How Wrong Can We Be?

Most models published in economic journals have a reasonable if loose match with human behavior.  If most published economic models says that in X-type situations people tend to do more Y, then more often than not, people in fact more often do Y when X.  This is evidence that economic theoryists actually know something about human behavior.

However, when you ask people in situation X why they do Y, the reasons they give usually have only a weak connection to the reasons in related economic models.  Yes people who have been taught economics can find it easy to explain their actions in economic model terms, but this is not how most folks usually think.  Thus the practice of academic economics implicitly accepts that people often, perhaps even usually, do things for reasons other than the reasons they give.

Consider also that something similar holds in sales and marketing.  The rationale a marketer gives for why an ad or other product strategy works usually differs quite a bit from the reasons people give for why they like an ad or a product.  Similarly, the reasons dating and other relation consultants give for why their suggested strategies help people like or respect you are often quite at odds with the reasons people give for why they like or respect others.

In addition, I just posted on how seeing the hidden status games in most conversations makes one a better actor, and on how psychotherapy is all about exposing our hidden-to-ourselves agendas.   Standard social science accounts of religion say religion is quite functional for people, but for reasons rather different from the reasons religious people give.  Similarly, my “showing that you care” explanation of medicine suggests medical behavior is functional, but for quite different reasons than people usually give.

This all seems to add up to a consistent expert consensus that humans quite often, perhaps even usually, just don’t know why they do what they do.  And this is extremely disturbing, as it calls into question our own opinions about why we do what we do.  Worse, if each of these areas (econ modeling, marketing, acting, psychotherapy) experts call into question only a limited range of our opinions, while implicitly assuming most of our other opinions are correct, perhaps these experts seriously underestimate just how misinformed we all are.

I’d like to take this skepticism seriously, and in fact seem to be somewhat obsessed by this project.   I’m not sure exactly how, but I’ll plow ahead anyway.

GD Star Rating
loading...
Tagged as: ,
Trackback URL:
  • http://andrewkemendo.blogspot.com Andrew Kemendo

    I’d like to take this skepticism seriously, and in fact seem to be somewhat obsessed by this project.

    Is this not already the core of every field you mentioned; modeling human behavior?

    As an economist I know I realized early on that this is what economics does, or at least seeks to do – and I think the same is true for psychology and all other human-centric sciences.

    Worse, if each of these areas (econ modeling, marketing, acting, psychotherapy) experts call into question only a limited range of our opinions, while implicitly assuming most of our other opinions are correct, perhaps these experts seriously underestimate just how misinformed we all are.

    I think you project too much onto researchers about their thoughts on how people think of other things which lie outside of their field. I can only speak for myself when I say that I take other people’s stated preferences (on all things) with a grain or two.

  • http://sophia.smith.edu/~jdmiller/resume.pdf Jam

    Is this result just because experts have spent more time considering the issue?

    What if you found someone in situation X who was doing Y and said “spend an hour thinking about why experts think people like you do Y. If you give the right reason you will get $Z”

  • TGGP

    The most extreme examples of this are the rationalizations people give for their behavior under hypnotic suggestion or with severed corpus callosum (aka split-brain).

  • http://ambrosini.us/wordpress pushmedia1

    Economic models give rationalizations for average behavior. Its hard for us to average rationalizations (how do you take the average of words/stories?). Maybe you’re just realizing a sort of Jensen’s law of verbalizations where the verbalization of an average isn’t equal to the average of verbalizations.

    • Jess Riedel

      The expert explanation for the behavior is not just rule of thumb which is useful for compressing observed data; is has predictive value. The claim, I think, is that we can experimentally demonstrate that the expert explanation (rationalizations for average behavior) is more correct than the laymen explanation (average rationalizations). That is, in fact, which the experts hold these opinions.

    • Jess Riedel

      Also, I haven’t heard of Jensen’s law and a google search turned up nothing. Could you point me toward a description?

