Stone-Age Minds in Traffic

Evolutionary mismatch explains myopia, obesity and doomscrolling with one sentence, which is why it deserves less trust than it gets. One claim here is close to airtight, one is a live dispute, and one borrows a mechanism's vocabulary without its evidence.

Every claim this article audits eventually has to survive a room like this one: an instrument checked against a standard, a dose logged in real hours, a cart that weighs what a body actually burns. — Image prompt and art direction by Brecht Corbeel; generation pending.

Abstract

"Our Pleistocene brains can't handle modernity" is a sentence flexible enough to explain nearsightedness, weight gain and phone use with equal confidence, and a claim that flexible should worry a careful reader before it persuades one. This article proposes a three-part audit for any mismatch claim — a specified evolved mechanism, a dated environmental novelty, and causal evidence linking the two — and runs it against three cases. Myopia's indoor-childhood epidemiology clears every bar: randomized trials in China and Taiwan, a named retinal mechanism, and a prevalence curve that tracks exposure rather than ancestry. Obesity keeps a real mechanism but loses the specific evolutionary story, caught between Neel's thrifty genotype and Speakman's drifty-gene rebuttal, while the modern food environment's causal effect on intake stands on its own regardless of which theory wins. Screens and attention borrow supernormal-stimulus rhetoric that the causal evidence, so far, does not support at the scale claimed. The piece closes with the criteria that separate an audited mismatch claim from a headline.

A mismatch claim is a factual claim, not a mood

“Our brains evolved for the savanna, not the strip mall” is one of the most quoted sentences in popular science writing, and one of the least policed. It is invoked to explain why people overeat, why they doomscroll, why they panic in open-plan offices, why they fall for scams, why teenagers are anxious, and why commuters find traffic unbearable. A sentence that flexible is either a genuinely deep unifying insight or a rhetorical move so loose it can be fitted to any complaint after the fact. The only way to tell the difference is to stop treating “evolutionary mismatch” as a single claim and start treating it as a family of separate, falsifiable claims that happen to share a name.

The name has a real intellectual lineage, and it is worth taking seriously before auditing it. John Bowlby introduced the underlying concept in 1969 while building attachment theory, arguing that any evolved behavioral system can only be expected to function well inside the conditions under which it was shaped: “structure takes a form that is determined by the kind of environment in which the system has been in fact operating during its evolution… This environment I propose to term the system’s ‘environment of adaptedness.’ Only within its environment of adaptedness can it be expected that a system will work efficiently” [1]. Bowlby was explicit that this was not a place on a map or a decade on a calendar; it was a statistical description of the conditions a system had been selected against, which for most human psychological systems meant the long stretch of small, mobile, hunter-gatherer social life that preceded agriculture.

John Tooby and Leda Cosmides took that idea and made it the organizing principle of evolutionary psychology as a field. Their argument, restated in the field’s standard handbook, is that complex functional organization in any organism is never a coincidence: wherever a system shows the kind of intricate, well-fitted design that could not arise from entropy alone, that design has to trace back to a specific history of selection pressures, and the correct explanatory target is “the recurrent structure of the organism’s ancestral world, rather than modern, local, transient, or individual conditions” [2]. That is a more disciplined claim than the popular version. It says a mismatch hypothesis has to specify which recurring ancestral pressure built which mechanism — not simply gesture at “the Pleistocene” as an undifferentiated golden age the present has fallen out of.

The concept also has a genuine mechanistic seed, and it did not come from psychology at all. Niko Tinbergen’s mid-twentieth-century ethology supplied the first rigorous demonstration that an evolved releasing mechanism can be triggered more strongly by an artificial stimulus than by the natural one it evolved to detect. Herring gull chicks peck at a red spot on the tip of an adult’s bill to solicit regurgitated food; when Tinbergen and his collaborators tested chicks against a series of models — a lifelike three-dimensional head, a bare bill, and a thin red-and-white-striped rod with no head at all — the striped rod without any head attached elicited more pecking than the accurate model did [3]. Separately, when gulls and other ground-nesting birds were offered a choice between their own eggs and oversized, more intensely marked dummy eggs, several species preferred to incubate the exaggerated dummy, in some documented cases sliding off it repeatedly rather than settle for their own [3]. Tinbergen called the exaggerated stimulus “supernormal”: one that exceeds the natural range of the cue an evolved detector responds to, and therefore captures the detector’s response more completely than anything that detector ever encountered during its evolution. That is not a metaphor borrowed from psychology and applied loosely to modern life. It is a specific, replicated, experimentally demonstrated mechanism, discovered in birds, decades before anyone extended it to people.

