Amusing Ourselves to Death?

• Heritability of Leisure-Time Activities & Media Consumption
• External Links
• Appendix
  • Literature Review
    • Loehlin & Nichols1976: A Study of 850 Sets of Twins
    • Waller Et Al 1995, “Occupational and Leisure Time Interests, and Personality”

In one narrative of doom, developments in entertainment & recreation like the mass media or designer drugs continuously threaten humanity by creating ever more effective superstimuli —stimuli which are irresistible, as they exploit hardwired preferences, which humans find impossible to resist en masse, any more than the herring gull chick can resist begging for food from a colorfully-painted needle rather than an accurate model of herring gull heads, or other birds can resist nurturing bigger brighter more colorful eggs rather than their own, or Australian jewel beetles can resist mating with beer bottles.

Obesity is on the rise as the food-industrial complex engineers ever more palatable foods, stuffed with salt and fat and sugar, deliciously addictive. Tech companies harvest gigabytes of personal data, using ever more sophisticated & advanced AI tailoring of ads to individuals, pervasively inserted into their media streams, persuading them to buy or think whatever the advertisers wish (subliminal advertising being so dangerously effective it must be outlawed). Media and pornography and harlequin romance novels offer inhumanly perfect simulacrum of sexually attractive & high status people, selecting the most beautiful people out of hundreds of millions and then further enhancing them with special effects into hyperreality, with infinite amounts of porn for every preference and kink on tap via the Internet—and sexbots eventually to come. Social media like Facebook or Instagram expose us to highly-selective curated ideal lifestyles, inspiring envy for lives that never were, and spreading rage and depression. Cheap pervasive ‘firewater’ shattered aboriginal tribes and groups, already weakened by epidemic and dispossession, who had never been exposed to industrial-scale alcohol, even in England, where in the early Industrial Revolution, alcohol consumption spiked in the USA ¹ and “there were gin pushcarts working their way through the streets of London” as Shirky puts it (“Gin, Television, and Social Surplus”), and alcoholism & the “corner saloon” was seen as one of the most severe of all social ills (prompting Prohibition). Ultraviolence on TV desensitizes men to violence while early TV was a “vast cultural wasteland” sucking up hundreds of billions of man-hours a year watching vapid shows like Gilligan’s Island (even now, if time diaries are to be believed, Americans continue to spend multiple hours a day on average watching TV).² Penny dreadfuls, and then comic books, seduce innocent children with their gruesome lurid imagery blurring the line between fantasy & reality while glorifying criminality & aggression, triggering waves of murderous juvenile delinquency. New highly potent “research chemical” opiates manufactured in Chinese labs ravage the American hinterland. Slot machines & video games offer push-button experiences more compelling than real life, with video gamers taught to stalk & entrap vulnerable women, or collapsing in Internet cafes and streamers dying at their post, ‘click’ or ‘social gaming’ companies like Zynga cunningly exploiting helplessly addictive users by mastery of Pavlovian operant conditioning techniques embodied in Farmville, steadily draining them of all their cash in return for making meaningless numbers go up while the players cease even to enjoy the game (and VR increasingly offering a full replacement for dreary reality)—and need one mention how violent games glamorize gun violence and are are simulators training teenagers for school shootings like Columbine? “Fake news”, manufactured in bulk online & flooding Americans through social media, ensured the election of President Donald Trump & the reign of the Republican Party, threatening to end the American democratic experiment forever. Psychological research into human behavior has been weaponized as “nudges” by greedy marketers, suckering buyers into buying overpriced goods like hotel rooms by shady gimmicks such as claiming to almost be out of rooms. Worse, new online mediums and advanced ‘reinforcement learning’ algorithms manipulate innocent viewers into watching or reading dangerous materials, leading to a steady spiral of ever more extreme content consumption culminating in anti-vaccine or Flat Earth or Republican beliefs, with “ Elsagate ” and “ unboxing ” addling children into nonsensical violent sexual video loops interspersed with brainwashing consumerism into the pre-verbal. With continued progress in science & technology, one day we may invent the final entertainment or drug, a species-wide Infinite Jest, something so entertaining or addictive that humanity wireheads itself into being unable to feed, protect, or reproduce itself—eventually going extinct. They literally amuse themselves to death.

