Data Versus Democracy

by Kris Shaffer

  or why. It’s why a clean house or a well-kept garden makes us feel relaxed,

  even at home—when everything is “as it should be,” our brain has an easy

  time making sense of the environment, even if it’s a new one. (Remember,

  aeons of evolutionary history mean we’re always looking for danger! A clean

  environment, with everything in its place, makes it easier for us to get a

  handle on what’s out there and makes it safe to let our guard down.)

  A number of scientific studies have demonstrated these principles at work. In

  one particularly famous study, researchers showed their subjects (who had no

  knowledge of the Chinese language) a series of Chinese characters and asked

  them to provide an adjective that they thought the character might stand for.

  Characters that the subjects had already seen earlier in the study tended to

  be attributed adjectives with a more positive connotation than characters the

  subjects were seeing for the first time.10

  While it seems like a simple enough study, it’s actually rather remarkable that

  previous exposure to a character can lead to a discernible and statistically

  significant increase in the positivity associated with that character. And this

  kind of phenomenon happens regardless of whether the subject recognizes

  the object! That’s because we build up this “statistical” awareness unconsciously.

  I’ve always been taken by music cognition studies, and they provide some of

  the most interesting insights into these kinds of phenomena. A number of

  studies have been done to measure how humans internalize the patterns in

  the music they hear. 11 From testing infants’ musical expectations by tracking

  the movement of their head or eyes, to asking people to sing the note they

  think might come next, to having subjects rate how well a note “fits” with a

  10Jennifer L. Monahan, Sheila T. Murphy, and R. B. Zajonc, “Subliminal Mere Exposure:

  Specific, General, and Diffuse Effects,” Psychological Science 11/6 (2000), 462–66.

  11See, for example, Aniruddh D. Patel, Music, Language, and the Brain, Oxford: Oxford

  University Press (2008); Huron, Sweet Anticipation; Patrick N. Juslin and John A. Sloboda,

  Music and Emotion: Theory and Research, Oxford: Oxford University Press (2001); and the

  journal, Music Perception—just to name a few.

  26

  Chapter 2 | Cog in the System

  melody, to giving subjects money and asking them to bet on which note will

  come next, these studies come at the question from just about every angle.

  But they generally come to the same conclusion: humans are really good at

  learning patterns through repeated exposure and making predictions and

  judgments in accordance with the patterns we learned, even if we can’t explain

  the reason behind those judgments.

  One study12 presented Western musicians with a melody one note at a time,

  but the melody was from a style that was unfamiliar to all of them—Balinese

  gamelan music. (If you’ve never heard, or seen, Balinese gamelan music

  performed before, find a few videos on YouTube. It’s incredible.) This music

  not only follows different rules than Western classical music, pop music, jazz,

  blues, country, bluegrass, you name it. It is based on an entirely different

  system of scales than Western music. You cannot play Beethoven on a

  gamelan, nor could you perform a gamelan piece on a piano. It simply wouldn’t

  work. That, plus the musicians’ lack of exposure to the style, made it an

  excellent means to test how a human, in this case an expert Western musician,

  might approach “learning” a completely new musical style from scratch.

  The experimenters played first one note of the gamelan melody and asked

  each subject to predict what note they thought most likely to come next. The

  guesses were just that: guesses. No better than chance at predicting the next

  note. After two notes, three notes, still fairly wild guesses. But as the melody

  progressed, and the subjects had more context for their predictions, their

  predictions improved. To be clear, they were not learning the rules of the

  musical style. In fact, they missed some core elements of the musical style

  completely. But they were learning the statistics of the style, the basic

  tendencies and proportions. After hearing enough of the melody, if the melody

  so far had more C’s than E-flats, they would predict C more often than E-flat.

  And by the end of the melody, though they weren’t exactly gamelan experts,

  their predictive powers were noticeably better than chance.

  This ability to learn patterns quickly, with detail, and unconsciously is a core

  part of our humanity. It’s a large part of how we learned our first language. We

  weren’t born with an English gene, or a Japanese gene, or a Farsi gene. We

  didn’t sit down at the blackboard for lessons on how to move the lips, jaw, and

  tongue to say “mama” or “dada.” We are born with a genetic predisposition

  to learn, and our brain hones in on the patterns we hear most often and does

  all it can to emulate them—and to improve upon our ability to emulate them.

  The ability to make predictions and evaluations based on those patterns is

  also part and parcel to being human—even leading to our deepest flaws,

  including responding negatively, even violently, to the unfamiliar. Our racism,

  sexism, homophobia, transphobia, xenophobia, … they all stem from a

  12Huron, Sweet Anticipation, pp. 53–55.