Put those three pieces together and a real audit standard falls out, rather than having to be invented from nothing. A mismatch claim earns the name only if it specifies, first, an evolved mechanism — not “the brain” in general, but a particular system with a describable function; second, an environmental novelty — a measurable, dated way in which the present differs from the conditions that shaped that mechanism; and third, causal evidence linking the two, ideally a dose-response relationship or an experiment, rather than a plausible-sounding correlation. Formally, treat a claim as confirmed only when all three hold at once:

\text{mismatch confirmed} \iff \text{mechanism specified} \;\land\; \text{novelty dated} \;\land\; \text{causal link demonstrated}

Figure 1. Before a mismatch claim about myopia can be trusted, the instrument measuring it has to be checked against a known standard first — the discipline the rest of this article asks other mismatch claims to meet. — Image prompt and art direction by Brecht Corbeel; generation pending.

Three modern cases will be run against that standard here: myopia, obesity, and the claim that screens and social media exploit evolved attention systems the way Tinbergen’s striped rod exploited a gull chick’s pecking reflex. One of the three clears every bar. One keeps a real mechanism but loses the specific evolutionary story to a genuine, still-unresolved dispute between two named theories. One borrows the vocabulary of supernormal stimuli without yet producing evidence of the kind Tinbergen actually had for gulls.

Myopia’s prevalence curve is the strongest circumstantial evidence mismatch has

Start with the numbers, because they are unusually large and unusually well documented. The U.S. National Academies’ 2024 consensus review of myopia reports that 80 to 90 percent of 17- and 18-year-olds in the urbanized populations of Hong Kong, Singapore, mainland China, Taiwan and Japan are myopic, drawing on the widely cited tallies assembled by Morgan and Rose and by Morgan and colleagues; in Seoul, among 19-year-old male conscripts, the rate exceeds 96 percent, and in Singapore the rate among young adults defined by a spherical equivalent worse than −0.50 diopters reaches 81.6 percent [6]. Within a single Chinese cohort tracked between 2012 and 2015, the same review reports a grade-by-grade gradient that both rural and urban children share in its shape and diverge in its endpoint: roughly 20 percent myopic by grades one through three in both settings, rising to 69 percent by grades seven through nine in rural schools versus 63 percent in urban ones, and reaching 81 percent in rural grades ten through twelve against 88 percent in urban grades ten through twelve [6].

That pattern is exactly what a mismatch claim should look like before any mechanism is even discussed. The trait tracks schooling intensity and urbanization far more tightly than it tracks ancestry: the same broadly East Asian gene pool produces dramatically different myopia rates depending on when and where a given cohort went through childhood, and the rate climbs in lockstep with each additional grade of indoor academic pressure. A purely genetic account has no obvious way to explain a curve that moves that fast within a population’s own recent history; an account built around a specific, dated environmental change does.

The environmental novelty leg of the audit is where randomized evidence, not just correlation, enters. He and colleagues ran a cluster-randomized trial across twelve primary schools in Guangzhou, China, assigning six schools to add a mandatory 40-minute outdoor activity class to every school day for grade-one children while parents were separately encouraged to increase outdoor time on weekends, and six schools to continue as usual. Among 1,903 children followed for three years, the cumulative incidence of myopia was 30.4 percent in the intervention schools against 39.5 percent in the control schools, a 9.1-percentage-point absolute reduction (95 percent CI, −14.1 to −4.1; p<.001), with a correspondingly smaller shift in spherical equivalent refraction, −1.42 diopters against −1.59 diopters (p=.04) [4]. A separate Taiwanese trial of a policy requiring at least 80 minutes of daily outdoor time, summarized in the 2021 International Myopia Institute consensus report, found myopia incidence falling from 17 percent to 8 percent under the policy, with the associated myopic shift reduced from 0.38 to 0.25 diopters [5]. Two independent trial populations, in two countries, using different outdoor-time thresholds, both show incidence falling in the same direction by a comparable relative magnitude — which is the closest thing epidemiology offers to a converging, dose-sensitive causal signal rather than a single lucky result.

