Is Selfless Evil Far?

Brainstorming came from Osborn in 1939 as a method for creative problem solving. He was frustrated by employees’ inability to develop creative ideas individually for ad campaigns. … Osborn claimed that two principles contribute … “1. Defer judgment,” and “2. Reach for quantity.” (more)

In the last decade or so, psychologists have confirmed one of the most robust mind patterns ever seen: construal level theory, which I call near vs. far thought. In brief: humans think more abstractly, and in less detail, about things far away in time, space, social contact, and probability, and assume that things near or far in some ways are also near or far in other ways.

Since far mode thoughts tend to have weaker decision consequences, I’ve suggested that far mode is better adapted to managing social images, relative to making helpful choices. This fits with far mode being more associated with confidence, high power/status, positive moods and reasons, pride and shame, self-control, trusting others, resisting conformity pressure, supporting underdogs, love over sex, words over sounds, polite speech over slang, and ideal values over practical constraints.

But even if its greater role in managing social images makes far mode beliefs less accurate, far mode is built too deeply in us to do without it. If we must use it, how can we best use it, to avoid bias? My tentative answer starts from the observation that in far mode we are better at creativity, while near mode we are better at analysis.

Mental tasks can be roughly divided into generation and evaluation. Our minds must search a vast space of possible thoughts, generating possible thoughts to explicitly consider. We must also evaluate such explicit thoughts. Since far mode is better at creativity, while near mode is better at analysis, we should prefer to generate in far mode, and evaluate in near mode. First see if idealism can be made practical, before resorting to cynicism. This fits with claims that groups create better when they temporarily avoid criticism and evaluation.

Of course we can’t make this a strict rule; circumstances will often force us to evaluate in far mode, and to generate in near mode. But we should at least be aware of our handicaps in such situations. Which brings us to the subject of evil.

Humans evolved a sense of morality, helping us to coordinate to discourage many specific forms of selfish behavior that hurt groups. We thus evolved to tell stories of evil villains who engaged in such harmful behaviors, and of good heroes who opposed them. Such stories often depicted villains who are tempted in near mode by concrete personal gain, such as loot or sex, and heroes who thought in far mode about a wider good.

But today, most evil is probably not of this selfish sort. Instead, very bad things are caused more by far thinking. Consider the prototypically-evil Nazis. Their urge to exterminate Jews came less from unhappy personal experiences with individual Jews, and more from abstract fears gone wrong – killing Jews probably hurt Germans overall. Similarly, most xenophobia comes less from personal interactions and more from abstract, and largely incorrect, fears. People tend to have satisfactory and mutually advantageous relations with immigrants, even as they politically support policies to prevent such relations.

Similarly, democratic regulation usually goes wrong by supplanting direct consumer evaluations of products, services, and practices, which tend to be made in near mode, with abstract public opinions about good policy, which tend to be formed in far mode. Autocratic regulation goes wrong similarly, since power tends to put leaders in a far mode. I’m not saying that there should never be regulation, but rather that an important and neglected cost of regulation is displacing reliable near mode evaluations with unreliable far mode evaluations.

We can’t think without far mode, but we can use it most where it works best: to suggest candidate actions, products, policies, theories, etc. We should minimize biases from a far mode system designed more for social image management, by using near mode where it works best, to analyze and evaluate these candidates. Science experiments, computer engineering demos, policy trials, prediction bets, and business profits all offer such crucial concrete near-mode feedback. We need these to avoid the all-too-common selfless evil of far mode evaluation errors.

Added 8Sept: There is some tension between this post and my older post on The Felt & The Unfelt.

GD Star Rating
Tagged as: , ,
Trackback URL:
  • Good synthesizing post bringing your experiments with near/far into you improvement of social epistemology project.

  • Matt Young

    Th two modes should be related, far term thought gives us a goal, short term thought arranges events to match the goal.

  • mjgeddes

    Bayesian Induction/Decision Theory = Evaluation/Optimization = Near Mode
    Categorization/Information Theory = Creativity/Signaling = Far Mode

    Also Far is intimately connected with conscious experience, whereas Near does not neccesserily require conscious deliberation.

    Near mode is Slytherin/Ravenclaw-like (mechanics of winning), far mode is more Hufflepuff/Gryffindor-like (social image, heroic ideals).

