Oddly Influenced

Welcome to oddly influenced, a podcast about how people have applied ideas from *outside* software *to* software. Episode 23: Personality and destiny.

Maybe 15 years ago, I stumbled across John M. Doris’s 2002 book, /Lack of Character: Personality and Moral Behavior/, which introduced me to situationist psychology and the “person vs. situation debate”. The question under debate was: “how much of behavior is driven by personality and how much is driven by the situation or environment?” Now, any good consultant would know that the answer is “it depends”, and indeed that seems to be the consensus after some 40 years of what seems to have been a fairly spicy debate. For example, here’s what a 1990 paper surveying the history of personality trait research says about the other side: “a strong attack was launched upon the entire field of trait research by Mischel (1968), Peterson (1960), Ullmann & Krasner (1975), and other born-again fundamentalists, who excoriated trait theory as akin to scientific sin.” I’m sure all the participants in the debate got righteously angry and so had a grand old time.

It’s tempting to dismiss all this, especially because psychology has been having a hard time lately, what with famous results not replicating and all. However, the studies are interesting and I’ve personally found them useful in explaining some of my own experiences. Specifically, why I have sometimes been successful at doing things that don’t match my personality… either my personality as I conceive of it or as personality tests summarize me.

The message of this episode is, mainly: your personality may not constrain you as much as you think. Be wary of assuming you know how you’ll react to a novel situation. And: sometimes small adjustments to your environment will make a surprisingly large difference.

Note: I am not a therapist. I’m not going to give you life advice. I’m summarizing some academic literature and, to provide examples, extrapolating to my own life. I’ve personally never cared for software consultants who try to “fix people”, and I don’t want to be that guy in this podcast. I’m the “hey, maybe this is worth trying” guy.

≤ music ≥

In at least the modern world, it seems natural to divide personality into a list of traits, each of which can be represented as a point on a number line. Sometimes, as with Myers-Briggs, the endpoints of the scale are labelled, so you can be more toward the “judging” end or more toward the “perceiving” end. The so-called “Big Five” personality traits have one name for a whole scale: So you can score high for “openness to experience” or low for openness to experience. Myers-Briggs is popular in business; the Big Five is popular among psychology researchers, who generally regard Myers-Briggs as snake oil. This episode doesn’t have to pick a side, so I won’t. The studies I’ll describe apply equally to both. They also apply to so-called “folk psychology” definitions of words like “extroverted”, “conscientious”, “helpful”, and so on.

So I’m going to leave aside personality classifications for now. I may add a bonus episode, because their nature and what it’s valid to say using them is interesting.

≤ short music ≥

The great grand-daddy of situationist studies was Newcomb’s 1929 /The Consistency of Certain Extrovert/Introvert Behavior Patterns in 51 Problem Boys/. The study was done in a summer camp. At this time, the word “extraversion” had conflicting definitions (as it still does today), having only recently been popularized by the psychiatrist Carl Jung. The study consisted of camp counselors (who just happened to also be psychology students) recording behavior during certain activities. For example, after boys organized a game, a counselor might tick off one of four possible behaviors for a particular boy:

* Insisted on doing the organizing himself.
* Gave constant advice to the leader
* Helped to plan, but loyal to the leader
* Let others do all the planning.

The assumption was that an extroverted boy would have one of the first behaviors, but an introverted boy would probably let others do all the planning. The counselors also observed what percent of time a boy talked at table (5% or less, 15%, 33%, or 50% or more) and what percent of the time did he play or work alone (using the same categories). There were a total of 30 activities monitored. The boys didn’t know they were part of an experiment, and they weren’t even told that they were all “problem boys” (although some guessed). Aside from the covert record-keeping, this was an ordinary summer camp, and most wanted to come back next summer.

I bet most people would say a person who monopolizes table conversation tends toward extraversion, as does someone who almost exclusively plays in a group. So the measurement of the first for a boy ought to do a fairly good job of predicting the second. But, quoting Mischel in 2004, “Newcomb was shocked to find that the average correlation coefficient based on daily behavior records across the situations was about 0.14.” When I gave a talk on these results at XP Universe 2011, I made a slide of what a scatter plot with a correlation of 0.14 looks like. It’s in the show notes. I defy you to see any relationship between X and Y.

Newcomb’s last sentence in his monograph is “And if [this work] points in the direction of a more individual technique instead of a dependence on [using] type psychology for educational and therapeutic purposes, it has perhaps been worthwhile.” Well, that matches my biases, so: case closed.

Oh, all right.