Figure 2. The randomized trials that made myopia the strongest mismatch case ran on exactly this kind of instrument: a dose of daylight measured in hours, not inferred from a headline. — Image prompt and art direction by Brecht Corbeel; generation pending.

It is also worth naming, in the spirit of the audit standard rather than against it, a live question the field has not fully closed: whether the protective factor is outdoor activity specifically, or simply light intensity, which outdoor environments happen to supply in far greater quantity than almost any indoor one. If sufficiently bright artificial indoor lighting turned out to confer the same protection without requiring children to go outside at all, that would falsify “the outdoors” as the operative variable while leaving the underlying light-dependent mechanism intact. That the question can even be posed this precisely, and is being tested this way, is itself evidence of a mature research program rather than a slogan.

The mechanism is not a metaphor: dopamine, light, and an eye that keeps recalibrating itself

The reason myopia clears the mechanism leg of the audit so cleanly is that “the eye evolved for outdoor light” is not an untestable evolutionary story bolted onto a correlation after the fact; it names a specific, independently studied biological feedback loop. The human eye does not simply grow to a fixed length and stop. Its axial length is actively regulated during childhood by a local process called emmetropization, in which the retina — specifically its peripheral and central regions — senses the eye’s own optical defocus and adjusts the signaling that controls how fast the sclera elongates, continuously matching the eye’s physical length to its optical power [5]. That is an evolved calibration system, not a passive default; it had to evolve because an eye’s final focal length cannot be specified by genes alone with the precision vision requires, so natural selection built a feedback loop that lets the growing eye tune itself against the actual visual environment it develops in.

Light intensity turns out to be a direct input to that loop, mediated by retinal dopamine. Animal studies summarized in the same consensus review found that brighter ambient light drives greater dopamine release from the retina, that dopamine and dopamine agonists slow axial elongation, and that sufficiently high light intensity can block the development of experimental myopia in animal models entirely, without any other experimental parameter being changed [5]. That gives the “evolved mechanism” leg of the audit real specificity: a system whose calibration signal depends on retinal dopamine release, itself driven by ambient light intensity, and shaped over evolutionary time by an environment in which children’s daylight hours vastly outnumbered their hours spent focusing at close range indoors.

Figure 3. Emmetropization is the eye's own feedback loop, not a metaphor: retinal dopamine signalling that keeps recalibrating how fast the eye grows against the light it actually receives. — Image prompt and art direction by Brecht Corbeel; generation pending.

The environmental novelty is equally specific rather than a vague appeal to “modern life”: sustained indoor schooling during exactly the developmental window in which this feedback loop is most active, replacing what for most of human history would have been long daily stretches of outdoor daylight combined with far less sustained close-range focusing. Put the three legs together — a described biochemical calibration mechanism, a dated and measurable change in daily light exposure and near-work load, and two independent randomized trials showing that restoring outdoor light exposure changes the trait in the predicted direction with a plausible dose-response — and myopia is the case the mismatch framework was made for. It is not surprising that it is also the case most often cited by the framework’s defenders. It is, on the evidence actually assembled here, the case that deserves to be.

Obesity keeps a real mechanism but loses the specific evolutionary story

Obesity is where the audit gets harder, and where a careful writer has to resist the temptation to declare a winner the literature has not declared. The starting hypothesis is James Neel’s, proposed in 1962: a “thrifty genotype,” defined as one “exceptionally efficient in the intake and/or utilization of food,” would have been favored by natural selection across the vast majority of human history, because, as Neel put it, “during the first 99 percent or more of man’s life on Earth, while he existed as a hunter and gatherer, it was often feast or famine” [7]. Under recurring famine, the capacity to store fat efficiently after a period of abundance would have improved survival odds; under the caloric abundance and physical inactivity of industrialized life, the same genotype becomes, in Neel’s own phrase, “rendered detrimental by progress” [7]. For decades this was treated as the obesity field’s founding mismatch story, and it is easy to see why: it names a mechanism, a novelty, and a plausible causal link in a single elegant sentence.

