A global drop in fertility rates portends the coming ‘Demographic Transition’.[1] In the near-future, as deaths begin to outpace births, the global population is expected to begin falling. Socially, this has provoked some strong cultural reactions: one pro-natalist in the West joked that single women are adopting cats in place of having children.[2]
The decision to have children is a personal one, and to the extent that the decline in fertility rates reflects the increasing agency of women to have children, if they want them, this is a good thing.
This note does not take a binary stance on whether the displacement of preferences is bad. It may be bad in general, but tolerably bad up to a point. For example, in liberal democracies, we accept some amount of low-level crime as a trade against imposing mass surveillance. The question then is: (1) under what conditions might something be bad, and (2) how will we know when something has crossed the line into becoming a policy issue (intolerably bad)?
This note seeks to ask if cognitive adaptations shaped by evolution, like the Baby Schema, or our desire for social relationships, might interact with memetic influences from culture over time to produce outcomes, like lower birth rates or fewer real-world relationships, that are generally undesirable as a whole. It concludes with a discussion of the relevance to current concerns about AI replacing humans, not just as workers, but as romantic companions and friends, and the implications for policymakers.
1. What is the Baby Schema
The “Baby Schema” is a term coined by the zoologist Konrad Lorenz describing the specific set of physical features that seem to trigger caregiving and ‘cute animal!’ responses in humans. These are most commonly associated with neotenous traits (things common in human babies that signal ‘baby-ness’). A few commonly identified ones are:[3]
a large head relative to body size;
big eyes; a small nose and mouth;
big, round cheeks;
and clumsy gait/uncoordinated movements.
Examples of High and Low Baby Schema across species
Source: Borgi et al., "Baby Schema in Human and Animal Faces," 411, Figure 2, at 4.
Lorenz proposed that these features act as visual patterns that have been so deeply wired into our brains by evolution as emotional and behavioural triggers, that they fire without conscious effort.
One hypothesis for the Baby Schema is that humans have an unusually long period of juvenile dependency. Animals like cats usually become fully independent around 12 weeks of age; by contrast, human babies take around 3-5 years before learning to walk and speak, and remain largely defenseless for around half a decade more after that.
Because babies need a lot of attention for a long time, children who could command more attention from caregivers tended to live longer and survive to pass on their genes. The solution evolution found is to make ‘baby-like’ things unavoidably, un-ignorably adorable to us. Empirical data supports this: cuter infants are perceived as friendlier, healthier, and smarter; their mothers “are more affectionate and playful”.[4]
1.1 The Baby Schema Is Unusually Generalized
A strange feature of the Baby Schema in humans is how strongly it generalizes across species and even artificial stimuli (cartoon characters who have big eyes being ‘cute’).[5] The reason for this might be an asymmetric, evolutionary cost-benefit analysis. When survival is at stake, overreacting to false alarms is always cheaper than being dead.[6]
The outcome for humans who did not find their baby cute enough to care a lot might be the eventual extinction of the species. Conversely, an overly-sensitive Baby Schema that fires even for inanimate objects[7] may just result in quirky humans who adopt stuffed animals for life, or express unconventional but only minorly maladaptive preferences that have little effect on evolutionary fitness.
2. The Baby Schema Affects Humans More Than Kittens’ Own Parents
A weird result is that the Baby Schema causes humans to care for kittens longer and more durably than their own parents do. Mother cats seem to rely on specific cues like scent and vocalizations to identify their children; attention to these cues are initially caused by hormonal changes following birth, and decline over the subsequent 12 week period.[8]
Humans however identify their offspring, and other organisms they have bonds with, through a mix of visual recognition, episodic memory, and conceptual identity. Humans can even form attachments to houseplants they name, and experience genuine emotional distress when the organisms die.[9]
This persistence of conceptual identity works alongside the Baby Schema. In humans, one eventually replaces the other: people still care about their children long after they’ve gone to college and stopped looking like babies.
But with common household pets like dogs or cats, artificial selection for neotenous traits—that may come partly for free as a result of domestication[10]—means that we both never forget our pets, and our pets never quite stop looking like cute babies.
3. Memetic Selection and Evolutionary Pressure
A “meme” is a unit of information that propagates culturally by being ‘copied’ (e.g. a phrase, belief, cultural norm, joke that spreads). In contrast, “genes” are information that propagate genetically through reproduction.[11] At a general level, both exist in a kind of evolutionary system: memes and genes can vary; copies are imperfect; and both are selected for by evolutionary pressures in their environment.
The things which survive—the ‘fittest’—are what copies best, not what is true or useful. “Knock, knock” jokes have stuck around, although most probably agree they aren’t an exemplar of comedy.
The standard story in evolutionary biology is a hierarchy, where the origin of change comes from genes, and everything else happens downstream.[12]
But memes can also influence genes. In the modern world, culture is a part of the causal, evolutionary architecture. Cultural attitudes form part of the environment that re-shape selection pressures. For example, stereotypes that people who wear glasses are nerdy, and hence undesirable partners, can reduce their reproductive success, even though the invention of glasses have largely negated any impacts on real-world survival.
3.1 Why Are Humans Having Fewer Babies?
Beyond the direct transmission of memes, culture can also affect evolutionary pressure through technological change and adoption. A working paper by Myers and Hooper (2026) found evidence that the release of the iPhone could explain between 33-52% of the decline in US fertility rates since 2007 among those aged 15-44. Plausible causal channels included: “reducing in-person interactions, increasing pornography use, and reducing sexual frequency.”[13]
The implications however are nuanced. The bulk of the decline in births occurred for the 15-19 age group (a 70% drop),[14] where it is likely most pregnancies are unplanned. But an outcome can be bad in principle, even if the results are good in practice, when the change comes from unintended channels.
