Apr 19, 2013

Interesting Video about Jungian Types and Math Learning

A link to this TEDx presentation was recently posted by Jonathan in the comments of my "Fooled by Socionics" post.


My critique of this video will be similar to what I've been saying about typology for a long time.

It's really interesting to hear about individual differences that affect learning, and I will be excited to hear more about this. On the surface it would seem that the research conducted supports Jungian typology. Students are divided into types, and far-reaching differences in learning processes are discovered.

What this proves is that whatever algorithm was used to divide the students into groups (types) was to some degree predictive of their behavior when learning math. What is not clear is:

  • The relationship between the algorithm used and the theory of Jungian typology. In other words, what assumptions were made in producing the algorithm that do not necessarily directly derive from the theory?

    What I'm hinting at here is that the researchers' explanations of types may differ from those of other Jungian typologists, because the theory is basically unquantifiable. Different researchers might therefore end up with different divisions of schoolchildren with higher or lower predictive value.

  • Is the algorithm used by the researchers actually the best possible one for the purposes of this study? In other words, is there another way of dividing up the children that would be even more predictive of their behavior when learning math?

Occasionally socionists in Ukraine and Russia produce similar kinds of studies which are often tantalizing, but always inconclusive. For typologists in general, the focus is to generate empirical support for the theory rather than the best possible predictive outcome. I don't have much respect for this approach anymore. Maybe it's okay for the early investigation stage of new hypotheses, but it's not okay for a mature theory.

What I mean is that if we have a method "A" that we suppose influences outcome "B," then what typologists always do is — "let's apply our A and see what B we get." They never focus on maximizing the outcome! That would be — "let's see what A will maximize B."

This is probably the single most important idea to take away from my recent posts about why I have left socionics behind. This is what science is all about, and typologists — some bluntly, others more subtly — resist it for whatever reason (laziness and boredom, overconfidence, lack of acceptance of such research among their reference group, lack of research funds, etc.).

Going back to the TEDx video, let's imagine the research was not to "see what this way of applying Jungian Typology can predict" but to "see what method we can find for grouping children into learning groups with the greatest predictive power."

They could even start off using their existing typological method just to get things going. Then, they would look at how these children responded to different math learning methods and environments. Then they would consider whether they might move some children from one group to another based on their observed behavior, or whether there were large enough differences within a single group to justify dividing it further. Then they would try to formulate the key characteristics separating one group from another. They would have to decide whether these characteristics were more discrete or continuous and at what threshold level a person might be switched from one group to another. Finally, they would work on developing the most effective method for dividing children into these learning groups.

At some point, they would have to deal with the problem of the 99 percentile and 1 percentile math students who learn so much faster or slower than the other people with their supposed learning style that it is simply impractical to leave them in the same group.

The method produced might be a 5-minute test instead of a hours-long talking session on the differences between types. Or it might be a brain scan or physiological test done while performing different tasks. Maybe the behavior of interest is so complex that, for now, the only way to confidently identify it is in the classroom, over a longer time span. This might even lead researchers to say — "let's just use our talking session on Jungian types plus the following additional tweaks."

At any rate, the method would be optimized for enhancing math learning and would not necessarily predict any of the other things that Jungian typology is supposed to predict. But it would be rock-solid and other researchers could take it and build upon it.

Apr 13, 2013

Preview of My Changing Views on Personality and Relationships

(This post started out as a response to Jonathan's comment on my previous post, then grew to article length)

Part of why it may seem I'm going a little too far in turning away from socionics (I acknowledge that's possible) is that I haven't yet talked about the views that are taking the place of socionics in my mind. That's because I wanted to first do a methodical run-through of socionics before getting to my new perspectives, but it's going slow because I have almost no one to discuss it with these days. Hence this post, which is a kind of preview of my emerging ideas — "convictions," you might say — on phenomena previously described by socionics.

At this point, allowing myself to skip past some of the methodical stuff and cut to the chase will probably be useful to both me and my readers. Here are some of the main points in no particular order: 

1. Identifying people as "the same type" is useful only when they have significant obvious similarities. I'm done with sticking highly sensitive or highly intelligent/creative/refined/whatever individuals into the same type as people with a completely different background and sensitivity just because they share some esoteric "information preferences." If other people don't see obvious parallels between two people, then calling them the same type does more harm than good (this conviction comes from personal experience). 