The trouble is that the elegance outran the demographic evidence, and a serious rival account now occupies the same journal issue as its most cited defense. John Speakman’s 2008 critique argues that the thrifty-genotype logic does not survive scrutiny of how famine mortality and famine fertility actually interact: differential survival between leaner and fatter individuals during a famine is, on its own, too small and too rare an event to drive strong selection, and — more importantly — famines are almost universally followed by periods of enhanced fecundity that offset the reproductive cost incurred during the famine itself, so that the net reproductive advantage of carrying “thrifty” variants across a full feast-famine cycle is far smaller than Neel’s account assumes [8]. In place of positive selection, Speakman proposes a “drifty gene” hypothesis: a release from heavy, size-dependent predation pressure on hominins roughly two million years ago would have relaxed an earlier selective ceiling on body fat — being conspicuously large or slow was a liability against predators before that release — after which genetic variants affecting adiposity were free to drift neutrally, unconstrained by either strong positive or strong negative selection. On this account, the wide modern variance in obesity susceptibility is not evidence of an ancestral adaptation for thrift at all; it is the accumulated, non-adaptive byproduct of several million years of relaxed selection on a trait that used to matter and, for most of that period, largely stopped mattering [8].

That is not the end of the argument, and it should not be treated as one. Andrew Prentice, Branwen Hennig and Anthony Fulford published a direct rebuttal in the same 2008 issue of the same journal, arguing that a mixture of famines and seasonal food shortages during the post-agricultural era did exert real natural selection favoring fat storage — but operating chiefly through differential fertility rather than differential survival, meaning individuals who stored fat more efficiently reproduced more successfully across recurring lean seasons even without a large mortality gap [9]. Their paper explicitly frames the dispute as unresolved and looks to newer genetic and bioinformatic methods to adjudicate it. Nearly two decades later, that adjudication has not produced a clean verdict either way; reviews of the cardiometabolic-disease literature continue to treat the thrifty-genotype and drifty-gene accounts as live, competing hypotheses rather than a settled question with one surviving theory. The honest summary is that obesity keeps a real, biologically plausible evolutionary mechanism — heritable variation in metabolic efficiency and appetite regulation clearly exists and clearly matters — while the specific story of how natural selection produced that variation remains a genuine, named dispute between serious researchers, not a case either side has won.

Figure 4. Neel's thrifty genotype and Speakman's drifty-gene alternative are two live evolutionary accounts of the same metabolic variance, and the dispute between them is still open. — Image prompt and art direction by Brecht Corbeel; generation pending.

The food environment does not need the ancestral story to be true

The discipline the audit standard demands here is separating two claims that get bundled under one label. One claim is about which evolutionary process shaped the genetic variance in human appetite and metabolism — that is the thrifty-versus-drifty dispute, and it remains open. The other claim is that the modern food environment causally drives people to eat more than they otherwise would, independent of which evolutionary story about genetic variance turns out to be correct. That second claim does not need the first one settled, and it has much stronger direct evidence behind it.

Kevin Hall and colleagues ran the relevant test at the NIH Clinical Center in 2019: twenty weight-stable adults were admitted as inpatients and given, in randomized crossover order, two weeks of an ultra-processed diet and two weeks of an unprocessed diet, with every meal in both conditions matched for presented calories, energy density, sugar, sodium, fiber and macronutrient composition, and participants told simply to eat as much or as little as they wanted [10]. Despite that matching, ad libitum energy intake was 508±106 kcal per day higher on the ultra-processed diet than the unprocessed one (p=.0001), driven by increased intake of carbohydrate (+280±54 kcal/day, p<.0001) and fat (+230±53 kcal/day, p=.0004) but not protein; the excess appeared at every meal, with breakfast up 144±39 kcal/day (p=.0014), lunch up 248±39 kcal/day (p<.0001) and dinner up 108±41 kcal/day (p=.017) [10]. Because the same twenty people ate both diets, two weeks each, with nutrient content held equal on paper, the design isolates the food matrix itself — something about how ultra-processed foods are engineered, independent of their stated nutrient content — as the causal driver of excess intake. Nothing about that result depends on resolving whether human appetite regulation evolved under thrifty selection, drifty selection, or some third story entirely.