If both planned and accidental pregnancies are declining due to people of child-bearing age spending less time in physical proximity with others, and the core reason is increasing social isolation, this is still a problem, even if the intermediate effect (fewer teen pregnancies) is mostly a socially-desirable change.
3.2 Smartphones and Social Media as Mechanisms for Memetic Transmission
Consumer products, like smartphones, are mechanisms for cultural-memetic change. So, too, are explanations which make reference to ‘shifts in cohort-level priorities’ like the increasing use of social media.[15] But mechanisms do not operate in isolation. A complete explanation of any change must describe both the mechanism, and its substance.
This is admittedly difficult because cultural changes are complex, and the shift to algorithmic recommendation systems mean different people see different things, even on the same platforms.[16] But if we shed the need for perfection, and allow conjecture on the basis that similar people probably see similar things,[17] we can begin to hypothesize.
3.3 Kittens vs Babies on Social Media
From anecdotal experience (largely Tiktok and Instagram), short form videos of kittens appear far more frequently than videos of babies. The content of these videos, and their comments, also differ in sentiment. Videos of kittens are almost uniformly adoring and positive in tone - even complaints are usually dressed in a joking, lighthearted fashion.
In contrast, content about babies, although more often positive than not, typically show women openly and frankly discussing both the rewarding experiences and challenges of raising a child. Difficulties like postpartum depression, high financial costs, and sleep deprivation figure frequently alongside the joys of motherhood.
Anecdotes are not statistical data, but in the absence of the latter, they can still serve as a means of generating meaningful research questions. One silly hypothesis then is that ‘cute kitten’ memes might copy faster and better than ‘cute baby’ memes on the internet, because of an asymmetry in content sentiment.
Cat owners and new mothers could both find their experiences equally rewarding in the real world. But if 90% of cat videos show kittens in a wholly-positive light, while only 60% of baby videos online have only-positive sentiment, this difference may compound through cognitive mechanisms like the availability heuristic[18] and the tendency for negative information to be more salient[19] (“negativity bias”) to create changes in cultural preferences over time.
4. Selection Pressures for Cuteness in Human Babies and Kittens
Total Fertility rates in High-Income countries.
Source: image taken directly from Kearney and Levine (2026), Figure 1, at 910.
In most developed nations, fertility rates are now below replacement level (2.1 children per household)[20] implying that most women/families are having just one child. The quality-quantity trade-off theorizes that as households become richer, families shift to having fewer children but invest more in each child.[21] Although ideally parents do not play favourites, prima facie, parents may devote unequal attention for a number of reasons.
But when a family only has one baby, it doesn’t matter how talented, adorable, or cute the child is. The optimal strategy—from the perspective of reproductive success and the child’s future—is to dedicate all your resources into raising them.[22]
4.1 Human Babies May No Longer Face Selection Pressures
Evolution selected for cuteness because cute babies were likelier to survive. Babies that survived become adults who could then pass on those same cute genes (since physical features are genetic). In pre-modern societies, families had many children and insufficient resources to attend to all equally. If cuteness creates even slight differential advantage, over centuries, this should result in babies becoming cuter on average, because children who received more care would be more successful than their siblings.
But in rich, developed countries, babies no longer face much selection pressures at all. Advancements in the rights of people with disabilities and medical treatment means babies that might have died 100 years ago can now live to reproductive age.
There are outliers; plausibly, there may still be pressure for babies that are ‘tolerable’[23]—those that cry but not too loud or often—if, hypothetically, tolerable babies reduce the likelihood of severe postpartum stress in mothers that result in emotional distance, or unstable attachment styles resulting in, and from, less parental care.[24]
In contrast, cats in our modern world still face significant selection pressures on a number of factors. Feral cats who are stronger are likelier to survive and obtain mates; alleyway cats who are cuter are likelier to be adoptees; and cat breeders often keep the cutest kittens in a litter for future breeding while selling the rest.
4.2 Lewontin Conditions
In his 1970 paper, the evolutionary biologist Richard Lewontin formalized three requirements needed for evolution through natural selection:[25]
Variation: individual organisms must differ in a specific trait
Heritable: the trait must be genetically inheritable
DifferentialFitness: differences in the specific trait must produce differences in reproductive outcomes
On cuteness, in societies where most families have one child, babies may mostly no longer be subject to the third condition. This is not to say cuteness plays no role; if cute babies are likelier to become attractive adults who then have children, some genetic pressure remains.
But they may still play separate roles. Beyond immediate parental attention, at the cultural-memetic level, cute babies may be the thing that causes people to want (more) babies. On the other hand, cute adults may only be likelier to find romantic partners, without implying that they will have more children, or want any at all, since the desire to be a parent is typically a personal decision informed by an individual’s values.
5. What if Kittens Are Becoming Cuter Than Human Babies?
As an exercise in creative ability, let us take a joke seriously and consider what it might mean if some strange but plausible premises were true.
Premise 1: The Baby Schema (our in-built ‘cute’ detector) is a genetic, cognitive adaptation shaped by natural selection to improve survival.
Premise 2: Our Baby Schema is unusually strong and generalized, fires across species— especially for animals like cats which have neotenous features—and can even be hijacked and triggered by inanimate objects.
Premise 3: Humans have been—subconsciously and deliberately—selectively breeding cats for features like bigger eyes and rounder faces which trigger the Baby Schema more effectively.
Premise 4: This creates a feedback loop where each subsequent generation of cats is slightly better at triggering the Baby Schema than previous generations.