It’s expensive to do such studies, but there have been replications by those “born-again fundamentalists.” For example, Mischel and Shoda looked at a more recent residential camp for problem children, taking advantage of the surveillance tech used there. They looked at aggressiveness in different situations like “when teased or provoked by peers”, “when warned by adults”, or “when approached positively by peers”. They concluded, quote “aggressive behavior in one type of situation did not strongly predict the individual’s behavior in a different type of situation. […] The same individual who is one of the least aggressive when teased may be well known for his characteristically high level of anger and irritation when flattered and praised.” (That actually resonates with me. While I’m not a problem child, not exactly, and I don’t get *angry*, many people have noted how awkwardly I receive praise, how eager I am to deflect it.)

These results are troubling because it sure *seems* like we have more consistent personalities than the studies show. One reason might be that we let our conclusions determine our observations. In the Newcomb study, there was both the daily recording of events and later summary ratings by the counselors. (I think at the end of the summer.) Both used the same list of activities, including the three I’ve already mentioned plus ones like “Did he tell of his own past or of exploits he had accomplished?” and “Did he spend quiet hour without loud talking or running about?” For each such behavior, they assigned one of five summaries:

1. never occurred at all
2. rare; very exceptional
3. occasionally; not frequent
4. fairly frequent
5. very frequent; prominent characteristic

The counselors knew of the hypothesis that the boys had an underlying level of extraversion or introversion, and that the level predicted behaviors.
Newcomb wrote of the summaries, “These ratings tended to confirm the relationships demanded by the hypothesis of extroversion-introversion [, whereas the daily records did not]. The conclusion can scarcely be avoided that these relationships are found in the minds of the raters to a larger extent than in the boy’s actual behavior. […] One behavior connotes another, in accordance with these mental habits, and in reporting behavior, those that do not fit into the picture tend to drop out, unless retained by some means of objective recording.” In other words, once I’ve concluded Bobby is an extravert, I’ll remember that he both talks a lot at table and mostly plays in a group, even if one of those isn’t true.

Here’s another study, one that also shows how we may place more weight on favorite summary words like “conscientious” than on the more complex actual behaviors. (/Seeing Like a State/ comes to mind here. See episodes 17 though 19.)

I couldn’t get the original description, either legitimately or via Sci-Hub, so I’m working off a brief description in one of the papers linked at the show notes. The psychologists studied the conscientiousness of college students as they went about their lives. As part of the setup, the college students were asked in what situations conscientiousness was important. If I’m interpreting the summary correctly, two things were true:

1. Over all situations, there was the usual lack of consistency: people varied a lot in how conscientious they were.
2. In the situations the students themselves identified as important, they *were* consistent.

In other words, we may think we’re consistent in our conscientiousness, extraversion, helpfulness, and so on because we only pay attention to those things in cases where we *are* consistent. Moreover, we can predict our own conscientiousness in a new situation when we can determine that the situation is “close enough” to a situation we already know about.

The question is: how good are we at knowing the new situation is close enough?

≤ music ≥

Apparently, we are not very good. Here’s an example from my own experience.

I heard about Extreme Programming sometime around 1999. I’d spent the 10 years from 1981 to 1991 working in two-person offices. Since then, I’d been an independent consultant, working alone except to fly in for site visits.

My experience of a two-person office was that I could concentrate and be productive if neither of us talked. But people would come and talk to my officemate. There was nothing like it for breaking my concentration and destroying my “flow experience.”

When I heard of the idea of pair programming within a team room, I was sure I would hate it. How would I ever get a chance to *think*?

Fortunately, I tried it, and it was just fine.

So, where had I gone wrong?

My guess is that I fixated on the *similarities* between a two-person office and a team room, things like doing focused work and hearing people having a conversation that I wasn’t part of. I didn’t look at the differences: when many people are having a conversation, it tends to fade into white noise. Other people’s conversations are less distracting if you’re having one of your own.

Because of my poor understanding of the situation, I couldn’t make a decent prediction. I had to try it to know.

Here’s another example.

Most people would peg me as a classic introvert. I’m not good at parties. Throughout college, I hung back in group discussions. Heck, I even switched my degree from straight Computer Science to Math and Computer Science partly because I didn’t want to take the required compiler construction course that used team projects. I preferred to work alone. I never did “study groups”, either.

However, as a public speaker, I’m extroverted. It’s not even an “acts extroverted but needs to be alone to recharge” kind of thing. I’m *pumped* after I give a talk. I feed on the response of the audience, and the bigger the audience, the more I feed.

So what’s going on?

One answer would be that public speaking is simply one of those cases where a person – me – acts introverted in one situation and extroverted in another, but I think it’s a bit more complicated, because I responded like an introvert in my early speaking career. I was dull, monotonous, and nervous, whereas now I’m positively flamboyant. I’m not faking extroversion: I *am* extroverted, because I changed the situation to facilitate it.

I made one change deliberately and one half-deliberately. The first was to make the talk-giving situation closer to an airplane-landing situation.

See, there was a point at which I was learning to fly gliders. Given that I’m somewhat afraid of heights, that was an odd thing to do, but it’s not actually a problem during most parts of flight. It’s more of a problem during landing, when you’re focused on the ground that’s coming up at you.