Figure 5. Hall's ad-libitum trial isolated the food matrix from genetics entirely: twenty adults, the same two weeks each way, only the processing of matched-nutrient meals changed. — Image prompt and art direction by Brecht Corbeel; generation pending.

That is the useful way to hold the obesity case in mind: leg (b), the environmental novelty, is about as well specified as this kind of claim gets — an industrially engineered food matrix with no close ancestral analog, tested against a matched control in a randomized trial; leg ©, the causal evidence, is a clean crossover RCT with a large, precisely quantified, mechanistically legible effect. Leg (a), the specific evolved mechanism responsible for the genetic variance the modern food environment now acts on, is the part that remains genuinely contested between Neel’s heirs and Speakman’s. Treating obesity as a single mismatch claim that stands or falls together collapses two questions that the evidence answers very differently, and an honest account keeps them separate.

Screens borrow the mechanism’s rhetoric and have not yet earned its evidence

The claim that social media, infinite-scroll feeds and algorithmic notifications function as supernormal stimuli — artificial triggers exploiting an evolved reward system more intensely than anything that system evolved to handle — is the most direct modern extension of Tinbergen’s own logic, and it is worth taking that lineage seriously rather than dismissing it by association. Variable-ratio reward schedules, infinite content, and engagement systems explicitly optimized against human attention are real design features of modern platforms, and dopaminergic reward-prediction-error signaling is a real, well-studied neural mechanism. The question the audit has to ask is narrower and less flattering to the popular version of the claim: does the causal evidence show this mechanism producing effects anywhere near the scale the rhetoric implies?

Amy Orben and Andrew Przybylski’s 2019 specification curve analysis is the largest attempt yet to answer that question at the population level. Applying the method across three major datasets totaling 355,358 adolescents, they found that digital technology use explains at most 0.4 percent of the variation in adolescent wellbeing — a negative association, but one comparable in size to, or smaller than, the negative associations between wellbeing and other mundane behaviors measured in the same datasets, including eating potatoes and wearing glasses [11]. That is a genuine practical-significance check, and it cuts hard against any claim that screen use is doing to population-level wellbeing what a supernormal stimulus does to a gull chick’s pecking rate. Whatever screens are doing, at the scale of a representative national sample it is not obviously more consequential than a handful of other ordinary correlates of adolescence.

That is not, however, the whole picture, and reporting only the null-leaning correlational result would itself be a failure of the audit standard’s insistence on both sides of a live dispute. Two well-identified studies of specific interventions find real, if narrower, causal effects. Luca Braghieri, Ro’ee Levy and Alexey Makarin exploited the staggered rollout of Facebook across individual American college campuses between 2004 and 2006 in a generalized difference-in-differences design, and found that a college’s Facebook rollout caused a measurable decline in student mental health and increased the share of students reporting that poor mental health had impaired their academic performance; their evidence on mechanism points specifically to unfavorable social comparison, not to attention capture or reward-circuit hijacking [12]. Hunt Allcott, Luca Braghieri, Sarah Eichmeyer and Matthew Gentzkow ran a genuine randomized experiment, paying a sample of 2,743 Facebook users to deactivate their accounts for four weeks before the November 2018 U.S. midterm election; deactivation reduced online activity, increased offline activities including watching television alone and socializing with family and friends, reduced both factual political news knowledge and political polarization, increased self-reported subjective wellbeing, and produced a persistent reduction in Facebook use and in participants’ own valuation of the platform even after the experiment ended [13].

