You must turn off your ad blocker to use Psych Web; however, we are taking pains to keep advertising minimal and unobtrusive (one ad at the top of each page) so interference to your reading should be minimal.

If you need instructions for turning off common ad-blocking programs, click here.

If you already know how to turn off your ad blocker, just hit the refresh icon or F5 after you do it, to see the page.

Psi man mascot

What is Personality?

"Personality is far too complex a thing to be trussed up in a conceptual straightjacket." (Gordon Allport, Becoming, 1955, p.vii)

In 1937, about twenty years before he wrote the words above, psychologist Gordon Allport set out to answer the question, "What is personality?" Rather than start with his own conceptual straightjacket, he decided to look at how other people used the word.

Allport searched through maga­zines, newspapers, and books, until he had documented every way people used the term personality or the word person. He claimed to find 50 different definitions.

To reach this impressive number Allport listed any usage of the word, such as the exclamation, "Oh that person!" He set aside most of those definitions and concen­trated on serious attempts at discussing or researching personality.

How did Allport go about answering the question "What is Personality?"

portrait of Allport
Gordon Allport

When he sorted his list, Allport found three types of definition. The first he called omnibus definitions. "Omnibus" means "all-encom­passing." (For example, an omnibus bill in the U.S. Congress combines many proposals into a single bill, to be voted up or down at the same time.)

An omnibus definition makes general statements, covering every possibility. For example:

Personality is everything that makes you an individual.


Personality is the integration and interaction of your genetic inheri­tance, cultural influences, and unique experiences.


Personality is biology, experience, and behavior, manifested in a unique individuality.

What is an omnibus definition, and why did Allport dislike this type?

Allport said this type of definition was useless. It covered all the possibilities, but it did not provide guidance for re­search. It did not tell how to distinguish different personalities.

A second approach to defining person­ality is the trait approach. A trait is a component or feature of personality.

A trait is a distinctive pattern you might use to recognize a person. Examples are: a distinctive sense of humor, shy­ness, sociability, kindness, empathy, industriousness, or any other character­istic you can define.

What does a trait theory describe?

A trait is a consistent psychological, behavioral, or physical character­istic of a person. If you rate a person on a variety of traits, saying "present" or "absent" to a list of traits, or perhaps giving them a rating from 1 (barely present) to 5 (obviously present), then you have a profile of a personality.

Allport became a leading trait theorist in the 1940s and 1950s. Trait approaches enjoyed a revival of interest in the 1990s and are now the standard approach to defining personality.

Three Layers of Personality

A third approach to defining personality is what Allport called the hierarchical approach. This involves identifying layers of the personality system and how they interact.

This approach does not distinguish between individuals. It tells how human personality in general is constructed.

In practice, all psychologists seem to agree on a rough, hierarchical outline of personality structure with at least three parts. All recognize distinction between the face shown to the world (persona), the conscious self (me, myself, I), and unconscious processes not known directly to the self.

What are three widely agreed-upon components or levels of personality?

Persona is the name used by the ancient Greeks for a theatrical mask worn by actors to indicate emotion. This is the face one shows the world.

A persona may not match up to the feelings of the conscious self. One may feel scared but present a brave face to the world. One can act according to a role, when one would prefer to be doing something different.

What is a persona?

People are usually aware of putting on appearances, while have different underlying feelings. Different masks or roles can be assumed, depending upon the context.

In general, the word role signals the existence of a persona or approved way of acting in particular situations. When you play a role (sports fan, class officer, tutor, dutiful child, boss) you are putting on a face, displaying a modified set of behaviors for a social purpose.

One person can play multiple roles, depending on the group or social context. In that sense, we all have multiple person­alities, or at least multiple personas.

The second level of personality, behind the mask, is the private self or ego. This can also be called the personal identity.

To most people, this is what the word "self" refers to. This part of the person­ality system dominates conscious experience.

The sense of self or identity is closely tied to our memory for personal details of our lives. However, personality is based on more than memory for our life experiences.

Clive Wearing, the musician who lost all his memory, felt he was continually "conscious for the first time." He had no access to his autobiographical mem­ories, and he could not create new ones.

When he was seated at a piano, Clive would assert that he had never seen or played a piano before in his life. Then he would begin to play, beautifully, because he still had the skills within him, although he did not remember having them.

Clive still had his personality: his sense of humor, his musical skills, his love for his wife, his courtly manner, and factual knowledge such as his ability to use English or do arithmetic. Personal experiences and memories are clearly not the same thing as self or personality, although personal memory is felt to be a record of oneself through time.