This theory can even be extended to explain the Fermi paradox’s Great Silence: why do we see absolutely no trace of alien civilizations when astronomical data indicates habitable worlds should be quite common and technological/statistical considerations show the universe could be rapidly colonized? Many explanations for the Great Silence show a spectacular failure of imagination by postulating extremely narrow mechanisms, like “Western liberalism” or “nuclear war”, for the Silence, which do not do a good job of explaining why the hyperintelligent eusocial ants of Knara Prime went extinct or the immortal sentient silicon crystals of Delti 10 never spread across the universe or the absence of the self-stabilizing chaotic storm cell civilizations of the Jovian planets. But the “amusing oneself to death” hypothesis is plausibly universal: since precognition is impossible and it’s impossible to know the true fitness of actions & choices in advance, all evolved species must use proxies for fitness; if they advance in science & technology to the point of interstellar civilization being possible, they must also have developed a wide variety of tools and understanding of their own physical substrates, which allow faking fitness signals and hijacking preferences; if that is possible, then at least some individuals will do so out of curiosity or for incentives; hence, all interstellar civilizations are at risk of amusing themselves to death and if the risk is sufficiently great, it could be the Great Filter.

In another narrative, superstimuli may be problems, but they tend to be self-limiting ones: people adapt, culturally and biologically, to them, and the problems gradually go away, if indeed they were ever remotely as big as they were portrayed to be, in an endless Sisyphean cycle of technology panics. Early advertising strikes a modern reader as being laughably transparent, crude, and unconvincing ³; the first Internet banner ads in the 1990s reportedly had clickthrough rates >10%, while now the most sophisticated & carefully targeted ad is doing well if it can get a 0.01% clickthrough rate—assuming the traffic isn’t all bots or accidental misclicks—as audiences experience “advertising wearout” & adapt to ignore ads (Kinnucan et al 1993, Blair1987/2000, Braun & Moe2013); and there are serious questions about to what extent Internet advertising works at all & if the tiny effects can actually be measured (see references in my ad A/B test). Alcohol ravaged American Indian tribes and contributed to the destruction of many of them, but while alcoholism and drug abuse remain endemic problems on Indian reservations, the pitch does not appear so apocalyptic these days; Prohibition to reduce alcoholism, rather than being such a burning issue that the very Constitution must be amended, is invoked mostly as a historical fiasco & criticism of the War on Drugs. TV faces competition from other forms of recreation, and is no longer so dominant, with the intellectual quality of TV massively increasing over time (in part thanks to peculiar tech economics like Wall Street investing tens of billions of dollars into Netflix/Amazon to try to replace Hollywood), leading to what has been widely called a golden age of television. Consumers, just as they did in response to earlier advertising, gradually wise up to gimmicks when they seen them too often, and are increasingly cynical about “nudges”. Fake news, whatever contribution it did make to the 2016 US Presidential election, saw its viewership among Americans then crash 2016–2018 (and a disastrous 2018 mid-term election for the Republicans/Trump). Video and computer games, for all the moral panics over them, have proven to be mostly substitutes for other forms of socializing and media consumption, and the occasional “young man collapses dead while playing MMORPG/FPS/etc game” news report remains but a highly rare event, of no more consequence than “young man collapses dead when tackled in sports game” (and often perhaps for the same reason, congenital heart defects), with the various laws passed like curfews being largely unnecessary overreactions; Zynga, poster child of the new wave of exploitative games forecast to be a boot stomping on the wallet of humanity forever, has since reeled from FB changes & loss of most of its users, its 2018 stock now a quarter of its 2012 _14ya high; social games have taken over the gaming world in the form of “e-sports” competitive gaming leagues & professional players with their own stadiums, streaming, casual mobile gaming, and multiplayer games (bearing a deliberate and striking resemblance to regular sports, and a far cry from playing Super Mario Brothers by oneself on an NES). Media consumption often appears motivated far less by the entertainment or esthetic value of the media, than by the desire to participate in trends, signal affiliation with particular groups, or create common referents with other people—all of which inherently curtails media consumption because there is no point in too-obscure media or consuming too much, as that eliminates or crowds out the true function of the media consumption (one might say ‘entertainment’ gives itself far too much credit for being entertaining). Norms for social media are already evolving rapidly, with widespread awareness of the falsity of self-presentations on social media and counter-memes for reducing use, or at least containing the harm with methods like auto-deleting posts (eg. Snapchat). Porn and media are more accessible than ever, indeed, but nevertheless, access to them appears to have little to do with reproduction—American fertility rates are stable and appear far more affected by economic recessions & real estate prices & college attendance than by access to broadband. Likewise, the demographic transition worldwide appears linked mostly to female education, independent of pornography. This is despite the admittedly enormous time soaked up by media—peaking at 9 hours of TV per day per American household around 2009, and yet, people managed to hold down jobs and live their lives and do science and have children. Inherently, any new problem triggers responses and backlashes as people burn out, learn, and pass on knowledge of how to avoid the problems & use it healthily, the technology is modified, societal mechanisms like laws & regulation kick in, the new thing becomes integrated into existing social norms/rituals, and if nothing else, people adapt to it either by cultural inheritance (families/subgroups/ethnicities with adaptive memes flourish while others fall into decadence) or genetically (alcohol abuse may be an example).