      • http://www.cawtech.freeserve.co.uk Alan Crowe

        Pushmedia1 is alluding to Jensen’s Inequality. What can we say about the expected value of a function of a random variable, compared to the values of the function evaluated at the expected value of the random variable? In general we can say very little. If the function is convex then the expected value of the function is larger; this is Jensen’s Inequality. Pushmedia1 is riffing off the set-up for Jensen’s Inequality, in which we contrast applying the function before and after taking the expectation.

  • Roger Depledge

    people often, perhaps even usually, do things for reasons other than the reasons they give.

    quite at odds with the reasons people give for why they like or respect others.

    functional, but for quite different reasons than people usually give.

    This all seems to add up to a consistent expert consensus that humans quite often, perhaps even usually, just don’t know why they do what they do.

    This seems to me to be a non-sequitur. People may very well know but give distorted or false reasons.

  • Curt Adams

    This is highly analogous to biology. Biological organisms do the “right thing” for their reproduction (frequently, anyway) not because they understand what would accomplish this goal, but because genes that produce the right effects by arbitrary biochemical processes have become dominant in the population through evolution. It doesn’t matter *why* a salesman pursues a particular strategy – if successful salesmen keep their strategies while unsuccessful ones quit or change, then salesmen will tend to do things that help them sell whether or not they understand why. Indeed, when the “right things” involve deception, coupling incorrect beliefs with correct action could be actively beneficial by freeing the actor from worrying about getting caught.

    • Robert Koslover

      I agree. Robin seems concerned that people have unreliable self-explanations for their behaviors. Well, just try asking any other animal why it behaves the way it does. The problem is not just that it can’t tell you the answer, but rather that (except perhaps for some chimps or dolphins) it literally doesn’t even have a clue! People are obviously capable of far more self-reflection than animals, but it would still seem reasonable to believe that we only have some, not all, the information we need to truly accurately explain our own behaviors.

    • http://hanson.gmu.edu Robin Hanson

      Having no explanation or being very uncertain about an explanation is very different from humans, who act rather confident in their mistaken explanations.

      • Robert Koslover

        Yes, I agree with that too. I would also speculate that it is a defensive mechanism, even for innocuous behaviors. The split brain experiments mentioned by TGGP for example,suggest that the conscious human mind is desperate to always have some kind of excuse or explanation for both conscious and unconscious behavior. After all, to acknowledge that one’s outward behavior may actually be due to in unconscious factors means accepting that we are not fully in control of ourselves. A second comment: When a child misbehaves, and and adult demands to know why, the child may sometimes look sheepish and answer “I don’t know,” even if asked repeatedly. Interestingly, that answer is very seldom considered acceptable by the parent (even when it may be true, in some cases). And so, for better or worse, I would assert that children (at least, those in our culture) are effectively trained from an early age that: (1) explanations for their behavior are mandatory, and (2) that it is a useful skill to be able to come up with them quickly.

  • http://brokensymmetry.typepad.com Michael F. Martin

    Isn’t this why economists tend to take only revealed preferences seriously?

    • Douglas Knight

      If you apply the cynicism of economists to their profession, you might instead conclude that their position is to minimize verbalized preferences as a bar to publication. At the same time, a publication gains novelty from contradicting the verbalized preference.

    • http://williambswift.blogspot.com/ billswift

      I just posted a comment on Less Wrong (http://lesswrong.com/lw/e9/fighting_akrasia_finding_the_source/#comments) that suggested “akrasia” is really just a mismatch between our real (revealed) preferences and the preferences we want to signal.

      • fburnaby

        That was a fantastic article, by the way.

  • Douglas Knight

    Most models published in economic journals have a reasonable if loose match with human behavior.

    Of course: they wouldn’t be published if they predicted obviously false behavior. Journals reward novelty and behavior is cheaper to observe than motivation, so they’re more likely to reward novelty in motivation than behavior.

    Marketing has better incentives to understand people, though not necessarily to share this knowledge. Also, the relevant behavior (purchasing) is easy to observe.

  • Norman

    “[…] perhaps these experts seriously underestimate just how misinformed we all are.”