2. This inevitably suggests a different set of types — either a similar number of types but with a much more uneven distribution, or a greater number of types. I personally don't care much anymore to name the types or create a consistent system such as socionics. However, identifying and describing the important common traits between two people I still find to be very worthwhile. 

3. The brain is not organized into socionic functions the way socionics suggests. The contents of our ongoing stream of conscious mental activity cannot be categorized by socionic function. In other words, the majority of the time, you will not be able to clearly relate what you're thinking about — or how you're thinking about it — to some socionic category. Rather than trying to force a socionics categorization, I'm more interested in just letting the information speak for itself and kind of self-categorize based on principles of pragmatism.

3.1. I would find it interesting to go through some of the music we once examined (back in 2007) using socionics terminology and allow different kinds of categories to emerge from that exercise. The questions I would start out with are, "what effect does this music have on the listener, and what does this music say about the composer's personality and state(s) of mind, and possibly the culture in which he/she grew up in?" Along the way we might discover that some music is just "better" than other, that level of sophistication is just as important a factor as the "states of mind" we were trying to describe using socionics functions, or that we come up with a number of states worth describing that does not equal 8.

3.1.1. I almost forgot to acknowledge, however, that Jung's and Augusta's idea of dividing up thought processes and basic traits into co-equal mental functions is a powerful and liberating idea that teaches one to see the other side of things and identify possible alternatives to nearly any approach to anything. That makes for a very useful mental tool, even if the details are not strictly correct. 

4. A result of point #3 is that interaction mostly does not occur on the basis of socionic functions. I believe that applying a general psychology/science perspective in examining specific cases brings one to different (simpler and better) conclusions than socionics about why people do or do not get along.

5.1. There is still something to functions and their impact on relationships. It's like they are bundles of values, but not actual mental processes or modules of information perception, processing, and output. 

5.2. There is still something mysterious about why people who are so different can sometimes get along so well. What socionics has done, however, is to put complementary differences on a pedestal. For the most part, people hang out with people of similar personalities who — perhaps — differ from them in some more subtle (and perhaps simpler and physiologically definable) way than being "a completely different type." 

6. Duality as described by Augusta is basically equivalent to falling in love. Remove the love, and duality is more mundane a phenomenon and barely preferable — on average — to other relationships. In modern culture, we expect and even require love for long-term relationships and generally prefer any relationship with love to one without it, regardless of the intertype relationship. There are good reasons for this coming from the logic of biological success.

6.1. Love does not obey the "laws" of intertype relations, and the idealizations that people project onto objects of passion do not necessarily come from the person's supposed Super-Id (dualizing) functions. That is, if you listen closely to what people want to have in an ideal partner — not just their conscious preferences but their emotional reactions to different people — I think you see that the preferences are 1) indeed significant and generally there for good reason; 2) not reducible to an "ideal dual"; and that 3) some people are [much] more universally desirable than others, and again, not because people are stupid and don't know what will make them happy, but because there are [nearly] universally positive and negative traits, habits, and life circumstances of great signifance to human interaction to which socionics is blind (see point #8).

6.1.1. It could be true that dual relations are more often magical than others. However, the fact that they tend to be described as magical feeds a tendency to see duals any time there is magic. Furthermore, a "magical" dual might have more in common with a "magical ___ type" than with a non-magical dual. 

7. There do exist particularly potent combinations of people that socionics is unable to predict (because its model of the psyche is not just incomplete, but fundamentally incorrect). The cause of the potency is unknown to socionics but can be discovered on a case-by-case basis and then, perhaps, generalized into a set of patterns and general rules. I would speculate that these combinations are highly symbiotic on grounds that are more or less permanent (constitutional) rather than situational (e.g. your specific current needs complement the other person's). Such pairings are consistently able to elicit symmetrical positive emotions in each other. Why? I don't know, but I'm now quite confident it is not socionics.

7.1. People with more extreme traits may have different patterns of interpersonal compatibility than those with traits closer to the norm. In particular, they may experience incompatibility much more often and have greater unmet needs for understanding and connection. The causes of this are fairly straightforward and obvious and probably have little or nothing to do with socionics.

8. There are many very important universals that socionics is blind to. For instance, that some people are almost universally annoying while others are almost universally liked/admired. There are not 16 strategies of success, but rather just a handful, plus variations. People achieve success not only by relying on unique strengths, but also by developing universal qualities common to most varieties of success. Trying to follow a somewhat contrived and esoteric type-specific path to success seems to me generally less useful than working on improving the universals.