Wild populations constantly increase in fitness (Burt1995, Hendry et al 2018), and larger selective pressures produce larger changes, as demonstrated by many successful artificial selection experiments. Further, there is both genetic & cultural evolution at work.

Heritability of Leisure-Time Activities & Media Consumption

MaTCH

The Polderman et al 2015 twin-study mega- meta-analysis of k = 2,748 studies picks up a few results relevant to the question of media/entertainment/leisure-time, with the most relevant categorization being “Recreation and Leisure”, but their online interface for visualization has no way to get the original studies rather than just the model-fitting & wordclouds AFAIK (considerably limiting MaTCH’s usefulness as a database to consult for meta-analyses about specific traits):

MaTCH (“Meta-Analysis of Twin Correlations and Heritability”) database visualization of twin study results on traits classified under ICF/ICD10 subchapter “Recreation and Leisure”

General Literature

For a more fine-grained description, I consulted my bibliographies and used a Google Scholar search along with followup searches of music heritable / game heritable, searches of reverse citation & “related articles” for a few key articles, and checking relevant-looking citations in the body of articles.

I looked primarily for heritability estimates of things which could be described as media or entertainment consumption patterns, media or esthetic preferences, leisure time activities, and the like. I exclude most studies of religiosity because while highly heritable & relevant, it’s arguably not perceived as optional or recreational by most people; I excluded many studies of musical pitch perception/tone identification because those reflect a basic musical aptitude which apparently is not even causally increased by musical practice ⁴; I also exclude measures of vocational aptitude/interest (like the inventories collected in several large-scale twin registries) as those may reflect economic considerations & local opportunities & wishful dreaming more than actual life activities; for physical activity, I try to include only measures of voluntary/leisure rather than work-related or raw physical movement (eg. accelerometer logs, which would include things like fidgeting); similarly, I don’t try to comprehensively cover food preferences or eating habits (much less the vast literature on various kinds of drug consumption & abuse), even though those are arguably primarily recreational activities. Where multiple models fits are reported, I try to use the heritability from what the authors regard as the best-fitting model, and when a CE or E model is selected rather than ACE/DCE/DE, I report as 0.00, and I sum a ² and d ² if both are available; if only monozygotic & dizygotic twin correlations are reported, I (or another author) use the Falconer formula (2 × (r _MZ − r _DZ)).