    Do experts really think people are just mistaken? While it is possible that people don’t engage in introspection, that doesn’t seem to be the simplest explanation. Perhaps it is as Dr. House from Fox’s House, MD puts it: “Everybody lies.”

    I think assuming that people will tell the truth or give the best answer they know whenever asked, regardless of giving any incentive for truthfulness, assumes a false default position. Most of the time we will make ourselves look better by lying about our motives for any given activity.

  • Unnamed

    For the social psychology angle, you should see Nisbett and Wilson’s (1977) classic paper, Telling More Than We Can Know, and Tim Wilson’s book Strangers To Ourselves.

  • Granite26

    Is it possible that people realise EXACTLY why they prefer things one way, but lie to the survey takers? Especially when they KNOW that they’ve got off-topic or shallow reasons for revealing a preference.

    Another possibilty is sufficiency. Any preference is made up of a significant number of factors. It’s really boring to say “Well, I thought she was cute because she’s got a balanced face, a nice bust and an hourglass shape’, because ALL cute girls have this. So your mind is picking out not the things that make her stand out in the group you’ve put her in (sic: cute girls). In short, the dimples aren’t what make her cute, they’re what make her memorably cute.

  • Z. M. Davis

    Are the laypersons’ and experts’ accounts are really so opposed to each other? There can be multiple levels of explanation: the individual might be perfectly sincere in their stated reasons for doing something, while the economist or the sociologist might have another complementary explanation for why social conditions exist in which people do this thing, and the evolutionary psychologist has yet another complementary explanation for why humans evolved to want to do this thing under such-and-such conditions, and the physicist can just as correctly say that it’s all just quarks and leptons anyway.

    This is just standard textbook reductionism, isn’t it? What am I missing?

  • Yvain

    Someone on OB (can’t remember who) once posted this paper full of the psychological evidence for exactly the proposition you’re arguing. It’s pretty good. If you didn’t read it the first time around, I have to recommend it.

    However, there are also cases like the psychotherapists you described a few posts ago. They managed to do well and beat a control group, but the theories they used to explain their success were completely made up. The self-reports of patients probably would have been a better guide to what was going on in their minds than the therapists’ Freudian models.

    Compare this to the evolutionary psychologists, who usually try to confirm their theories by predicting weird facts that would not have been obvious without their theory and then confirming them in experiment (like how the grief parents feel at a child’s death is proportional to the child’s reproductive opportunities).

    Some economists have managed to do this; others have not. I am only fully comfortable believing the economists’ theories once they meet with such experimental success. If you want to convince people to believe those theories, you should publicize their successful non-retrospective predictions as much as possible.

  • http://www.hopeanon.typepad.com Hopefully Anonymous

    I think it’s great that you’re obsessed with this project. One way I think you could usefully proceed is in making quantitative models that attempt to forecast behavior. I think a strand of research that does tie together the various combinations of self-deception and social functionality is the area of repugnancy. Self-deception may be optimizing when a functional activity or motivation is repugnant to ourselves. Goffman’s concept of “covering” may occur when a functional activity or motivation is repugnant to others.

    Also, I think it’s really useful to look at how this works on the various social orders of magnitude from psycho to micro to macro. Countries may not have internally transparent propagandas about why they have an officially stated belief or religion, why they elect a president with particular traits, or why they go to war. Even at the most “elite” levels, where decisions are made, there may be internal micro-deception, with nontransparent semiotics for the purpose of coordination, rather overt conspiracy (“we should lie to the masses so they’ll do this, because the truth would be repugnant to them”).

    The previous paragraph applies to smaller social units too, like corporations and neighborhoods. Building up from primate aesthetics to optimized algorithmic behavior may be the key here.

    A useful model may be flexible rational agents optimizing for persistence (corporations, countries, trait distributions), playing a game where to persist they need to motivate primates with relatively inflexible, irrational (from the perspective of persistence optimizing) biases (repugnancies, motivations).

  • http://www.hopeanon.typepad.com Hopefully Anonymous

    “rather overt conspiracy” is a typo, I meant “rather than overt conspiracy”. Coordination with nontransparent semiotics rather than overt conspiracy to pursue an outcome where the underlying motivation may be repugnant for an elite subpopulation to acknowledge to themselves as a subpopulation.