  Data versus Democracy

  27

  negative reaction to the unfamiliar, and our ancient association of the unfamiliar

  with danger and threat.

  Now, of course there are other factors in how we appraise things. As discussed

  previously, the associations our brains have formed over time, as we have

  repeatedly engaged things in context (rather than in an experiment), also play

  a huge role. But all things being equal, familiarity, even unconscious familiarity,

  leads to a more positive appraisal of an object, a person, or an idea.

  This core psychological principle forms the basis of much practice in advertising.

  Think of product placement. The subtle inclusion of a product in a film or

  television show, often paid for just like a commercial, by making that product

  slightly more familiar, slightly more preactivated, primes you to appraise it

  more positively when deciding whether or not to purchase it in the future.

  They might hit you over the head with it—like Reese’s Pieces in the movie

  E.T.—or it might be more subtle—like the label on the can of soda sitting on

  the edge of the table just inside the frame—but the principle remains the

  same: repeated engagement leads to positive appraisal, all other factors being

  equal.

  One wildly successful example from the world of advertising is the white

  Apple earbud.13 Part of the marketing around the iPod itself was its small size,

  even its hideability (“1,000 songs in your pocket”). The iPod couldn’t fulfill its

  own identity as a stealthy device while also being its own marketing campaign—

  “Look at that! I must have one!” However, at that point in time, white earbuds

  were extremely rare. By bequeathing a pair of standout white earbuds with

  every ne
w iPod (and later, iPhone), the device could keep its low profile,

  tucked away in every listener’s pocket, while the earbuds could play the

  marketing game.

  Apple capitalized on this with its visual advertising campaign. Remember the

  commercials, posters, and magazine ads with a black silhouette of a person

  dancing in front of a background made up of a single, solid, bright color? 14 The

  person dancing was always holding an iPod in their hand (also part of the black

  silhouette). The only other element in the photo or video was the pair of

  Apple earbuds. White. Stark against the black and fuchsia/orange/green. The

  visual focal point of the advertisement. The earbuds highlighted in this ad

  campaign and the earbuds seen on the street or in the train reinforced each

  other, generating a sense of ubiquity and raising the familiarity of the symbol.

  Never before had the color of the wire between someone’s ear and their

  13Sasha Geffen, “The iPod May Be Dead, but Those Iconic Ads Still Shape the Way We See

  Music,” MTV News, published Mary 12, 2016, www.mtv.com/news/2879585/ipod-ads-

  in-music-culture/.

  14“2003: Apple Releases its Silhouette Campaign for iPod,” The Drum, March 31, 2016,

  www.thedrum.com/news/2016/03/31/2003-apple-releases-its-silhouette-

  campaign-ipod.

  28

  Chapter 2 | Cog in the System

  pocket primed us to think (usually positively) about a specific brand. But by

  picking something rare yet visible and reinforcing it with their visual advertising

  materials in a way that took advantage of the way our brain works, Apple was

  able to do just that.

  Now, familiarity itself is no guarantee of a positive response. Those same

  white earbuds, once they reach a point of market saturation, can start to

  stand out too much. Some people recognize the gimmick. Others simply want

  to avoid being like everyone else. And so how much is too much? is a question

  advertisers have been asking themselves for decades. A single Superbowl ad

  isn’t worth the expensive investment if you have no marketing budget left

  afterward. But the same commercial seen during every commercial break of

  every show is also counterproductive. Remember that when something is too

  familiar, but arrests our attention anyway, the waste of resources can lead to

  a negative reaction. Finding the sweet spot (which is different for every

  individual) is a major part of the advertising game.15

  Also remember that when two things are associated in our memory, activating

  one will activate the other at the same time.16 So while repeated exposure to

  white earbuds might boost familiarity with both the earbuds and the device

  they come with, that only leads to a positive appraisal if you don’t already have

  a negative association with the earbuds, the iPod, or the Apple corporation.

  This is why after major scandals, companies often rebrand, either renaming a

  product or renaming the company altogether.

  I imagine you can already see potential implications here for social media and

  big data analytics. If seeing something too often is bad, not often enough is

  ineffective, and the sweet spot is different for each person, then individualizing

  how often someone sees an ad would optimize its effect. Likewise, if

  associations in a person’s memory affect the positive or negative appraisal

  they have when seeing or hearing something, and if different people have

  different associations, then individualizing the content of an ad would likewise

  optimize its effect.

