An Update on ‘Face Reading’ (2014)

By Eva Gregersen

One of the current trends in Jungian typology is “face reading,” that is, the idea that people’s types can be determined by either their physical appearance or their eye movements or facial gestures. At CelebrityTypes, we have been roundly critical of “face reading” (here, here, here, and here). However, we have perhaps left the general reader with an overly negative impression of the merits of face reading.

There is still no evidence at all that “face reading” can be used to determine personality types, so if you dislike the mixture of typology and face reading there is nothing to worry about. However, in the interest of fairness we must state that face reading and physiognomy approaches to psychology are not completely useless, but have been shown to have some merit in three specific areas.

One is that people are seemingly able to predict the sexual orientations of strangers by simply looking at their faces. One line of studies has determined that sexual orientation can statistically be deciphered through facial information but has held itself back from determining exactly which facial features makes a person seem gay in the eyes of the observer. Meanwhile, another study has found that the faces of homosexual men are typically shorter and broader than the faces of heterosexual men (though this study concluded that people are not actually able to predict the sexual orientation of strangers by staring at their faces.)

The other is that people have been shown to be able to accurately predict survival-related traits such as a stranger’s attractiveness (reproductive potential) and aggression level (threat to own survival) based on just 40-100 milliseconds of watching that person’s face. There is also some evidence to suggest that less survival-related traits like intelligence can be detected by face reading, although this evidence is more scant. As the researchers conjecture, since survival and reproduction are the two most basic features of biological existence, it makes good sense that humans are able to garner some basic facts about each other intuitively, without the use of protracted intellectual analysis.

Third we find that people are able to tell a stranger’s political orientation by simply looking at a photo of that person’s face. The rational explanation for this ability, as proposed by the scientists John R. Hibbing, Kevin B. Smith, and John R. Alford, among others, seems to be that a person’s typical level of emotional expression is on average discernible even from static photos of that person’s face. On average, people who are more emotionally expressive at their base are also more prone to be left-wingers. And conversely, people who are emotionally inexpressive are – statistically speaking – more likely to be right-wingers. Thus the reason that people are able to predict the political orientation of strangers by “face reading” seems to be that left-wingers are statistically more prone to be smiling and appear relaxed in their photos (e.g. Alan Alda) while right-wingers will typically have less of a smile and more of frown (e.g. Clint Eastwood).

So it would seem that “face reading” is not completely useless. However, even when factoring in this evidence, we are still a far cry from the way “face reading” is usually meshed with Jungian typology.


As I have taken care to point out, all of these studies rely on statistical averages rather than one-to-one assessments of the sexual or political orientation of specific individuals. The evidence only supports what the scientists call “chance guessing with above chance accuracy.” But the way “face reading” is used in conjunction with Jungian typology, a 1:1 connection between a single person’s facial movements and that specific person’s type is usually postulated. However, there is no evidence to suggest that face reading works well when applied in depth as opposed to superficially. Beyond giving a certain quick hunch where the accuracy is better than random, face reading appears to be useless.

That is because an important finding in the studies is that “face reading” ceases to be of value after as little as 40 milliseconds. In 40-100 milliseconds, people get a quick hunch from the face that is better than random and further time spent analyzing the face is then no longer of value. This should be contrasted with the protracted research that the “face readers” of the Jungian type community make use of in order to arrive at their findings.

In fact, one of the studies cited below (Rule, Amady & Hallett 2009) actually found that reflecting on the impressions gained through face reading was bound to lead to worse results than merely going with the initial 40 millisecond snap judgment of a person’s face. So in any case, face reading is only supported by evidence when used superficially, subliminally, and instinctively. When we want to assess a person’s character in depth and with reference to intellectual concepts like Jungian functions, we must leave face reading behind.

It must further be said that none of these scientific approaches are used prescriptively, that is to say, no one would claim that someone actually is gay or left-wing because a bunch of strangers have asserted it to be likely based on his face. In both cases, we would seek to verify whom the person actually votes for and whom he finds arousing before venturing any judgment as to whether he actually is gay or is left-leaning.

In other words, we would never rely on face reading as a determinant in itself, but would do everything possible to verify the person’s preferences independently of face reading and with direct reference to the definition of sexual or political orientation.

In Jungian typology, the way to type someone with direct reference to the definitions involved is by meta-representing for oneself what goes on in the other person’s psyche through psycho-dynamic inference. But in the face reading community involved with Jungian type, people will simply claim that someone is a certain psychological type because of that person’s physical gestures and eyes movements.