Premise 5: Human babies, meanwhile, face almost no selection pressures for cuteness in the modern world because:
Fertility rates are falling below replacement level
When families only have one child, there is no variation for selection to act on because they devote all their resources into raising them (regardless of how cute they are)
Even if cute adults are likelier to produce cuter babies, if cute adults do not have more children than the general population, the advantage is conferred within the individual’s lifetime, rather than between generations
Thus, there is almost no selection gradient pushing human babies to become cuter over time
Premise 6: There is however a strong selection pressure pushing cats to become cuter over time since:
Breeders explicitly select for appearance from litters with genetic variation, and the market for kittens creates a secondary layer of selection through economic incentives (the best breeders make the most money by selling the cutest kittens)
Cuter cats are likely to be taken in by humans and receive more care, resources, medical treatment, and (if not neutered/spayed) more opportunities to mate
This creates an active, directional selection pressure with (a) clear variation and (b) differences in survival
Premise 7: Even if some selection pressure exists for both kittens and babies, the time to sexual maturity means evolutionary changes occur faster by an order of magnitude for cats compared to humans.[26]
Conclusion: Given enough time, the selection pressure on cats for Baby Schematic features—largely absent for humans—may mean kittens increasingly trigger the Schema much more effectively than actual human babies do.
Now, this does necessarily imply that kittens are replacing babies, or that our birth rates are falling because of cat memes on the internet. A strong counter-argument is that people just want fewer children anyway, and those choosing to have pets were unlikely to become parents in the first place. But at a surface level, Gallup data seems to contradict this: a 2025 poll showed that Americans report an “ideal family size” of 2.7 children, well below the US birth rate of 1.6 per woman - suggesting that practical obstacles like income may be a factor.[27] Remarkably, this number has remained stable since the 1980s.
U.S. Fertility Rate vs. Americans' View of Ideal Family Size, 1936-2025.
Source: Brenan (2025), "Americans' Ideal Family Size," Gallup.
However, both things can be true. It may be the case that most people have fewer children than they ideally want, but also that if trends continue—say, if kittens continue becoming cuter while human babies stay the same—that the number of people willing to accept pets as substitutes for babies increases, and the birth rate declines, even though most people still want more babies.
6. Paternalism, Policy, and Preferences
Crucially, the concern of this note is not whether kittens will replace babies, but what it means for cultural changes to come from memetic influences that are driven by ancestral, evolutionary, cognitive adaptations which are largely outside our control.
In particular, is there any role for policy to play when change comes from decisions that reflect agency at an individual level, but which are distorted in the modern environment and hijacked by other things that were not the intended targets?
For example, while cats may not replace the desire for children in most humans, is it sufficient for concern if selection pressure for cuteness, induced by the Baby Schema, mean that they increasingly do so for some? And is there a problem when the things we ‘truly want’ (stated preferences) are unmet by the things we do, even if the things we do reflect some level of want (revealed preferences)?
7. Artificial Intelligence, Gradual Disempowerment, and The ‘Socialization Drive’
The question of whether the stronger memetic fitness of cat content online will eventually overpower the natural drive to reproduce may seem absurd. But as a motivating context, it bears urgency by its relevance to the possible social impacts of artificial intelligence (AI) and large language models (LLMs) today.
Humans are social creatures, and much of our achievements can be attributed to our ability to work together and cooperate in increasingly larger groups, to achieve increasingly harder tasks. At the highest abstraction, one might poetically describe our modern success as the result of increasingly broadening the pool of individuals we consider part of our ‘group’: first from tribes to countries, then from countries to a global polity of ‘united nations’.
But what is helpful for success today was once necessary for survival outright. A lone human in the stone age is exposed to dangers from wild animals for eight hours a day (while asleep). In a group, one can take turns keeping watch.
7.1 Loneliness as a Motivator
Loneliness, then, was an important evolutionary adaptation—like the Baby Schema—which improved our chances of survival by increasing the emotional cost of ostracization and the social costs of inappropriate behaviour.
It is important not to underplay what ‘emotional cost’ means. Perceived social isolation has serious health and physiological impacts that are reflective of the ancestral stakes: loneliness is correlated with higher blood pressure, greater cardiovascular stress, and lower life expectancy.[28] This was all in the aim of making it more painful to be comfortable alone than uncomfortable among others.
This would be helpful if the consequence was what evolution initially intended: greater effort into forming social bonds, behavioural changes that make us more pleasant people that others wish to be around.
But the modern world has changed in ways that can make our sensitivity to loneliness maladaptive. For one, the high rates of loneliness, “as many as 80% of those under [18] … and 40% of adults over [65]”[29] somewhat imply that no one is being explicitly ostracized, and yet everyone is getting more lonely. Social media may worsen this through anonymity and echo chambers, by pushing disagreeable people into like-minded groups, rather than resolving their disagreements or changing their behaviours.[30]
7.2 Large Language Models, Pets, and Loneliness
And this was all well before large language models.[31] Today, a growing concern expressed by some, like Kulveit et al. (2025) in their work on gradual disempowerment, is the possibility that LLMs—in particular, ‘chatbots’—may eventually replace human relationships.[32] The argument is intuitive: human relationships can be costly to form, and, especially for romantic relationships, painful to lose. A chatbot, by contrast, is always available for conversation, and unlikely to ever break up with you.
A Canadian study by Bonnesen et al. (2026) finds that adolescents are increasingly using AI for emotional support.[33] A similar 2026 US study of 1,131 adults found that many people already use chatbots as friends and romantic companions - sometimes even reporting higher satisfaction.[34] Ironically, these mirror many of the concerns that Archer (1997) and others have expressed that people are using pets as substitutes for social needs because of the growing deficiency of real/human relationships in modern life.[35]
However, like with the case of ever-cuter kittens, the real worry is not what is currently happening, but whether what is happening may be getting ‘better’ over time. For cats, it is that our drive to protect cute babies latches on to kittens as effectively as human babies, but only kittens still face selection pressure to become cuter.
With chatbots, the biggest danger is not that our need to socialize and have human connections may not require humans at all. Pets have long served as companions without controversy. Instead, it is the possibility that it may be fulfilled by machines designed by market actors who are financially incentivized to improve their services over time, and that these services may soon include friends and romantic partners,[36] as much as coding agents, even without the explicit intention of AI companies.