8.1. For the most part, our lives are dominated by universals — standard situations that would elicit similar reactions in most people. Socionics suggests the opposite.

8.1.1. However, there are definitely times when a person needs to focus on discovering and enhancing the individual and specific. Socionics can give some broad hints, but nothing more. So can other typologies. So can non-typological empirical psychology.

8.2. There are "better" and "worse" states to be in that are remarkably universal, but these states are hard to describe from a socionics perspective. "Type development" is a clumsy and unparsimonious way to describe it. Modern science now has a ton to say about what contributes to happiness and well-being, and research results don't obviously suggest a typological approach.

9. The territory that we each stake out in life and build our self-identities upon is mostly based on unconscious calculations of our probability of biological success in the various roles/niches we have tested or know of given our existing investments and available energy and resources. These roles are often continually changing as different actors come and go and resource allocation patterns (i.e. the economy) evolve, since these things affect our personal prospects. While there may be patterns in which personalities gravitate towards which kinds of roles, situational and non-typical factors are generally more important. Since our connections with people are to a large extent determined by these roles, the things that bring people together and create a bond are best described using non-socionic language because socionics plays a small part in it.

9.1. I think lasting interpersonal conflicts can be parsimoniously described as reflecting threats to self-identities, current roles, and biological success. If you are a sadist and want to cause people psychological pain, don't nerdily attack their hypothetical "point of least resistance"; instead, doubt their self-identity, jeopardize their current roles, and question their biological success. 

- - - - - - - -

That's a glimpse of where my ideas are taking me.

Back to Augusta...

An important question I ask myself is, if things really are as I have suggested here, how can Augusta and so many followers have thought for so long that the system worked? I think that a key to the longevity of many not-quite-true (I don't mean that perjoratively) idea systems is their complexity. The structure and sophistication of systems of thought often have the effect of aweing their adherents. If the ideas are complex and extensive, they may take a very long time to prove or disprove. I, for one, only feel confident in calling socionics "inaccurate" as of the past 6 months.

A long-lived idea system must also be at least reasonably accurate at some level — or impossible to disprove. The fact that thinkers have been identifying types of people for millenia suggests that there is some basis for this approach. It is also plainly evident that different combinations of people are more or less compatible, and that once-established relationships and attitudes toward each other often last for a long time. However, our limited consciousness seems incapable of understanding why this is the case without the help of science. One of Augusta's errors (and mine too) was overstating the role of permanent, constitutional differences in personality. Her model treats people as essentially unchanging actors and isn't well suited for describing short- and long-term development. This error, as well as its opposite — that everything is situational — are easy to make if you are unaware of the last 30 years of psychology research.

Augusta attempted to explain phenomena which are still clouded in mystery — namely love and interpersonal attachment/rejection patterns — and did a half-decent job at it. Good enough to make a lot of people excited about the discovery. Because there are so many types and varieties of intertype relations, a place could be found for any person in the universe revolving around any particular person. To find points where the system breaks down requires comparing overlapping universes to see how the predicted relationships and perceived personalities play out. Like I said, this can take a lot of time — years and years — as one tries tweaking typings and one's understanding of socionics to see if things can be made to fit after all. 

Apr 9, 2013

"Fooled" by Socionics

This long and rambling post was inspired by this comment from Consentingadult.

I have a suggestion: instead of criticizing Augusta (which in itself is perfectly valid), why don't you actually criticize one or more of your own writings? E.g. the ILE description in the articlewhere you said, amongst other things: 
"I have known ILEs who were heads of research institutes, NGOs, language teaching schools, and consulting companies. Many more held various positions in all kinds of organizations where they had a great deal of independence. The common feature of all engaged ILEs seems to be that they are working on some far-reaching personal project that has yet to come to full fruition."
I would like to know: what was wrong about this, how did you fool yourself to arrive at such insights that you now claim cannot be true or valid?

I'll get to criticism of types eventually. I was hoping to start with information aspects, then move to functions and Model A, then to types and intertype relations.

If your question was not simply rhetorical (since I detect a bit of sarcasm), here is my response.

What's "wrong" about the ILE description

I described a subset of ILEs — those who are above-average, proactive, and highly engaged in the modern, growing economy. If I'd written something that described the under-average, reactive, and underengaged ILEs, or the subsistence farmer and hunter-gatherer ILEs equally as well as the more fortunate, modern ILEs, the resulting description would have been too abstract to "do any good."

