The Theory of Temperament

It is amazing how much emotional heat you can create by saying something like,

Humans have different temperaments, and genetic factors play a role in that.

Some people will read that statement and say that it is true and obvious. But others may read it and become indignant. They may see it as opening the door to racism and sexism.

The first part of the statement seems particularly difficult to deny. As we go through life, we believe that we can correctly make predictions about how different people will react in otherwise very similar situations. That is, we believe that we can predict the behavior of people (including ourselves) on the basis of what we have observed about their past behavior.

We know that these predictions are not always correct. However, we believe that they work well enough that we find it useful to make such predictions. In short, we treat people as having persistent inclinations to behave in certain ways. That is, we treat people as having temperaments. In everyday life, we operate with a theory of temperament.

But temperament is a slippery concept from a scientific perspective. Ideally, we would have instruments that measure temperament, studies that show us the causal factors that determine temperament, and studies that show us how people’s choices and performance are influenced by their temperaments. But the science is not that far along.

Dog Temperament

In the case of dogs, an organization called the American Temperament Test Society has a testing procedure that

measures different aspects of temperament such as stability, shyness, aggressiveness, and friendliness as well as the dog’s instinct for protectiveness towards its handler and/or self-preservation in the face of a threat.

The procedure involves putting the dog in various situations and coding the response according to certain criteria. I gather that the results of these tests are highly predictive of a dog’s future behavior. Of course, dog behaviors can be altered to some extent through training.

Temperament in dogs appears to have a very important genetic component. Genetic differences exist both across breeds and within breeds. The average golden retriever is less hyperactive than the average cocker spaniel. But it would be a mistake to claim that all golden retrievers are less hyperactive than all cocker spaniels.

What to measure,and how

When we turn from dogs to humans, it is less clear what we wish to measure. From the most ancient model (choleric, phlegmatic, melancholic, sanguine) to the five-factor model (openness, conscientiousness, extroversion, agreeableness, neuroticism) to astrological models to hormone-based models to “personality types” popular with corporate consultants and some psychometricians, there are many plausible ways to classify temperament.

Many of these approaches are binary. Supposedly, you are extroverted or introverted. You are melancholic or not. But it is likely that reality is more nuanced, with people placed on different points along a scale or spectrum.

Another challenge is to choose between general and specific measures. My grand-niece will try any food you put in front of her, but she is extremely reticent around new people. My grandson is the exact opposite. These differences would confound any attempt to generalize about their openness to new experiences.

If you like to let the data choose the level of generality, then you can employ a statistical technique known as factor analysis. This may be helpful, but factor analysis does not absolutely resolve the boundary between specific and general.

Another challenge is to find indicators that produce consistent results across different measurement instruments. Recently, I took the DISC personality test using two different online survey instruments. One showed that I was a strong “C,” and the other showed that “C” was my weakest characteristic!

Even if I my answers to surveys yield consistent results, there is still a question about whether my answers are reliable or deceptive. Experts establish the temperament of dogs by observing behavior, not by giving them surveys.

Still another challenge is that measures of temperament may not be consistent across time. On a scale measuring extroversion or some other element of temperament, you might now be different from what you were as a teenager.

Overall, few indicators of temperament are scientifically satisfying. The least problematic is IQ, in that it is consistently measured by different instruments and across time. The marshmallow test, which is intended to measure the ability to defer gratification, is another instrument that seems to have persistent predictive power.

Culture, race, gender, and temperament

Average measures of temperament differ by race and gender. Again, these are only average differences, and they only measure the difference between, say, the average man and the average woman. They do not show that there is a temperamental difference between every man and every woman.

Also, for some social outcomes, the average may be less significant than the extremes. For example, Jordan Peterson is fond of claiming, plausibly, that engineers tend to be people with a very high orientation toward working with things rather than working with people, and this inclines more men than women to go into engineering.

Suppose that, as Peterson reports, only 15 percent of women have the same orientation toward working with things as the average man. If the average man were inclined to go into engineering, then only 15 percent of women would be so inclined.

But consider that the average man does not go into engineering. Instead, suppose that the only men who go into engineering are those who are, relative to other males, especially oriented toward working with things. Very few women will be oriented toward working with things to that degree. This means that the proportion of women inclined to go into engineering will be quite small.

The end result may seem to suggest that women are under-represented in engineering because of cultural discrimination. But if culture is the driving factor, it must be the case that inclinations to work with things rather than people are set culturally rather than genetically. That may be difficult to demonstrate.

Is temperament culturally determined? Reading the work of researchers such as Robert Plomin, I am persuaded that there is a large genetic component underlying temperament. It seems that there is more compelling research pointing to genetic factors than there is research that would indicate that temperament is solely an artifact of culture.

Certainly, the manner in which the genetic component of temperament affects behavior is mediated by culture. If you were to embed identical twins in very different cultures, you would get very different behaviors. But by the same token, you see very different behaviors in people embedded in the same culture, and these behavior differences reflect differences in their genetic makeup.

Concluding Remarks

  1. Whether we admit it consciously or not, we believe in temperament. That is, when we observe and interact with other people, we take it for granted that differences in behavior reflect in part differences in inclinations that are persistent.
  2. Although we are drawn to binary classifications of temperament, it is probably the case that more often characteristics of temperament are arrayed along a scale of more or less, rather than either/or.
  3. The determinants of temperament are probably polygenic. That is, many genes and interactions among genes are likely evolved in shaping an individual’s temperament.
  4. Although we are interested in racial and gender differences in temperament, these differences do not predict individual cases. I may belong to a group that on average is more X than another group, and yet I can be less X than a particular individual within that other group.
  5. The science of measuring temperament is in a very primitive state. There is not good agreement on what and how to measure. I would say that IQ is an exception to that. Otherwise, the success of temperament indicators seems to be highly situational. Some indicators may prove useful for predicting particular behaviors in particular circumstances, but it is hard to find measures that are generally useful.
  6. There probably are important genetic determinants of temperament. But the way that temperament shows up in behavior is mediated by the cultural environment.