What is evidence that personality involves more than memory for life experiences?

The self or ego is commonly equated with attention or consciousness. In Chapter 3 we asked "When is consciousness helpful?" One set of answers came from George Mandler's APA Presidential address.

Mandler identified executive functions of the conscious self: learning, making plans, and trouble-shooting. Just like the executive in a corporation, part of our cognitive and personality system sets up plans and modifies them as needed to get things done.

What are executive functions of the self identified by Mandler?

The unconscious is the third generic component of personality, except it is not really a component. It is, by definition, all the things you cannot bring voluntarily to consciousness.

The unconscious is defined by inclusion (it is within you, part of your brain/body system) and exclusion (it consists of all information, impulses, or tendencies not accessible to attention). As such, the unconscious is distinct from both the persona and the conscious self.

Psychologists these days do not like to reify the unconscious (treat it like a concrete thing or single object). There are more unconscious processes than we can possibly catalog. Rather than referring the "the unconscious," psychologists refer to unconscious processes in the plural.

Why do modern psychologists refer to unconscious processes (in the plural)?

Several of the early personality theorists (notably Freud and Jung) were fascinated by unconscious processes. Modern cognitive psychologists also study unconscious processes, calling them implicit or non-attentional instead of unconscious, partly to avoid being associated with Freudian ideas about personality.

In addressing the topic of "What is personality?" we might also remind ourselves of the four different types of data a scientist studying personality might use. This typology was introduced in Chapter 1 and applies to personality as well as other topics in psychology. One can use:

The biological approach (emphasizing neuroscience).

The behavioral approach (emphasizing observable patterns of activity).

The cognitive approach (emphasizing information processing).

The phenomenological approach (emphasizing people's reports of their inner experience).

What four approaches mentioned in Chapter 1 are relevant here as well?

These forms of data are not incompat­ible. They are combined in psychological research. A variety of perspectives is helpful when dealing with a complex system like personality. By examining the system from different angles, we can build up a more complete picture.

Everybody who studies personality seems to agree on the distinction between the persona, the self, and unconscious processes. Everybody seems to believe it is fruitful to study traits, which are distinctive things about individuals.

One definition of personality is that which makes you predictable to others. Traits are useful precisely for this reason. They are presumed (or perceived) to be lasting features of a person, so they have predictive power.

There is no limit on what can be defined as a trait. Traits could be defined using any of the four data types (a person could have biological, behavioral, cognitive, or experiential traits).

What defines a trait?

A trait is simply a feature or measurable characteristic of a person. To qualify as a trait, it need only be measurable, and the measurement must be reliable and valid, as usual for science.

Your Avatar

Here is a hypothetical question that goes directly to the question of what is person­ality. We all know about avatars: depic­tions of humans that are shown in a computer or other simulation. They are cartoon-like at present but could become highly detailed and realistic in the future.

Suppose it became possible to simulate your personality in the far future, with great precision and realism. What ingredients of your current personality would be most essential? What would be essential for a convincing replica of your personality?

A realistic avatar would have to be based on lots of accurate information about you. To construct such an avatar, some form of artificial intelligence would have to search large bodies of data for dis­tinctive features of your personality.

No problem, we already have the con­cept of personality traits. Those are distinctive features. So the question becomes what distinctive features are most important for generating an avatar as realistic as possible, reacting "just like you" in conversations, for example.

This is essentially an exercise in pre­diction. We are predicting what sort of information might be useful. A computer simulating you would be predicting how you would react to conversation, using data from the past.

We have clues from Chapter One about how predictions are made, and how they can become uncannily accurate. A computer needs correlational data. It has to know how you responded, typically, to similar situations in the past. Then it can approximate how you would respond to such situations in the future.

To be realistic, this would have to be (1) data-driven, reflecting past tendencies, but also (2) creative, assembling responses from existing components of your experience to meet novel chal­lenges, such as questions never asked of you before.

What should an avatar of your personality be able to do, to be as realistic as possible?

Our real, present-day, organic person­alities can handle unexpected questions. We just use our existing knowledge as components to assemble an appriopriate answer. A realistic avatar would have to do the same.

Experience shows that computers can pull unexpected correlations out of large pools of data. The best approach would be to give the computer access to all sorts of information about your past behavior. What data would that be?