  • Jeet

    ‘In vino veritas’ – some people tell the truth when under inebriation. Can we actually describe our motives ( why we do what we do) under a bit of inebriation as well? Not sure if any research has been done on this (besides decision making stuff). Does the ‘sober self deception’ transform into realisation under inebriation? If the answer under inebriation is emotional – rather than a rationalisation of some factors, what does that mean?

  • http://moritherapy.org isabella mori (@moritherapy)

    interesting topic, with many angles. thanks for being obsessed with it!

    i’d like to offer an addition to what z.m. davis said re multiple levels of explanation. it’s pretty simplistic to run on the assumption that behaviours have ONE reason/cause. there’s usually a complex web of them, and it’s hard to put it all in words.

    also, let’s not forget how ill-worded most surveys are, and how quickly we tend to want to get through them. i don’t know if focus groups are any indication of how economists structure their methodology but the marketing-oriented focus groups i’ve attended, run by way-too-well paid industrial psychologists, were pretty abysmal from a research point of view.

    you may find it interesting that there is a branch of brief therapy that is becoming more and more popular (proposed by people such as steve miller) that thrives on taking people by their word, assuming that what they say is as close to the truth as can reasonably be expected.

    having said all this, i totally agree that there is a level at which humans are not overly familiar with, or interested in, the truth (whatever THAT is.)

  • http://liberalvichy.blogspot.com/ Vichy

    I don’t see what psychology or statistics has to do with economics proper (that is, economizing scarce resources). Economizing within a means-end framework, and their relation within an overall market/distribution scheme is quite different from ‘modeling’ actual preferences, prices and historical situations.

    • http://entitledtoanopinion.wordpress.com TGGP

      If you accept E. O. Wilson on reduction and consilience (as I do), then economics can ultimately be reduced to (and hence should be based on) psychology, then psychology on neurology and so on down to physics. Statistics is a tool for understanding a variety of things, including economics.

    • http://twitter.com/rotkapchen Rotkapche

      Vichy: The fundamentals of economics is ‘choice’ (scarcity is one element that plays into market pricing, which is a different part of economics). What we choose and how we choose it has a huge overlap in psychology.

      Take a look at the emerging field of Behavioral Economics

  • rbhui

    Robin, have you read “Rational Choice in an Uncertain World” by Reid Hastie and Robyn Dawes? While including behavioural economics, it has nice coverage of expert errors. The book is very readable, and the authors are very knowledgeable, an excellent combination.

    • http://lesswrong.com/ Eliezer Yudkowsky

      It’s the introduction I usually recommend.

  • http://pdf23ds.net pdf23ds

    This post is really inspiring me to be even more misanthropic. Of course, it’s really nothing new on this blog, but it’s nice to know you’re thinking about how to put all the pieces together.

  • karthik

    I dont know if this comment is appropriate. When i read this:

    This all seems to add up to a consistent expert consensus that humans quite often, perhaps even usually, just don’t know why they do what they do. And this is extremely disturbing, as it calls into question our own opinions about why we do what we do.

    I remembered Nassim Talebs book “Fooled by randomness”. There he talks about the role of human irrationality. He says all human decisions need not have reasons and could just be irrational and nothing much. May be when analyzing such questions we are too rational to not admit irrational behaviour that we are shocked at the “singularity” in human behaviour which is not knowing why they do what they do.
    Well another reason why experts may give different reasons(hopefully people take it in a right spirit) is that he has to find a justification,a reason, while there may be none. The following seems to fit perfectly:
    The authors of journals look for reasons and justify their point. People are also forthcoming with reasons;only that they are different. And there is no reason behind their decisions.
    It opens up another question. Suppose there is a case where reasons given by all coincides with the expert. It still does not mean it is the right reason . Probably there was no reason start with. Expert tries to find one and fit it. Even other people also try to explain their actions. But strangely the matter at hand is unique enough that there is less variance in reasons given by people and the mean happens to be that of the expert.
    Any comments?