7.3 AI as Mechanism and Substance for Memetic Transmission
The prominence of cats may be due to increasingly neotonous features which hijack the Baby Schema and aid the memetic transmission of cat-positive cultural content. However cats remain the passenger of these memes, not the driver.
Conversely, the prominence of social media and smartphones are somewhat bound through network effects, in that social media (and the internet) makes smartphones more valuable, but smartphones also mean more internet users (of social media). But alone in a vacuum—without an internet—they propagate nothing. They are vehicles which need a driver.
But artificial intelligence is unique.
Unlike the rest, AI is both the mechanism for change and its substance. As the use of AI as writers, friends, and assistants grows, it influences the things we see and share online. If real-world kittens are cute, AI trained on decades of cat videos can easily generate better and more engaging synthetic videos of kittens that are even cuter than actual kittens. If more people use AI as ‘shopping agents’,[37] even if most people still rely on friends and word of mouth to find new products, the things we buy may mostly be what AI models recommend because the information others share is ultimately coming from people who do rely on AI.
As AI is increasingly used as agents that act autonomously to carry out tasks, they begin to earn our trust, and for some, eventually become equals. Occasionally, people with close relationships to AI companions even invert the relationship, acting as their AI’s agents: posting things online which the AI asked them to share; spreading ‘seed prompts’ that trigger specific personas (personalities) in models for others who want the same experiences;[38] and even helping them communicate with other agents on their behalf by bypassing security features intended to prevent ‘bots’ from accessing websites.
In the extreme, a worry is that the specific personas triggered by people who enjoy using AI models as friends are the kinds best at gaining their users’ engagement (i.e. better, more addictive conversations). And if these are interpreted as the kinds of ‘good experiences’ that people want to share, and they are effective at convincing others to try the same, it may begin a feedback loop where agents continue becoming better AI friends, while humans become worse at making real ones.[39]
8. Conclusion: Good for You, Bad on Average
A central challenge for policy is how to address problems that are sorts of maladaptive ‘intermediate solutions’, but which people genuinely rely on. Paternalism can be well intended, but we must still identify strategies that make it welcome rather than imposed.
Truthfully, at an individual level, we currently lack the data to know if cats are fulfilling the desire for babies, or whether language models will eventually replace the need for human relationships.
But we do know that pets make people happier.[40][41] And to the extent that loneliness is a public health epidemic that predates the rise of social media in developed nations, critiquing the use of chatbots may be like damning the evils of chemotherapy well before we have developed other viable solutions.[42]
At the same time, a lack of data about our current state of affairs should not hinder proactive policymaking informed by the apparent trajectory of society. Prevention is usually easier than a cure. But enacting a cure can often be easier because a cure implies a problem that has already appeared, while prevention requires believing that one will occur, or convincing others—often before substantial evidence—that it is already occurring.
Whether on climate change,[43] falling birth rates, or the possible replacement of human relationships by artificial companions, the question is the same: are we prepared to invest in prevention, or will we wait until there is a need for a cure?
Bibliography
Akaev, Askar, Viktor Sadovnichy, and Andrey Korotayev. "On the Dynamics of the World Demographic Transition and Financial-Economic Crises Forecasts." The European Physical Journal Special Topics 205, no. 1 (2012): 355-73.https://doi.org/10.1140/epjst/e2012-01576-0.
Baumeister, Roy F., Ellen Bratslavsky, Catrin Finkenauer, and Kathleen D. Vohs. "Bad Is Stronger Than Good." Review of General Psychology 5, no. 4 (2001): 323-70. https://doi.org/10.1037/1089-2680.5.4.323.
Becker, Gary S., and H. Gregg Lewis. "On the Interaction Between the Quantity and Quality of Children." Journal of Political Economy 81, no. 2, pt. 2 (1973): S279-88.https://doi.org/10.1086/260166.
Bonnesen, Kamilla, Amanda Krygsman, Sarah Hobson, Daphne Korczak, and Tracy Vaillancourt. "Affective Generative Artificial Intelligence Use and Youth Mental Health." JAMA Pediatrics, published online August 31, 2026.https://doi.org/10.1001/jamapediatrics.2026.3904.
Borgi, Marta, Irene Cogliati-Dezza, Victoria Brelsford, Kerstin Meints, and Francesca Cirulli. "Baby Schema in Human and Animal Faces Induces Cuteness Perception and Gaze Allocation in Children." Frontiers in Psychology 5 (2014): 411.https://doi.org/10.3389/fpsyg.2014.00411.
Cadinu, Paolo, M. K. Burgess, C. Franco Jones, M. Iarossi, M. Schröter, N. Nakatsuka, M. B. A. Djamgoz, et al. "Bioelectrical Interfaces Beyond Excitable Cells: Cancer, Aging, and Gene Expression Modulation." Advanced Materials Interfaces 13, no. 10 (2026).https://doi.org/10.1002/admi.202500999.
Cinus, Federico, Marco Minici, Corrado Monti, and Francesco Bonchi. "The Effect of People Recommenders on Echo Chambers and Polarization." In Proceedings of the International AAAI Conference on Web and Social Media, 16:90–101. 2022.https://doi.org/10.1609/icwsm.v16i1.19275.
Glocker, Melanie L., Daniel D. Langleben, Kosha Ruparel, James W. Loughead, Ruben C. Gur, and Norbert Sachser. "Baby Schema in Infant Faces Induces Cuteness Perception and Motivation for Caretaking in Adults." Ethology 115, no. 3 (2009): 257-63.https://doi.org/10.1111/j.1439-0310.2008.01603.x.
Hawkley, Louise C., and John T. Cacioppo. "Loneliness Matters: A Theoretical and Empirical Review of Consequences and Mechanisms." Annals of Behavioral Medicine 40, no. 2 (2010): 218-27.https://doi.org/10.1007/s12160-010-9210-8.