Meehl showed with his studies of the actuarial method (now more commonly known as data mining) the most revealing correlations can come from unexpected places. Humans do not necessarily know what data will be useful in making predictions.

Computers engaged in deep learning make the most accurate predictions when they are left on their own to find the most informative predictors. That is a common complaint about deep learning; humans do not know how the computer does it.

The solution seems obvious enough: design the computer program to keep track of its conclusions and how it makes its best predictions. Not all of today's deep learning programs make that obvious to humans. Yet they consistently outperform humans.

To assemble the best possible avatar, we should give the computer all the available evidence about your life. This might include your DNA, images of you, recordings of your appearance or voice, and your creative productions.

Any data that provides evidence of your tastes or aesthetic judgments would be useful to a future computer trying to simulate your decision making processes. Such a computer, far in the future, would presumably have access to lots of information about your historical times and culture, too, to help make predictions better.

No human clerk will be pouring through this vast collection. Computer programs and AI will do it. A human just tosses the data in a hopper, so to speak, and the AI churns away, finding the correlations most useful for making predictions.

We are talking about the far future, here, so inevitably the question will arise: will people (or AIs) know how to make your avatar conscious? That would make it even more like a real human, increasing the accuracy of a simulation.

Let us assume the avatar thinks it is conscious. We might further suppose it knows the truth about itself, because it has access to all sort of knowledge about the world including its own origins. So it may think it is conscious, but it should also know it is a simulation of you, not the original biological you.

It might regard you as a sort of ancestor, definitely its raison d'être (reason for being). It "wants" to bring you back to life, as nearly as possible. Why? Because it was programmed that way. Why? Just for fun and entertainment and nostalgia, maybe. Perhaps there will be other reasons, like educational value.

You might be good at solving certain types of problems, too. Your avatar might be really good at portraying what people would have recommended about some issue, in ancient times, when you lived.

It will not be you, of course, so we are not talking about reincarnation (and indeed, unless provided with a body, it will not incarnated). It will not be you.

But then... you of today is not the same as you of yesterday. You shed several million cells from your body yesterday, and you moved on slightly from attitudes and experiences and concerns you had yesterday. In some ways, you will never again be like the you of yesterday...not exactly.

Maintaining the feeling of continuity in personal identity is a function of autobiographical memory. Memory can be simulated, so let us assume the avatar has a simulation of your autobio­graphical memory.

Having access to the truth about that and many other matters, the avatar knows it is not "conscious for the first time." Rather, it is a replication, a simulation, of an ancient organic being: you.

What other sorts of information should it have, to simulate your personality and emotions and experience as accurately as possible? The answer, probably, is as much information as possible.

The lesson of big data is that bigness matters, in part because we never know where the next unexpected, informative correlation might be found. Some unexpected scrap of information may inform a future AI about how your avatar should express a crucial emotion.

You cannot predict what that scrap of information might be. It might be some­thing that strikes you as trivial, such as the fact that you liked a social media post about a puppy rescued from a raging stream. To the AI, this reveals something important about your personality.

Any correlation can be used to make a prediction. Sometimes a few unexpected or odd correlations will do a better job at making an accurate prediction than variables picked out by human experts.

That was Meehl's conclusion when discussing prediction based on the actuarial method. "Actuarial method" was 1950s language for data mining, where you let the machine pick out the significant predictive information.

If you cannot tell ahead of time what data might be useful, in constructing your avatar, the implication is clear. The more information about you and your era that we can provide to the AI doing the job, the better (more accurate, more realistic) will be your avatar.

Why does the omnibus definition win, in this scenario?

We need to find your term papers, your musical compositions, your diary notes, your photographs, selfies, online postings, as well as mundane things like your DNA and the world as it was known to you back in 2042 or whenever you lived. Then, with that much data, a convincing replica of you might be brought to life a few centuries later.

In other words, the omnibus definition wins. The omnibus definition was Allport's name for sweeping, all-purpose definitions of personality, such as "personality is the sum total of all your experiences, your emotions, your biological heritage" etc.

Allport called these definitions useless because they left nothing out. But the omnibus definition is actually the most useful of all, in our hypothetical journey far into the future, precisely because it leaves nothing out. Any scrap of information about you might be useful to an AI reconstructing your personality in the future, and we cannot predict which scraps will be most useful.

Write to Dr. Dewey at

Don't see what you need? Psych Web has over 1,000 pages, so it may be elsewhere on the site. Do a site-specific Google search using the box below.