  Of course, individuals aren’t entirely unique. Though you and I might react

  differently to the same ad, we might react similarly enough to someone else

  that showing us the same content, on the same schedule, would have more

  or less the same effect. And so advertisers have been grouping us together for

  ages—single men ages 18–25, tweenage girls from upper-middle-class families,

  retired working-class widowers on medicare, city dwellers, suburbanites,

  homesteaders, you name it. Marketers divide people into groups based on

  demographics, do their market research, and target ads to groups accordingly.

  15Erin Richards, “Cognitive Efficiency Determines How Advertising Affects Your Brain,” in

  Science 2.0, December 9, 2008, www.science20.com/erin039s_spin/cognitive_

  efficiency_determines_how_advertising_affects_your_brain.

  16Donald O. Hebb, The Organization of Behavior, New York: Wiley & Sons (1949).

  Data versus Democracy

  29

  This is why there are a lot more toy commercials during children’s television

  shows than during daytime soap operas, for example.

  But TV ads can’t really target as precisely as advertisers might like. Are the

  people most likely to buy product X or vote for candidate Y really all watching

  the same show at the same time in the same broadcast/cable/satellite market?

  Surely our identities are not defined and clustered according to our television

  viewing habits, our morning-commute radio preferences, and our magazine

  subscriptions!

  The internet drastically altered this equation. Web technology allows

  advertisers (and purveyors of all kinds of information) to microtarget very

  small groups of people—even individuals, if they really know what they’re

  doing. (Targeting individuals with ads is forbidden on most major online

  platforms, but there are ways around that for the more talented, if less ethical,

  information operatives.) If the optimal content and timing is individual, if the

  groups of people with similar preferences and associations do not line up with

  TV show viewership or commute schedule, then targeting online starts to

  make more sense. And in a media environment where every web site visited,

  every link clicked, every image or video viewed, every song listened to, every

  like, every retweet, every :rage: emoticon is logged and available to advertisers

  (at least as filters), well … then we can really start to put our knowledge of

  human cognition to work in our marketing practices.

  In a world where every “sponsored post” is effectively an advertisement,

  advertising is not simply about selling things. It’s about political campaigns. It’s

  about journalists and freelance writers promoting their work. It’s about indie

  musicians finding an audience. It’s about inventors crowdfunding their

  prototypes. Anyone who wants to communicate information or effect human

  behavior, and who has the money to support it, can target individuals in finely

  grained ways without ever scraping a single bit of data themselves. And in

  spite of all the limitations put in place by the platforms, unethical operators

  can do some really shady things. (More on that in the chapters to come.)

  Summary

  Let’s sum up what we’ve learned. The brain is a large and complex information

  system, with optimized storage solutions and a powerful predictive analytics

  engine. It’s the real intelligence that the best AI systems are still trying to

  catch up to. The result of the brain’s optimized information storage and

  predictive analytics algorithms is a bottleneck called attention—the small

  amount of information we can “think about” at any given tim
e, including

  memories and signals from our five senses. Biologically, attention is incredibly

  powerful, but also incredibly expensive, and therefore incredibly limited. Over

  the course of human evolution, we have evolved processes for prioritizing the

  information that makes it into conscious attention, including both hard-wired

  30

  Chapter 2 | Cog in the System

  rules by which certain kinds of stimuli will always take priority as well as the

  capability to learn new priorities and connections that govern our attention.

  One of the main things we learn is statistical frequency (familiarity). Memories

  we engage frequently stay “preactivated,” ready for easy access. We also learn

  relationships between memories and stimuli, and when we interact with one

  thing in our memory, all of the things our brain has related it to will also raise

  in their level of activation, making it easier for us to process what we predict.

  We also know that, all else being equal, when a perception is easy to process,

  we form a positive emotional response. And that unconscious familiarity can

  be a powerful driver of positive responses to things we perceive.

  Of course, the degree to which something is familiar, whether or not we are

  aware of that familiarity, and what associations (positive and negative) we

  might have already made with that thing is highly individual. And while

  advertisers have been dividing us into groups for decades, and serving up ads

  likely to be the most effective for the particular group(s) we belong to, the

  optimal solution will be individualized.

  But there are a few problems with completely individualized advertising. First,

  in order to work, it requires access to tons of data about us as individuals—

  much of which we likely view as private. This includes knowledge of what

  we’ve seen, heard, smelled, tasted, touched, in what contexts we’ve interacted

  with it, and how we reacted to it in those different contexts. Where relevant

  data is absent, it requires a means of filling in the gaps (more on that in

  Chapter 3). Assuming advertisers have the data needed to manipulate us in

  the way they desire, this completely individualized advertising can have effects

  on both our behavior and our psychology that were not intended, or even

  thought of, by the advertisers. And the effects—both intended and