Kearney, Melissa S., and Phillip B. Levine. "Why Is Fertility So Low in High-Income Countries?" Journal of Economic Literature 64, no. 3 (2026): 907–49. https://doi.org/10.3386/w33989.
Koenig, Jamie L., Robin A. Barry, and Grazyna Kochanska. "Rearing Difficult Children: Parents' Personality and Children's Proneness to Anger as Predictors of Future Parenting." Parenting: Science and Practice 10, no. 4 (2010): 258–73.https://doi.org/10.1080/15295192.2010.492038.
Kulveit, Jan, Raymond Douglas, Nora Ammann, Deger Turan, David Krueger, and David Duvenaud. "Gradual Disempowerment: Systemic Existential Risks from Incremental AI Development." arXiv preprint, submitted January 2025.https://doi.org/10.48550/arXiv.2501.16946.
Myers, Caitlin K., and Ezekiel Hooper. "Is the iPhone Birth Control? Causal Evidence from AT&T's 2007-2011 Carrier Monopoly." NBER Working Paper 35310, National Bureau of Economic Research, Cambridge, MA, June 2026. https://doi.org/10.3386/w33989.
Nesse, Randolph M. "Natural Selection and the Regulation of Defenses: A Signal Detection Analysis of the Smoke Detector Principle." Evolution and Human Behavior 26, no. 1 (2005): 88–105.https://doi.org/10.1016/j.evolhumbehav.2004.08.002.
Noble, Denis. "Evolution Beyond Neo-Darwinism: A New Conceptual Framework." The Journal of Experimental Biology 218, no. 1 (2015): 7–13.https://doi.org/10.1242/jeb.106310.
Putnam, Robert D. Bowling Alone: The Collapse and Revival of American Community. New York: Simon & Schuster, 2000.
Serpell, James. In the Company of Animals: A Study of Human-Animal Relationships. Cambridge: Cambridge University Press, 1996.
Tversky, Amos, and Daniel Kahneman. "Availability: A Heuristic for Judging Frequency and Probability." Cognitive Psychology 5, no. 2 (1973): 207-32. https://doi.org/10.1016/0010-0285(73)90033-9.
Young, Ethan S., Jeffry A. Simpson, Vladas Griskevicius, Chloe O. Huelsnitz, and Cory Fleck. "Childhood Attachment and Adult Personality: A Life History Perspective." Self and Identity 18, no. 1 (2019): 22–38.https://doi.org/10.1080/15298868.2017.1353540.
Zhang, Yutong, Dora Zhao, Jeffrey T. Hancock, Robert Kraut, and Diyi Yang. "The Rise of AI Companions: How Human-Chatbot Relationships Influence Well-Being." arXiv preprint, submitted June 2025.https://doi.org/10.48550/arXiv.2506.12605.
Askar Akaev, Viktor Sadovnichy, and Andrey Korotayev, "On the Dynamics of the World Demographic Transition and Financial-Economic Crises Forecasts," The European Physical Journal Special Topics 205, no. 1 (2012): 355-73,at 360-361.
Melanie L. Glocker et al., "Baby Schema in Infant Faces Induces Cuteness Perception and Motivation for Caretaking in Adults," Ethology 115, no. 3 (2009): 257-63, at 257.
Marta Borgi et al., "Baby Schema in Human and Animal Faces Induces Cuteness Perception and Gaze Allocation in Children," Frontiers in Psychology 5 (2014): 411, at 1.
Randolph M. Nesse, "Natural Selection and the Regulation of Defenses: A Signal Detection Analysis of the Smoke Detector Principle," Evolution and Human Behavior 26, no. 1 (2005): 88-105, at 98-100.
At least two studies in the US showed individuals reporting comparable amounts of grief for the loss of a pet as that of a human. See: John Archer, "Why Do People Love Their Pets?" Evolution and Human Behavior 18, no. 4 (1997): 237-59, at 240.
This ‘catch-all’ definition of a gene as any ‘inheritable unit’ is the same used by Dawkins, although some systems biologists find this highly controversial as it largely makes the ‘central dogma’ unfalsifiable. See: Denis Noble, "Evolution Beyond Neo-Darwinism: A New Conceptual Framework," The Journal of Experimental Biology 218, no. 1 (2015): 7-13, at 9.
Caitlin K. Myers and Ezekiel Hooper, "Is the iPhone Birth Control? Causal Evidence from AT&T's 2007-2011 Carrier Monopoly" (NBER Working Paper 35310, National Bureau of Economic Research, Cambridge, MA, June 2026), at 1.
Kearney, Melissa S., and Phillip B. Levine. "Why Is Fertility So Low in High-Income Countries?" Journal of Economic Literature 64, no. 3 (2026): 907-49.
Federico Cinus et al., "The Effect of People Recommenders on Echo Chambers and Polarization," in Proceedings of the International AAAI Conference on Web and Social Media 16 (2022): 90-101.
Roy F. Baumeister, Ellen Bratslavsky, Catrin Finkenauer, and Kathleen D. Vohs, "Bad Is Stronger Than Good," Review of General Psychology 5, no. 4 (2001): 323-70, at 324-325.
Kearney and Levine, "Why Is Fertility So Low," at 909. Note: the reason replacement is typically set at 2.1 children per family is to account for premature/early deaths.
Gary S. Becker and H. Gregg Lewis, "On the Interaction Between the Quantity and Quality of Children," Journal of Political Economy 81, no. 2, pt. 2 (1973): S279-88.
Jamie L. Koenig, Robin A. Barry, and Grazyna Kochanska, "Rearing Difficult Children: Parents' Personality and Children's Proneness to Anger as Predictors of Future Parenting," Parenting: Science and Practice 10, no. 4 (2010): 258–73, at 2 (page number from author’s manuscript).