The full table is available in the appendices due to length (>500 entries).

Looking over the table, we can see that the predictions are borne out: all sorts of preferences & activities are substantially heritable, and a good guess at a mean heritability would indeed be ~0.50 (a simple unweighted average of all the heritabilities in the table is ~0.30, dragged down by the especially high measurement error /instability of many Loehlin & Nichols1976 items).

Some of the heritabilities are estimated at 0.00 but looking at those, they tend to be small samples where the confidence intervals are wide (often the model-fitting gives up & settles on a simple E model for lack of data) or where the trait is probably not being measured well (indeed, probably most of the heritabilities have substantial measurement error as they are based on self-reports and single-item binary/Likert scale questionnaires), so their true heritabilities are almost certainly much higher; generally, only the exercise/sport-related heritabilities have adequate sample sizes because data is routinely collected on those for investigation into basic demographics like weight/ BMI /health, while traits like bingo playing or TV watching are almost never measured. There is an unfortunate absence of any followups or investigations using modern molecular genetics, so there are no SNP heritabilities or polygenic scores to mention.⁵ Hopefully future studies will provide much more precise and broad estimates of these variables, such as using pedigrees extracted from social media—an example would be Facebook, where their researchers can easily extract large family pedigrees with rich detail on books/movies/etc, giving precise estimates of additive heritability & epistasis & estimating time trends, or even just directly extracting samples of tens of thousands of identical twins, and allow examination of other covariates to explain interests ⁶

Other things I noticed looking through the studies were that dominance genetics are reported unusually often (perhaps related to the influence of personality on preferences/activities, as personality traits appear likely to be under balancing selection /frequency-dependent selection which would reduce additive genetics ⁷), shared-environment effects show definite patterns of fadeout with age while heritability seems to increase with age and possibly peak in young adulthood (a Wilson effect?), and there may be sex-dependent effects where females have higher shared-environment and lower heritabilities than males (greater conformism/social concerns?).

External Links

• Kevin Kelly2010, What Technology Wants: chapter 10, “Lessons of Amish Hackers” (blog version)
• “The Amish understand a life-changing truth about technology the rest of us don’t”
• “A genetic perspective on the association between exercise and mental health in the era of genome-wide association studies”, de Geus2020
• “Analysis of genetic and environmental correlation between leisure activities and cognitive function in aging Chinese twins”, Xu et al 2020

Appendix

Literature Review

To demonstrate the point that there are pervasive genetic influences on all aspects of media consumption or leisure time activities/preferences/attitudes, I compile >580 heritability estimates from the behavioral genetics literature (drawing particularly on Loehlin & Nichols1976’s A Study of 850 Sets of Twins), roughly divided in ~13 categories.

Loehlin & Nichols1976: A Study of 850 Sets of Twins

A discussion of extracting ~376 behavioral items relating to recreation/leisure from Loehlin & Nichols1976: A Study of 850 Sets of Twins, which reports comprehensive summary statistic twin correlations from an early large-scale twin study (canvassed via the National Merit Scholarship Qualifying Test, 1962 _64ya). I transcribe them from the book, pool the weighted correlations by gender, and compute simple heritability estimates by Falconer’s formula for use in the recreation/leisure heritability literature review.