One might argue that I've described the "greatest potential" that people of this type can attain to. I agree that that might be inspiring, and people want and need to be inspired. But it's not necessarily true that you can group people into 16 types and then expect that any of those people has the potential to achieve the level of success, fulfillment, or happiness of the most fortunate individuals of their type.

I described the ideal type, not the borderline, overly contradictory, or untypeable ILEs. This is a universal problem in socionics, and it comes from being unable to type all individuals in a population. If socionists can only type what they themselves can recognize, then they only describe the most recognizable types. Then other people will perpetuate the bias towards recognizable types and intertype relations.

I don't expect that socionics will ever overcome this problem because it is not sufficiently accurate to ever develop foolproof empirical tests. An "ILE" will always be "what I've defined as an ILE," and thus any description is as good as any other.

Why do I use the word "accurate?" Because I am convinced that no matter how you define and split the 16 types, the resulting intertype relations will not be close enough to those postulated by socionics to call the system "accurate." The reason socionics has a "lack of empiricism" problem is simply because it has an "accuracy" problem. If socionics were more accurate, the empirical data would be flowing in all around. For instance, the phenomenon of duality would be scrutinized and elucidated, and researchers would be studying it from multiple angles to learn more about how and why it happens.

The fact is, socionics duality as described by Augusta is a rare subset of dual relations, and when it happens between a man and a woman, there is a lot more going on that has nothing to do with socionics. And it doesn't necessarily happen just to duals. In my mind this is a fatal blow to the socionics model of the psyche, since duality was socionics' most powerful hypothesis.

How I managed to "fool myself" 

A bigger question to ask is, "how did Augusta manage to fool herself?" I'll speculate about this further down.

In my case, my initial insights were simple and promising enough that I continued my study of socionics. I had only really deeply processed a small number of my personal relationships, but my insights were promising enough that I was hopeful that socionics could be applied generally, and not just in these specific cases. Of course, I based my understanding on my experiences of the most easily recognizable types and relationships in my environment.

At the moment I wrote the ILE description, my arsenal of experience had grown from a handful to maybe a couple hundred people of different types. However, my deeply processed personal experiences relating to each of them lagged behind. Based on my experiences so far, I was willing to take the system on faith to a degree. When eventually forced to face major contradictions, I began to systematically consider the less easily definable types and relations that I had been discounting and give them an equal place in my mind alongside the recognizable ones. After a while, I concluded that socionics was not accurate enough as it should be, considering the large amount of mental storage space required to house it. Furthermore, the things it could tell a person (less than I used to think) were of little use to me at this stage of life. This process provided some new insights that I now feel obligated to share with my socionics audience.


When one begins to get into socionics, one recognizes only a tiny sliver of types and intertype relations — those that most obviously fit descriptions and can thus be interpreted thoroughly from a socionics perspective. These "obvious fits" then form the foundation of one's understanding of socionics, giving the person a bias from day one that they will be unable to correct until much later, if ever. If a socionics aficionado recognizes the limitations and inaccuracies of socionics from the outset due to a more methodical and less exciteable personality, you can be sure that they will never become socionics writers!

Socionics is, in a way, worse off than astrology. With astrology you at least know when a person was born, making disagreement on their astrological type impossible. Then you can read a description and say, "well, that doesn't fit!" In socionics, if a person doesn't match their type description, then you've probably mistyped them. Keep switching types until you find the "best fit!"

Changing times

Yes, this has been an absolutely normal way of thinking about psychological matters for eons, but the times are changing and this approach is losing popularity as science has more and more to say about things that people used to only be able to speculate about. More and more people are sensing this and jumping on the train of empirical psychology and neuroscience. This is one of the reasons socionics is, in my opinion, slowly dying out.

Speculation about Augusta

I am quite certain that if Augusta were born today, she would not create socionics. She would read everything currently available on individual differences and go into psychology research or neuroscience. But she grew up before most of the really interesting research started being done, and was stuck in the Soviet Union where there was a official slant on psychology that downplayed personality differences. Thanks in part to typology, personality differences are now taken for granted.

Augusta created a system that allowed for much more healthy psychological variability than was commonly accepted at the time, but she was quite absolutist in her views on the types. For instance, she published a paper explaining why ILE is the best type to run a scientific research institute, ignoring other significant factors other than type. It seemed that she was proposing a new way of staffing posts and organizing society that could be implemented within the Soviet system.