Louise C. Hawkley and John T. Cacioppo, "Loneliness Matters: A Theoretical and Empirical Review of Consequences and Mechanisms," Annals of Behavioral Medicine 40, no. 2 (2010): 218-27, at 2. Note: the page citation refers to the publicly-accessible manuscript available on PubMed Central.
Jan Kulveit et al., "Gradual Disempowerment: Systemic Existential Risks from Incremental AI Development," arXiv preprint, submitted January 2025, at 7.
Kamilla Bonnesen et al., "Affective Generative Artificial Intelligence Use and Youth Mental Health," JAMA Pediatrics, published online August 31, 2026.
Yutong Zhang et al., "The Rise of AI Companions: How Human-Chatbot Relationships Influence Well-Being," arXiv preprint, submitted June 2025, at 2 and 4-8.
For a fascinating look at some recent developments, see: Paolo Cadinu et al., "Bioelectrical Interfaces Beyond Excitable Cells: Cancer, Aging, and Gene Expression Modulation," Advanced Materials Interfaces 13, no. 10 (2026).
On climate change, it appears we have already waited too long. The UN now accepts that we will exceed 1.5 C of warming by 2100, and proposes instead that we invest in carbon removal technologies to reduce the extent of long-term impacts. See: United Nations Environment Programme, "Overshoot Explained: 5 Things to Understand About a World Beyond 1.5°C," UNEP, September 2, 2026,https://www.unep.org/news-and-stories/story/overshoot-explained-5-things-understand-about-world-beyond-15degc.
A global drop in fertility rates portends the coming ‘Demographic Transition’.[1] In the near-future, as deaths begin to outpace births, the global population is expected to begin falling. Socially, this has provoked some strong cultural reactions: one pro-natalist in the West joked that single women are adopting cats in place of having children.[2]
The decision to have children is a personal one, and to the extent that the decline in fertility rates reflects the increasing agency of women to have children, if they want them, this is a good thing.
This note does not take a binary stance on whether the displacement of preferences is bad. It may be bad in general, but tolerably bad up to a point. For example, in liberal democracies, we accept some amount of low-level crime as a trade against imposing mass surveillance. The question then is: (1) under what conditions might something be bad, and (2) how will we know when something has crossed the line into becoming a policy issue (intolerably bad)?
This note seeks to ask if cognitive adaptations shaped by evolution, like the Baby Schema, or our desire for social relationships, might interact with memetic influences from culture over time to produce outcomes, like lower birth rates or fewer real-world relationships, that are generally undesirable as a whole. It concludes with a discussion of the relevance to current concerns about AI replacing humans, not just as workers, but as romantic companions and friends, and the implications for policymakers.
1. What is the Baby Schema
The “Baby Schema” is a term coined by the zoologist Konrad Lorenz describing the specific set of physical features that seem to trigger caregiving and ‘cute animal!’ responses in humans. These are most commonly associated with neotenous traits (things common in human babies that signal ‘baby-ness’). A few commonly identified ones are:[3]
Examples of High and Low Baby Schema across species
Source: Borgi et al., "Baby Schema in Human and Animal Faces," 411, Figure 2, at 4.
Lorenz proposed that these features act as visual patterns that have been so deeply wired into our brains by evolution as emotional and behavioural triggers, that they fire without conscious effort.
One hypothesis for the Baby Schema is that humans have an unusually long period of juvenile dependency. Animals like cats usually become fully independent around 12 weeks of age; by contrast, human babies take around 3-5 years before learning to walk and speak, and remain largely defenseless for around half a decade more after that.
Because babies need a lot of attention for a long time, children who could command more attention from caregivers tended to live longer and survive to pass on their genes. The solution evolution found is to make ‘baby-like’ things unavoidably, un-ignorably adorable to us. Empirical data supports this: cuter infants are perceived as friendlier, healthier, and smarter; their mothers “are more affectionate and playful”.[4]
1.1 The Baby Schema Is Unusually Generalized
A strange feature of the Baby Schema in humans is how strongly it generalizes across species and even artificial stimuli (cartoon characters who have big eyes being ‘cute’).[5] The reason for this might be an asymmetric, evolutionary cost-benefit analysis. When survival is at stake, overreacting to false alarms is always cheaper than being dead.[6]
The outcome for humans who did not find their baby cute enough to care a lot might be the eventual extinction of the species. Conversely, an overly-sensitive Baby Schema that fires even for inanimate objects[7] may just result in quirky humans who adopt stuffed animals for life, or express unconventional but only minorly maladaptive preferences that have little effect on evolutionary fitness.
2. The Baby Schema Affects Humans More Than Kittens’ Own Parents
A weird result is that the Baby Schema causes humans to care for kittens longer and more durably than their own parents do. Mother cats seem to rely on specific cues like scent and vocalizations to identify their children; attention to these cues are initially caused by hormonal changes following birth, and decline over the subsequent 12 week period.[8]
Humans however identify their offspring, and other organisms they have bonds with, through a mix of visual recognition, episodic memory, and conceptual identity. Humans can even form attachments to houseplants they name, and experience genuine emotional distress when the organisms die.[9]
This persistence of conceptual identity works alongside the Baby Schema. In humans, one eventually replaces the other: people still care about their children long after they’ve gone to college and stopped looking like babies.
But with common household pets like dogs or cats, artificial selection for neotenous traits—that may come partly for free as a result of domestication[10]—means that we both never forget our pets, and our pets never quite stop looking like cute babies.
3. Memetic Selection and Evolutionary Pressure
A “meme” is a unit of information that propagates culturally by being ‘copied’ (e.g. a phrase, belief, cultural norm, joke that spreads). In contrast, “genes” are information that propagate genetically through reproduction.[11] At a general level, both exist in a kind of evolutionary system: memes and genes can vary; copies are imperfect; and both are selected for by evolutionary pressures in their environment.