The way I got good enough at landing was to pretend the landscape outside the glider was just a two-dimensional image projected on the canopy. Then, the problem of landing correctly became a matter of moving the airplane’s controls to orient that image with respect to certain visible bits of the airframe. No illusion of depth: so no problem.

I did the same thing, conceptually, with audiences. It’s hard to explain, but the audience became an audiovisual image that I was adjusting. (This is almost the opposite of the standard advice of picking a person and speaking to them.) I’ve gotten looser with time, more focused on the audience as people, but that was a big breakthrough.

The half-deliberate thing I did was to allow myself to be physical. I don’t do well as a speaker behind a podium. If I’m allowed to walk around and gesture, that helps me sink into a zone where I’m somewhat talking to myself and somewhat talking to the abstraction of the audience.

What I think is interesting, and perhaps useful, is that I was able to, half-consciously, *engineer* my environment, how I moved in it, how I pictured it, in a way that made me successful at an activity few people in high school would have bet I’d ever be good at.

However, I have some bad news: the environment may be a pretty finicky machine to engineer.

≤ music ≥

In 1972, psychologists conducted a field experiment in a US suburban shopping plaza. In those days, there were no mobile phones. If you needed to make a call while you were out and about, you’d have to find a hardwired payphone, put in a US 10 cent piece, and then dial a number.

If no one answered, you got your 10 cents back. It would fall with a clatter into the coin-return slot. Nevertheless, people often walked away without retrieving their coin. In response, a lot of people got into the habit of checking for change in the coin-return slot whenever they made a phone call.

The 1972 experiment compared people who made a call, checked the coin return, and found the 10 cent coin planted by the experimenters, to people who checked but found nothing. As each person started walking away from the phone, one of the experimenters (always the same person) would position herself slightly in front and slightly to the side, walking in the same direction as the experimental subject. Then she would drop a bundle of folders to the ground, kneel, and start picking them up. The subject had the choice of swerving around her or stopping to help.

The question was: would the good luck of getting a coin affect how helpful the person was?

The results: 25 people did not find a coin. Only one of them helped. 16 people *did* find a coin. 14 of them helped, and only two did not.

That’s a pretty dramatic difference for a coin worth, today, about about 2/3 of a US dollar or an EU Euro. It fits Randall Monroe’s quip: “Always try to get data that’s good enough that you don’t need to do statistics on it.” On the other hand, some really jaw-dropping results in psychology – like the Stanford Prison Experiment – are now widely considered fraudulent or otherwise seriously flawed.

The literature I’ve read makes me think there have been enough replications. For example.

In 1973, a different experiment studied students at the Princeton Theological Seminary. The students were seminarians, people whose degree customarily prepared them for jobs as ministers in Christian churches. In a first session, the subjects filled out a survey about religiosity. In the second, they were called into an office, where an experimenter gave them a description to read and left. The description said the experimenters would like the subject to record an unscripted 3-5 minute talk on a particular topic. The experimenter came back, answered any questions, said space was tight in this building so the recording session would be in an empty office next building over, and drew a little map of the path to follow.

As the seminarian followed the path, he or she encountered a man slumped in an outside doorway, eyes closed, not moving. As the subject went by, the man coughed twice and groaned. The research question was: how much help would the seminarian provide?

There were two experimental variables:

First, how hurried was the subject. There were three cases:

*High hurry*: The experimenter’s script said, “Oh, you're late. They were expecting you a few minutes ago. We'd better get moving. The assistant should be waiting for you so you'd better hurry. It shouldn't take but just a minute."

*Intermediate hurry*: "The assistant is ready for you, so please go right over."

*Low hurry*: "It'll be a few minutes before they're ready for you, but you might as well head on over. If you have to wait over there, it shouldn't be long."

The second variable was which of two topics the seminarian was asked to speak on. The first was about whether, in these tumultuous times (the early 1970s), traditional clergy jobs are both satisfying and useful, or whether graduates should look for other ways to answer their calling, to fulfill their vocation. The other was the Parable of the Good Samaritan, which you may remember is all about stopping to help someone in distress.

The results!

Low-hurry seminarians offered help 63% of the time.
Medium hurry, 45%
High hurry, only 10% helped.

Whether they were to speak on the topic of helping strangers made no difference. In fact, “on several occasions, a seminary student going to give his talk on the parable of the Good Samaritan literally stepped over the victim as he hurried on his way!”

These results seem more plausible than the coin-return example. They’re not *as* dramatic, and we’ve all had the experience of being oblivious while rushed for time. Still…

A group of people studying for a helping profession, sometimes primed to be helpful with the Parable, not being paid much (around 10 dollars or Euros in today’s money), and going to meet a low-status person (some student with a tape recorder) whom they barely knew, nevertheless had their behavior *that much* affected by whether they were worried about being late.