So not only are “face readers” assuming 100% reliability of something that is most likely just “better than random.” They also let “face reading” data override the data that appears from typing someone with direct reference to definitions. Basically, their method is analogous to telling someone that he is a Democrat because he smiles in his photo while disregarding the fact that he has been voting Republican his entire life.

Regarding “face reading,” it will also be of interest to point out that while we are compelled to respect the scientific developments in the field, none of these innovations pertain directly to the core personality, but only to its derivatives. That is to say, there is still no evidence that you can tell a person’s preference for, say, Sensation or Intuition (and much less his cognitive functions) by staring at a photo of that person or by watching a video clip of that person with the sound turned off.

Finally, I would like to end by pointing out that mixing “face reading” with Jungian typology is not as new as its adherents often say. It was suggested and practiced by the Jungian James Oppenheim in the 1920s, but did not catch on.


Jacobs, Hibbing & Smith: Carrying Your Heart (and Your Politics) on Your Face: Ideology and Facial Muscle Responses Midwest Political Science Association Meetings 2012

Rule & Amady: Brief Exposures: Male Sexual Orientation is Accurately Perceived at 50ms Journal of Experimental Social Psychology 44 2008

Rule & Amady: Democrats and Republicans can be Differentiated from their Faces Plos ONE 5(1) 2010

Rule, Amady & Hallett: Female Sexual Orientation Is Perceived Accurately, Rapidly, and Automatically from the Face and Its Features Journal of Experimental Social Psychology 45 2009

Valentova, Kleisner, Havlicek & Neustupa: Shape Differences Between the Faces of Homosexual and Heterosexual Men Archives of Sexual Behavior February 2014, Volume 43, Issue 2

Willis & Todorov: First Impressions: Making Up Your Mind After a 100-ms Exposure to a Face Psychological Science 17(7) 2006


  1. Which communities, sites, etc of face readers are you basing the zero evidence claim on? Conversely what evidence is there that an approach using a persons works, quotes, interviews or preference tests can accurately predict a person type?

  2. We don’t understand the first part of your question. If there is no evidence, then how can we point to nothing? As for the second part, a professor of psychology just told us the other day that in 40 years of psychology, he had seen a new discussion about the scientific status of psychoanalysis a month. That question is probably never going to be resolved. It isn’t proper science, but nor is it entirely pseudoscience. There is, however, proof that psychologists and psychiatrists can type people according to the DSM types from short videos with them (like Jung’s types, the DSM types are also psycho-dynamic in origin).

  3. I’m not so sure about face-reading and political orientation. Compare your pictures for Sarah Palin and Vladimir Lenin, for example. Being left wing correlates with E and F, which correlate with emotional expressiveness. Correlation does not imply causation.

  4. Yes indeed. The point is that the method can hardly be applied to individuals as it is only a loose correlation. The claim is only that the hit rate of gauging political orientation from face reading is better than random.

  5. Let me first be clear that when I talk about face reading I am talking about body language only and not genetic traits, as appears to be what you are doing as well.

    My point was that since the process you use to type people isn’t based on a proven, evidence based scientific method, saying that another method (face reading) does not have any evidence is kind of pointless. Especially seeing as how you don’t appear to be looking at any specific examples like sites similar to this one, but based of face reading or anything else.

    It seems like the validity of face reading specific to typing a person would lie in either (or both) a sound theoretical principle that explains why it works or a high statistic amount of “correct” typings based only on face reading methods and nothing else. The problem with the latter is who decides what is a “correct” typing if there is no hard science to prove it?

    I’m sceptical of the possibility of reaching the same level of accuracy in typing an individual with face reading as opposed to traditional methods. But there are clear examples of correlations between body language and type. Enough to explore the possibility that functions might have clear visual cues that manifest subconsciously when they are in use.

    I would suggest taking a look at this site if you haven’t already:

    How accurate their methods are or how sound their theoretical framework is can be debated, but I wouldn’t dismiss it out of hand.

  6. It is not so much that the evidence is lacking as there is proof that psychologists and psychiatrists can type people according to the DSM types from short videos with them (like Jung’s types, the DSM types are also psycho-dynamic in origin). It is rather the fundamental assumptions of psycho-dynamic theory that are unfalsifiable.

    Since these assumptions are not removed from the system by anchoring functions in face reading rather than in psycho-dynamic inference, the fundamental problem of not conforming to genuine scientific standards persists.

    Sure, you can prove that some people tilt their neck at a certain angle more than others. But you can’t go from that to then being justified in postulating the whole nature of Fi. The nature of the functions will remain para-scientific the way all psycho-dynamic concepts are.