Figure 6. The same instrument that made myopia's case now sits mostly empty on the shelf reserved for screen and attention research — a thinner evidence base, shown rather than stated. — Image prompt and art direction by Brecht Corbeel; generation pending.

Reconciling those two bodies of evidence is more informative than picking one. The population-representative correlational evidence sets a real, tight upper bound: ordinary variation in how much time an ordinary teenager spends on a screen does not explain much of the variance in how that teenager is doing. But well-identified natural and randomized experiments studying a specific exposure — a platform’s arrival at a specific campus, a month away from a specific app — do find real, causally attributable, though still modest, effects, and the mechanisms those studies actually name are social comparison and the displacement of offline time and attention, not a generic story about dopamine being “hijacked.” That distinction matters for the audit’s first leg. Tinbergen could demonstrate supernormality directly, because he could put a real gull egg and an exaggerated dummy side by side and measure which one the bird chose to incubate; the natural stimulus and its exaggerated substitute were both concretely measurable objects with known dimensions. There is no equivalent measurable “ancestral analog” to a social media feed to run the same substitution test against — no known evolved releaser with a specifiable natural range that a feed can be shown to exceed. Dopaminergic reward-prediction-error signaling is engaged by nearly every rewarding or uncertain stimulus a person encounters, including food, novelty, and the light-driven pathway that governs myopia’s own mechanism; its mere involvement in screen use is not evidence that screens are supernormal in Tinbergen’s specific, falsifiable sense. That missing specificity, more than any flaw in the neuroscience, is why “dopamine hijacking” rhetoric currently outruns the evidence it claims to summarize.

What separates an audited mismatch claim from a headline

Running three cases through the same three-part test makes the differences concrete rather than rhetorical. Myopia specifies its mechanism precisely — retinal dopamine release regulating an emmetropization feedback loop that calibrates axial eye growth to ambient light — dates its novelty precisely, as a measurable reduction in daily outdoor light exposure concentrated in exactly the developmental years the feedback loop is most active, and supports the causal link with two independent randomized trials in two countries showing convergent, dose-sensitive reductions in incidence. It can also state a clear falsifier: if bright artificial indoor lighting, matched to outdoor lux levels, produced the same protective effect without requiring outdoor exposure at all, that would show the mechanism runs through light intensity rather than “the outdoors” as such, and the field is actively pursuing exactly that test rather than treating the current framing as beyond revision.

Obesity specifies a real mechanism in outline — heritable variation in appetite and metabolic efficiency — but the specific evolutionary account of how that variation arose remains a named, unresolved dispute between Neel’s thrifty-genotype descendants and Speakman’s drifty-gene alternative, with a serious rebuttal from Prentice and colleagues keeping the argument open on both sides rather than settled in either direction. That should lower confidence in any single evolutionary narrative about why obesity exists, without lowering confidence at all in the separate, well-supported causal claim that the engineered modern food environment drives excess intake through a mechanism that does not depend on that narrative being resolved.

Screens fail the audit differently again, and more instructively. The proposed mechanism — supernormal reward-stimulus capture — cannot currently be tested the way Tinbergen tested his gulls, because there is no measurable ancestral comparator to exceed. The causal evidence at the population level is small enough to fail a plain practical-significance test, even though better-identified studies of specific exposures find real, if modest, effects running through named mechanisms — social comparison, displaced offline time — that are not the mechanism the popular claim actually invokes. A claim that mixes a weak population-level correlation, real but differently-mechanized causal effects from separate specific studies, and a mechanism borrowed by analogy rather than demonstrated by substitution, is not yet an audited mismatch claim. It might still turn out to be one; the honest position is that it is not one yet, and that saying so is not the same as saying screens are harmless.

Mismatch is a research program, not an explanation

The lesson across all three cases points in the same direction. Evolutionary mismatch earns its keep not as a sentence that closes an argument but as a hypothesis generator that hands a specific, falsifiable prediction to the discipline actually equipped to test it. Myopia succeeded because the evolutionary hunch — a light-calibrated growth system meeting a newly dim, newly close-focused childhood — was immediately convertible into a cluster-randomized field trial with a measurable dose and a measurable outcome, and two independent research groups in two countries ran that trial and got a convergent answer. Obesity and screens have not failed as topics; they have simply not yet been handed off the same way, or the handoff has produced a genuinely split verdict that a responsible account has to carry forward rather than collapse into a single headline number.