I don’t know about you, but I don’t like feeling that I’m that fickle.

However…

I’ve gotten far enough beyond my suspicion of team-rooms that now my favorite place to work is in a coffee shop. I find the background conversation somehow helps me focus. BUT. There are people in this world whose voices cut through the background, drill into my head, are impossible to ignore, and drive me batty. It’s not volume, exactly, and it’s not so much that they project their voice – there’s just something undefinable about the tone. (Here’s a later addition to this paragraph: judging from the person who has just started ruining my current work session, it has something to do with the *certainty* with which they say *everything*.) Anyway, it’s another of those small changes that has a huge effect. Like being in a hurry for an inconsequential appointment.

≤ music ≥

Where does this leave us? It’s complicated. To recapitulate the summary from the start of the episode:

1. It’s hard to predict how personality traits will affect behavior in new situations.
2. We don’t have a good grasp of the difference between a “new situation” and “a variant of an old situation.” (By “we”, I mean both ordinary people and, as far as I can tell with my weak research skills, psychologists.)
3. Small differences in the situation (like recent good luck) can make a big difference in how traits like “helpfulness” are expressed.

Actually, rather than saying personality is complicated, I should say it’s “complex”, in the particular sense of that word in the Cynefin (kuh-NEV-in) framework. (Note: my expertise in Cynefin is limited to reading web pages, so be even more skeptical of me than usual.) In the framework, “complex” is described in comparison to “complicated”, so let me start with the latter.

A *complicated* domain is, quote, “the domain of experts”, one where the relationship between cause and effect may be subtle, but it’s known. (Or thought to be known, since experts can and do disagree.) David Snowden and Mary Boone use the example of where to put an oil well. Experts gather geological data, they analyze the implications of that data according to geological models, and then they triumphantly stab at a point on the map and say, “There!”. That kind of process goes by the slogan “sense-analyze-respond”.

A *complex* domain is one where you don’t have a predictive model. You can’t start by gathering the facts, because you don’t know what facts are relevant. Snowden and Boone say, “In this domain, we can understand why things happen only in retrospect.” My experience with pair programming in a team room is an example. *Now* I know why it works for me, but back then I didn’t have a model up to the task of predicting my own reactions. However, Snowden and Boone also say, “Instructive patterns, however, can emerge if the leader conducts experiments that are safe to [have] fail.” And that’s what I did. The slogan is “probe-sense-respond”.

To me, the experiments I’ve described strongly suggest that the problem of predicting behavior from personality is a complex problem. If you’ve always had trouble predicting how people will react, congratulations. It’s maybe not that you’re neuro-atypical, it’s that you’re clear-eyed about human limits.

If you think of yourself as being good at going from descriptions of personality (including your own) to predictions about behavior, I ask you to consider whether you may be fooling yourself. Maybe invest in making it easier to probe – to conduct experiments that are safe to fail.

Now, to get ironically recursive on you, you probably could have predicted where this episode would end up from what you’ve gleaned about my personality from earlier episodes. I can’t lie. It’s just *true* that I want to be the physicist Freeman Dyson when I grow up. He wrote the book /Infinite in All Directions/, of which a reviewer wrote, quote “What recommends [Dyson] is his ability to communicate […] the sheer delight he takes in the universe. He loves diversity. Frequently throughout the book a passage will reveal his pleasure at being alive and seeing and thinking.” And I’m attracted to Heinz von Foerster’s ethical principle “Act always so as to increase the number of choices.” I score *very* high on “openness to experience” in Big 5 tests, despite the apparent contradiction of being notoriously and excessively a creature of habit. (What I say to that, *Dawn*, is that I’m open to any experiences compatible with my daily routine of sitting with a laptop on my lap in the morning, then riding my bike to where I can sit with the same laptop on a table, then to the gym, then sit more at a table, then ride back for dinner at the usual time. The apparent inconsistency is another example of what this whole episode has been about. So there!)

≤ short music ≥

For *me*, well-off retired white guy, *most* experiments are safe to have fail. Plus, I get a lot of gratification from the ones that work out. I’m *glad* I found a new – therefore, *by definition*, for me, interesting – way to work. But lots of people aren’t like me. Lots of people couldn’t care less. If a successful experiment warrants nothing more than a shrug, and a failed experiment *isn’t* safe, why wouldn’t a person keep to situations he or she understands? In that case, as the conscientious college student study suggests, within a fixed and controlled set of situations, things like Myers-Briggs maybe have some predictive power, or at least serve as a shorthand and scientific-sounding way to explain why you don’t want to go to that party. If being able to say you’re an INTJ makes you happier, I say, “Have at it”.

It takes all sorts to make a world. You do you, I’ll do me, and thank you for listening.