    The key question to which no one can give me a straight answer is what is the exact relationship between the two? Is a general intelligence that runs on near mode alone possible, or is far mode actually a neccessary feature of general intelligence?

  • Doug S.

    The Well-Intentioned Extremist is indeed a popular villain type these days…

    • Evan

      I’ve noticed this too. If you read older works, most villains tend to be selfish and evil in both near and far mode. But in modern works villains who believe in some evil cause in far mode, but are disciplined and selfless in near mode, seem to be very popular. I wonder what motivated this change?

      • Alice

        Good point, Evan!

        I think there is a trend towards more nuanced characters in that way – the tragedy of moral ambiguity. Also, it’s more terrifying because more realistic to have a villain who is nice to children and helps old ladies across the road but just happens to be building an anti-asian army.

        The boorish villain particularly of victorian earlier works didn’t even need to be attributed far-mode thinking, he was repulsive enough in his proximity that it was rarely necessary to paint in the distance.

        I’d say the corollary of your villain-who-likes-flower-arranging is the hero with the drinking problem and the estranged wife who can’t help fighting for justice.

  • E. H.

    I had a writing class in my freshman year of Fine Arts college that was focused on the Holocaust. The final project was “What can Art do in face of the Holocaust?” I wrote about how the frame of mind and paradigm of art itself was at the root of what made the Holocaust possible… I decided it was too contraversial an idea, so I presented on how I was afraid to present my idea… Had I had this post as a reference, I might have had the courage-through-validation to present my original thought.

  • Noumenon

    I was just thinking when I read your last post “How is he referring to a ‘standard finding’ of near-far mode? I thought that was just his own organizing framework.” So it’s very helpful in this post to be reminded that “near-far mode” means “construal level theory,” which is what I’d have to call it to make it sound official to friends (even though your near-far framework is what makes me remember it.)

  • Thanks for the post– that lays the ideas out very neatly.

    I think a typical move in politics is to claim that the other side is moved by bad abstract theory.

    I never quite believed that the Pure Blood project made sense for Slytherin– too abstract, too idealistic. Slughorn is true Slytherin.

    • mjgeddes

      Yes Slytherin is certainly not selfless. Slytherin simply symbolizes our baser near-mode desires, which people like to pretend don’t exist. Ravenclaw and Hufflepuff symbolize the ‘passive voices’ of the mind, Sytherin and Gryffindor are the ‘active voices’. But evil is indeed far more likely to come from Gryffindor (heroic narratives, far) than Slytherin (practical hacks, near).

  • Keith

    As read your posts on construal theory, there’s something about the nature of the dichotomy between near and far that seems to be changing between contexts.

    I’m not sure if it’s that the position intermediate between near and far should be looked at more closely, or that there’s another axis nearby that bears examination. I feel like I’m missing something important here.

    I can’t quite put my finger on it though, not enough to come up with a test. I’m not quite sure what it is, but it’s tickling my brain every time I read one of these posts.

  • mjgeddes

    Interesting that the classic game ‘Pacman’ AI was programmed with ‘near’ and ‘far’ modes’ for enemy behaviour.

    I also used ‘near’ and ‘far’ modes in my trading spreadsheet for horse racing prediction markets on Betfair. The price movements have a ‘far mode’ behaviour further out from the race (lower volatility, looking for general horse race categories), and there’s a sudden phase-change to a ‘near mode’ behaviour as the event gets close in time (higher volatility, need more precise info on specific horses). Could be a general feature of prediction markets? If so, I can well image any working AGI needing the near-far mode split.

  • Pingback: Overcoming Bias : Forget 9/11()

  • Pingback: The Spamlist!: » Overcoming Bias: Forget 9/11()

  • Pingback: 9/11: Should America move on? « The Melbourne Urbanist()

  • Pingback: » Mammoth Reads: Attraction, Death, Medicine, and Punctuation » They Will Rise Again From the Tundra()

  • Added 8Sept: There is some tension between this post and my older post on The Felt & The Unfelt.

    There is! And which one is more right is closely related to the question of whether signaling is conscious (and accomplished in near mode, as the earlier article entails) or subconscious (and accomplished in far mode, as follows from the present posting).