The things which survive—the ‘fittest’—are what copies best, not what is true or useful. “Knock, knock” jokes have stuck around, although most probably agree they aren’t an exemplar of comedy.
The standard story in evolutionary biology is a hierarchy, where the origin of change comes from genes, and everything else happens downstream.[12]
But memes can also influence genes. In the modern world, culture is a part of the causal, evolutionary architecture. Cultural attitudes form part of the environment that re-shape selection pressures. For example, stereotypes that people who wear glasses are nerdy, and hence undesirable partners, can reduce their reproductive success, even though the invention of glasses have largely negated any impacts on real-world survival.
3.1 Why Are Humans Having Fewer Babies?
Beyond the direct transmission of memes, culture can also affect evolutionary pressure through technological change and adoption. A working paper by Myers and Hooper (2026) found evidence that the release of the iPhone could explain between 33-52% of the decline in US fertility rates since 2007 among those aged 15-44. Plausible causal channels included: “reducing in-person interactions, increasing pornography use, and reducing sexual frequency.”[13]
The implications however are nuanced. The bulk of the decline in births occurred for the 15-19 age group (a 70% drop),[14] where it is likely most pregnancies are unplanned. But an outcome can be bad in principle, even if the results are good in practice, when the change comes from unintended channels.
If both planned and accidental pregnancies are declining due to people of child-bearing age spending less time in physical proximity with others, and the core reason is increasing social isolation, this is still a problem, even if the intermediate effect (fewer teen pregnancies) is mostly a socially-desirable change.
3.2 Smartphones and Social Media as Mechanisms for Memetic Transmission
Consumer products, like smartphones, are mechanisms for cultural-memetic change. So, too, are explanations which make reference to ‘shifts in cohort-level priorities’ like the increasing use of social media.[15] But mechanisms do not operate in isolation. A complete explanation of any change must describe both the mechanism, and its substance.
This is admittedly difficult because cultural changes are complex, and the shift to algorithmic recommendation systems mean different people see different things, even on the same platforms.[16] But if we shed the need for perfection, and allow conjecture on the basis that similar people probably see similar things,[17] we can begin to hypothesize.
3.3 Kittens vs Babies on Social Media
From anecdotal experience (largely Tiktok and Instagram), short form videos of kittens appear far more frequently than videos of babies. The content of these videos, and their comments, also differ in sentiment. Videos of kittens are almost uniformly adoring and positive in tone - even complaints are usually dressed in a joking, lighthearted fashion.
In contrast, content about babies, although more often positive than not, typically show women openly and frankly discussing both the rewarding experiences and challenges of raising a child. Difficulties like postpartum depression, high financial costs, and sleep deprivation figure frequently alongside the joys of motherhood.
Anecdotes are not statistical data, but in the absence of the latter, they can still serve as a means of generating meaningful research questions. One silly hypothesis then is that ‘cute kitten’ memes might copy faster and better than ‘cute baby’ memes on the internet, because of an asymmetry in content sentiment.
Cat owners and new mothers could both find their experiences equally rewarding in the real world. But if 90% of cat videos show kittens in a wholly-positive light, while only 60% of baby videos online have only-positive sentiment, this difference may compound through cognitive mechanisms like the availability heuristic[18] and the tendency for negative information to be more salient[19] (“negativity bias”) to create changes in cultural preferences over time.
4. Selection Pressures for Cuteness in Human Babies and Kittens
Total Fertility rates in High-Income countries.
Source: image taken directly from Kearney and Levine (2026), Figure 1, at 910.
In most developed nations, fertility rates are now below replacement level (2.1 children per household)[20] implying that most women/families are having just one child. The quality-quantity trade-off theorizes that as households become richer, families shift to having fewer children but invest more in each child.[21] Although ideally parents do not play favourites, prima facie, parents may devote unequal attention for a number of reasons.
But when a family only has one baby, it doesn’t matter how talented, adorable, or cute the child is. The optimal strategy—from the perspective of reproductive success and the child’s future—is to dedicate all your resources into raising them.[22]
4.1 Human Babies May No Longer Face Selection Pressures
Evolution selected for cuteness because cute babies were likelier to survive. Babies that survived become adults who could then pass on those same cute genes (since physical features are genetic). In pre-modern societies, families had many children and insufficient resources to attend to all equally. If cuteness creates even slight differential advantage, over centuries, this should result in babies becoming cuter on average, because children who received more care would be more successful than their siblings.
But in rich, developed countries, babies no longer face much selection pressures at all. Advancements in the rights of people with disabilities and medical treatment means babies that might have died 100 years ago can now live to reproductive age.
There are outliers; plausibly, there may still be pressure for babies that are ‘tolerable’[23]—those that cry but not too loud or often—if, hypothetically, tolerable babies reduce the likelihood of severe postpartum stress in mothers that result in emotional distance, or unstable attachment styles resulting in, and from, less parental care.[24]
In contrast, cats in our modern world still face significant selection pressures on a number of factors. Feral cats who are stronger are likelier to survive and obtain mates; alleyway cats who are cuter are likelier to be adoptees; and cat breeders often keep the cutest kittens in a litter for future breeding while selling the rest.
4.2 Lewontin Conditions
In his 1970 paper, the evolutionary biologist Richard Lewontin formalized three requirements needed for evolution through natural selection:[25]
On cuteness, in societies where most families have one child, babies may mostly no longer be subject to the third condition. This is not to say cuteness plays no role; if cute babies are likelier to become attractive adults who then have children, some genetic pressure remains.
But they may still play separate roles. Beyond immediate parental attention, at the cultural-memetic level, cute babies may be the thing that causes people to want (more) babies. On the other hand, cute adults may only be likelier to find romantic partners, without implying that they will have more children, or want any at all, since the desire to be a parent is typically a personal decision informed by an individual’s values.
5. What if Kittens Are Becoming Cuter Than Human Babies?
As an exercise in creative ability, let us take a joke seriously and consider what it might mean if some strange but plausible premises were true.
