics, questionnaires or samples of thousands of people. They had only long years of observing patients, and even so they hit two dimensions that are among the first to jump out in every modern analysis. The remaining three waited another twenty centuries to be discovered.