Loehlin & Nichols1976’s Heredity, Environment and Personality: A Study of 850 Sets of Twins (see also the briefer discussion in Heredity and Environment: Major Findings from Twin Studies of Ability, Personality, and Interests, Nichols1976/1979 ¹⁴) is a twin study which attempted to compile a relatively large-scale twin sample by an extensive mail survey of the n = 1,507 11th-grade adolescent pairs of participants in the high school National Merit Scholarship Qualifying Test of 1962 _64ya (total n ~ 600,000) who indicated they were twins (as well as a control sample of non-twins), yielding 514 identical twin & 336 (same-sex) fraternal twin pairs; they were questioned as follows:

…to these [participants] were mailed a battery of personality and interest tests, including the California Psychological Inventory (CPI), the Holland Vocational Preference Inventory (VPI), an experimental Objective Behavior Inventory (OBI), an Adjective Check List (ACL), and a number of other, briefer self-rating scales, attitude measures, and other items. In addition, a parent was asked to fill out a questionnaire describing the early experiences and home environment of the twins. Other brief questionnaires were sent to teachers and friends, asking them to rate the twins on a number of personality traits; because these ratings were available for only part of our basic sample, they have not been analyzed in detail and will not be discussed further in this book. (The parent and twin questionnaires, except for the CPI, are reproduced in Appendix A.)

Many of the questions asked about recreation & leisure, including hobbies, preferences, honors/achievements, and items available in their house (most of which would be useful for hobbies or non-school-work); Loehlin & Nichols1976 has been cited by eg. Plomin et al 1990 for the TV watching time use item demonstrating heritability of TV viewing, but most or all of the other items have not been cited much (if at all). The 2 exceptions appear to be the re-analyses Coon & Carey1989, and Hambrick & Tucker-Drob2015.

Unusually, the book includes twin-pair correlations for all of the reported items, not just full test-scales or subfactors, so it’s possible to extract all relevant-looking items and run Falconer on them to go far beyond just TV watching. I have done so below for ~376 items, skipping Vocational Preference Inventory, obligatory religious questions, some dating questions that seem to reflect other parties’ actions rather than preferences/activities, “Ideal Self” preferences, and most school-related or misbehavior questions. Because the respondent sample sizes are not always balanced by gender, I combine male/female correlations before estimating heritability, by transforming them into Fisher’s Z and then taking an average of the 2 correlations weighted by respondent n; results are rounded to 2 digits and floored at 0.

In interpreting the results, it’s worth remembering that single-item responses have severe measurement error and many of the items exemplify this by being extremely specific or suffering from dichotomizing or range restriction and floor / ceiling /sex differences even though most of them have high response rates and superficially seem like large n: for example, in the entire sample of ~1,700 respondents, only around 2 total have ever engaged in skydiving (unsurprising given the era & parents not generally endorsing that activity); only 1 twin managed to publish a scientific paper; 0 twins report getting a tattoo in the past year; 0 twins report national debate successes or inventing a patentable device; while in the other direction, ~96% report having a TV in the house and ~99.9% report ever using a dictionary; and it is not surprising that only a handful of twins have engaged in stock trading (or read the stock listings regularly), have attended prize fights, or that female twins never report growing beards in the past year. (And others are obsolete: teenagers are no longer “pinned” or go for “casual cokes”, and I have never pushed a stalled car to get it started in my life nor seen anyone do so, although it’s interesting to note that apparently almost as many households had sewing machines as had TVs but only ~25% had tape recorders). Since there is no variance or only a little variance, the heritability estimates will be 0 or will be extremely imprecise and could take on any value 0–1.

Some of these are rare for anyone, others are asked prematurely; but of course, we know that something like dictionary use would be heritable if we measured it better in terms of something like “number of dictionary uses per year”, or that publishing scientific papers or attending burlesque shows would be more easily shown to be heritable if the question was asked of the high schoolers a decade or 2 later, and TV ownership will probably be more heritable now that it is rarer. I could have tried to exclude any items which didn’t have a reasonable number of mean affirmative responses like 50, but that would risk cherrypicking, so I include all entries which I initially selected as relevant while reading the questionnaires and then transcribed the raw numbers for.

Entries are categorized by the survey instrument, and presented in the same order as in the Loehlin & Nichols1976 appendix; n refers to respondents, not how many endorse an item (which occasionally is as low as 0).