None of this makes the evolutionary framing decorative. Without Bowlby’s environment of adaptedness and Tooby and Cosmides’s insistence that function must be explained by a specific history of selection, nobody would have gone looking for a light-dependent retinal feedback loop in the first place, and without Tinbergen’s gulls there would be no rigorous concept of a supernormal stimulus to test screens against and find wanting. What the myopia case actually demonstrates is the right division of labor: evolutionary theory generates the hypothesis worth testing, and ordinary epidemiology — cluster randomization, dose-response curves, independent replication across populations — decides whether it survives contact with data. A mismatch claim that cannot say in advance what result would prove it wrong is not yet a scientific claim, whatever else it might be. The three-part test is not a hurdle the field resents clearing. It is the entire difference between a theory that explains one thing carefully and a phrase that explains everything and therefore nothing.

Sources

1. Kevin Bennett. Environment of Evolutionary Adaptedness (EEA). Encyclopedia of Personality and Individual Differences (Springer) (2018). DOI: 10.1007/978-3-319-28099-8_1627-1.
2. John Tooby and Leda Cosmides. The Theoretical Foundations of Evolutionary Psychology. The Handbook of Evolutionary Psychology, 2nd ed., Vol. 1 (David M. Buss, ed.), John Wiley & Sons (2015).
3. Wikipedia contributors, summarizing Tinbergen (1948, 1951, 1953). Supernormal stimulus. Wikipedia (2026).
4. Mingguang He, Fan Xiang, Yingfeng Zeng, and colleagues. Effect of Time Spent Outdoors at School on the Development of Myopia Among Children in China: A Randomized Clinical Trial. JAMA (2015). DOI: 10.1001/jama.2015.10803.
5. Jost B. Jonas, Marcus Ang, Padmaja Cho, and colleagues. IMI Prevention of Myopia and Its Progression. Investigative Ophthalmology & Visual Science (2021). DOI: 10.1167/iovs.62.5.6.
6. National Academies of Sciences, Engineering, and Medicine. Understanding Myopia and Its Prevalence. National Academies Press (2024).
7. James V. Neel. Diabetes mellitus: a "thrifty" genotype rendered detrimental by "progress"?. American Journal of Human Genetics (reprinted, Bulletin of the World Health Organization, 1999) (1962).
8. John R. Speakman. Thrifty genes for obesity, an attractive but flawed idea, and an alternative perspective: the 'drifty gene' hypothesis. International Journal of Obesity (2008). DOI: 10.1038/ijo.2008.161.
9. Andrew M. Prentice, Branwen J. Hennig, and Anthony J. Fulford. Evolutionary origins of the obesity epidemic: natural selection of thrifty genes or genetic drift following predation release?. International Journal of Obesity (2008). DOI: 10.1038/ijo.2008.147.
10. Kevin D. Hall, Alexis Ayuketah, Robert Brychta, and colleagues. Ultra-Processed Diets Cause Excess Calorie Intake and Weight Gain: An Inpatient Randomized Controlled Trial of Ad Libitum Food Intake. Cell Metabolism (2019). DOI: 10.1016/j.cmet.2019.05.008.
11. Amy Orben and Andrew K. Przybylski. The association between adolescent well-being and digital technology use. Nature Human Behaviour (2019). DOI: 10.1038/s41562-018-0506-1.
12. Luca Braghieri, Ro'ee Levy, and Alexey Makarin. Social Media and Mental Health. American Economic Review (2022). DOI: 10.1257/aer.20211218.
13. Hunt Allcott, Luca Braghieri, Sarah Eichmeyer, and Matthew Gentzkow. The Welfare Effects of Social Media. American Economic Review (2020). DOI: 10.1257/aer.20190658.

Originally published at https://absolutedigitalpublishers.com/articles/stone-age-minds-in-traffic-auditing-the-mismatch-hypothesis.