Obviously, the zeitgeist has changed dramatically since then, even in the former Soviet Union. As I understand from my interaction with the socionics community in Kiev, in the late 1980s Augusta's interests drifted to other areas, and she became a less and less active participant of the socionics community, while continuing to occasionally voice her views on the subject. I was also told that she was in a rather unhappy marriage with an LSE (she was ILE), which could have contributed to her idealization of duality.

Was Augusta "fooled" by her own ideas? I suppose you could say so. She would have had the same problems with confirmation bias that any later student of socionics would have, alleviated by the process of discussing ideas and hypotheses with a group of associates. They noticed some fascinating new patterns, broke new ground, and made some mistakes that are easier to recognize decades later.

Watering down socionics

If socionists were forced to type everyone and, particularly, to examine every single relationship among these people, and then to describe types and intertype relations using a strictly statistical method, I am certain that we would see a drastic watering-down of both type and relation descriptions.  Socionics would lose its "teeth."

However, the more watered-down one's views on a subject are, the less motivated one is to propogate one's views. This is why the people who have built up the experience to help water down socionics usually can't be bothered to do so. I personally view this as my duty to society, but my motivation is much weaker than it was to promulgate classical socionics, which engaged my need for recognition, connection with others, and mental discovery. Now, the inspiration is just not there. There's no "vision" to be found by picking apart someone else's somewhat erroneous theory to a dwindling and disinterested audience. In fact, the main reason I am writing this post is probably to avoid doing my taxes.

From a kind of esoteric psychology perspective, it may be good to keep socionics classical, albeit full of illusions. People come through the system, receive rigorous training that excites them and fills them with energy, and they move on to do other things. But it's already too late; socionics is dying a long, slow death. Successive generations of socionists are less and less inspired and inspiring, career momentum and entropy are taking over, and there is just too much new and exciting information on personality and human interaction coming from research that takes a fundamentally different (empirical) approach than socionics.

Apr 8, 2013

Reexamining Socionics: Information Aspects

This is a continuation of my effort to critically examine the foundations of socionics, which I began in January.

Information aspects (note: NOT "elements of information metabolism," i.e. "mental functions") are a not-terribly-productive construct introduced by Augusta as she developed her unique perspective on Jung's Typology. Instead of being merely "modes of processing information" as in Jung's Typology, the psychic functions became instruments for "perceiving, processing, and conveying" information. Augusta saw the functions as responding to different streams of information coming in from the outside world. A stream of information could then be labeled by the corresponding Jungian function that processed it. The extraverted sensing-function processes extraverted sensing-information, and so forth. One person's extraverted sensing responds to extraverted sensing-information and conveys extraverted sensing-information to the outside world, and another person's extraverted sensing picks up that extraverted sensing-information and responds to it in a way defined by the position of extraverted sensing in his socionic type. The response might tend to be confident and authoritative, it might be curious and accepting, helpless, narrow-minded and categorical, etc. depending on the person's type.

I say that information aspects have not been very productive as a construct because there is really not much to be said about them other than "extraverted sensing-information is what the extraverted sensing-function perceives and produces," etc. They can't be defined in isolation from psychic functions; something is only "information" if it is a message that is potentially perceivable by a human observer. Augusta gave the information aspects more abstract definitions (see here, for instance), almost suggesting she first divided up reality and types of information, then found that her division matched Jung's functions. I'm certain it was actually the other way around.

Augusta's descriptions have been revised and concreticized by subsequent authors to more closely match how people of different types perceive the world in practice. And, to be honest, nothing much has been said or done about information aspects (also called "information elements") since Augusta wrote about them. Usually, they are simply confused with IM elements (elements of information metabolism), which are another name for the Jungian functions. In fact, most professional socionists even confuse them (another symptom perhaps?).

So, is there any use in dividing up information into different types? I think yes, if the division is arrived at empirically — rather than off the top of one's head — and improves understanding of the real world. In the case of socionics it appears that neither is true. The kind of division Augusta suggested is far from obvious. If one were to attempt to categorize information coming from a general science perspective, one would probably take a different approach:

  • By sensing organ/receptor: light and visual information, sounds, scents, tastes, tactile information, various bodily sensations, emotions and moods, etc.
  • By relevance to the perceiver: important/unimportant/potentially important, pleasant/unpleasant/neutral, etc.

If you had to divide up information, how would you go about it?