    Hence the advantage of psycho-dynamic inference over face reading is that, though the concepts assumed in both cases are ultimately unfalsifiable, inference is at least conducted with direct reference to the definitions involved and exhibits a consistency between method and concepts (both being psycho-dynamic), whereas face reading applies a behavioristic method to psycho-dynamic concepts.

    Now as you say, since psycho-dynamic methods and concepts are not truly scientific, that leaves the door open for anything better to come along and supplant that. But as we say, as long as this “better” approach also assumes psycho-dynamic concepts (such as the functions) as its end, they would never be able to prove that they actually *were* better. In principle, it may be that wholly unscientific approaches like looking into a crystal ball would provide us with better type assessments, but even if that were the case, the “crystal ballers” would have no method of proving that they actually were better as long as the ends involved are unfalsifiable. Therefore, the best we can do is to look at the results of the proposed method when applied to ends that *are* falsifiable, such as we have done with “face reading” above. Crystal balls seem to have no evidence going for them, and therefore no academically minded person takes them seriously. Face reading, it seems, *does* have something going for it, but as the studies consistently find, the merits of face reading are pretty much exhausted after 100 milliseconds, and using face reading in conjunction with intellectual reflection on the hunches gained by it is prone to lead to bad results. Hence the type of information that face reading seems to yield is not at all the type of indication we would need in order to supplant the usual method of protracted observation and inference with reference to unfalsifiable intellectual concepts. Even if you (or others) disagree with our assessment on this point, there is still the delicate matter as to *why* psycho-dynamic concepts should be kidnapped to serve other, incongruous aims and why people who are *not* psycho-dynamic in orientation have not been able to come up with other and better concepts. It stands to the credit of the psycho-dynamically oriented practitioners that the very concepts of E/I, S, N, F, and T were derived from their very method and had nothing at all to do with face reading or anything of that kind. But seeing as the concepts are ultimately unfalsifiable, it remains awkwardly true that none of these arguments constitute first rate proof and justification the way we would be able to conceive of proof if the end concepts actually were falsifiable.

    (As a side note, in order to conform more closely to scientific standards than the people who reject face reading, the face readers would have to reject cognitive functions and take a Keirsey-like, behavioristic approach to typology. At that point we must ask ourselves why we should use the Jungian type codes at all, since we could just as well use the Big Five, or other such inductive systems to suit a method of determining type [face reading] that is indeed inductive.)

    We looked at the database of typings that you suggested. Even assuming that we are always right (which we aren’t) it would be hard to gauge the accuracy of their method fairly because (1) it seems that some of their research is not extensive (e.g. Hitler ENTJ) – to gauge the merits of their method, we would have to assume equal degrees of research. (2) We don’t know to what extent they have relied solely on face reading. For example, they may have used the traditional method as a sidekick to the physiological basis of their method. Likewise (3) in their setup, the celebrities listed may be assumed to be known to the observer beforehand because of their fame. He may thus be influenced by his pre-existing knowledge of the celebrity, which should again be excluded from any viable assessment of their method.

  7. Thank you for the reply. Not quite sure on what to comment, this seems like one of those cases where you can talk it to death without getting anywhere.

    I agree more or less with most of your points, especially the last paragraph, although I’m not quite sure I understand the second to last paragraph.

    As always the end result is the most interesting. So if visual reading could better predict a persons type than say the best tests out there (like your own) or one or more “professionals” poring over a persons written works, history, quotes, etc or conducting an interview would you consider it legitimate? It would be possible to test this in a lab environment, although not realistically doable sadly :/

  8. Indeed it can be talked to death. That’s why we quoted the psychology professor who said that he had seen a new discussion of the scientific status of psychoanalysis each month for 40 years without ever seeing the question resolved.

    I (Eva) think that “face reading” as practiced on various sites where the observers most likely *also* listen to what a person says and unconsciously structure his cognition can easily supersede the accuracy of tests, including the official MBTI and also our own.

    Given the current state of the research on face reading as known from other fields, I don’t think it can supplant current methods of type assessment, like the ones we use on the site. In debates, even on this site, people often throw physiological information on the table as if it were a trump card. But if we assume that face reading can indeed say something about a person’s type, that information is hardly likely to be a trump card.

    But if face reading could be shown to be a reliable method in other, falsifiable fields and the obvious limitations of that method, as outlined in the post above, could be scientifically refuted, it would naturally have consequences for how we should regard face reading in relation to Jungian typology.

Comments are closed.