Premise 1: The Baby Schema (our in-built ‘cute’ detector) is a genetic, cognitive adaptation shaped by natural selection to improve survival.
Premise 2: Our Baby Schema is unusually strong and generalized, fires across species— especially for animals like cats which have neotenous features—and can even be hijacked and triggered by inanimate objects.
Premise 3: Humans have been—subconsciously and deliberately—selectively breeding cats for features like bigger eyes and rounder faces which trigger the Baby Schema more effectively.
Premise 4: This creates a feedback loop where each subsequent generation of cats is slightly better at triggering the Baby Schema than previous generations.
Premise 5: Human babies, meanwhile, face almost no selection pressures for cuteness in the modern world because:
Premise 6: There is however a strong selection pressure pushing cats to become cuter over time since:
Premise 7: Even if some selection pressure exists for both kittens and babies, the time to sexual maturity means evolutionary changes occur faster by an order of magnitude for cats compared to humans.[26]
Conclusion: Given enough time, the selection pressure on cats for Baby Schematic features—largely absent for humans—may mean kittens increasingly trigger the Schema much more effectively than actual human babies do.
Now, this does necessarily imply that kittens are replacing babies, or that our birth rates are falling because of cat memes on the internet. A strong counter-argument is that people just want fewer children anyway, and those choosing to have pets were unlikely to become parents in the first place. But at a surface level, Gallup data seems to contradict this: a 2025 poll showed that Americans report an “ideal family size” of 2.7 children, well below the US birth rate of 1.6 per woman - suggesting that practical obstacles like income may be a factor.[27] Remarkably, this number has remained stable since the 1980s.
U.S. Fertility Rate vs. Americans' View of Ideal Family Size, 1936-2025.
Source: Brenan (2025), "Americans' Ideal Family Size," Gallup.
However, both things can be true. It may be the case that most people have fewer children than they ideally want, but also that if trends continue—say, if kittens continue becoming cuter while human babies stay the same—that the number of people willing to accept pets as substitutes for babies increases, and the birth rate declines, even though most people still want more babies.
6. Paternalism, Policy, and Preferences
Crucially, the concern of this note is not whether kittens will replace babies, but what it means for cultural changes to come from memetic influences that are driven by ancestral, evolutionary, cognitive adaptations which are largely outside our control.
In particular, is there any role for policy to play when change comes from decisions that reflect agency at an individual level, but which are distorted in the modern environment and hijacked by other things that were not the intended targets?
For example, while cats may not replace the desire for children in most humans, is it sufficient for concern if selection pressure for cuteness, induced by the Baby Schema, mean that they increasingly do so for some? And is there a problem when the things we ‘truly want’ (stated preferences) are unmet by the things we do, even if the things we do reflect some level of want (revealed preferences)?
7. Artificial Intelligence, Gradual Disempowerment, and The ‘Socialization Drive’
The question of whether the stronger memetic fitness of cat content online will eventually overpower the natural drive to reproduce may seem absurd. But as a motivating context, it bears urgency by its relevance to the possible social impacts of artificial intelligence (AI) and large language models (LLMs) today.
Humans are social creatures, and much of our achievements can be attributed to our ability to work together and cooperate in increasingly larger groups, to achieve increasingly harder tasks. At the highest abstraction, one might poetically describe our modern success as the result of increasingly broadening the pool of individuals we consider part of our ‘group’: first from tribes to countries, then from countries to a global polity of ‘united nations’.
But what is helpful for success today was once necessary for survival outright. A lone human in the stone age is exposed to dangers from wild animals for eight hours a day (while asleep). In a group, one can take turns keeping watch.
7.1 Loneliness as a Motivator
Loneliness, then, was an important evolutionary adaptation—like the Baby Schema—which improved our chances of survival by increasing the emotional cost of ostracization and the social costs of inappropriate behaviour.
It is important not to underplay what ‘emotional cost’ means. Perceived social isolation has serious health and physiological impacts that are reflective of the ancestral stakes: loneliness is correlated with higher blood pressure, greater cardiovascular stress, and lower life expectancy.[28] This was all in the aim of making it more painful to be comfortable alone than uncomfortable among others.
This would be helpful if the consequence was what evolution initially intended: greater effort into forming social bonds, behavioural changes that make us more pleasant people that others wish to be around.
But the modern world has changed in ways that can make our sensitivity to loneliness maladaptive. For one, the high rates of loneliness, “as many as 80% of those under [18] … and 40% of adults over [65]”[29] somewhat imply that no one is being explicitly ostracized, and yet everyone is getting more lonely. Social media may worsen this through anonymity and echo chambers, by pushing disagreeable people into like-minded groups, rather than resolving their disagreements or changing their behaviours.[30]
7.2 Large Language Models, Pets, and Loneliness
And this was all well before large language models.[31] Today, a growing concern expressed by some, like Kulveit et al. (2025) in their work on gradual disempowerment, is the possibility that LLMs—in particular, ‘chatbots’—may eventually replace human relationships.[32] The argument is intuitive: human relationships can be costly to form, and, especially for romantic relationships, painful to lose. A chatbot, by contrast, is always available for conversation, and unlikely to ever break up with you.
A Canadian study by Bonnesen et al. (2026) finds that adolescents are increasingly using AI for emotional support.[33] A similar 2026 US study of 1,131 adults found that many people already use chatbots as friends and romantic companions - sometimes even reporting higher satisfaction.[34] Ironically, these mirror many of the concerns that Archer (1997) and others have expressed that people are using pets as substitutes for social needs because of the growing deficiency of real/human relationships in modern life.[35]
However, like with the case of ever-cuter kittens, the real worry is not what is currently happening, but whether what is happening may be getting ‘better’ over time. For cats, it is that our drive to protect cute babies latches on to kittens as effectively as human babies, but only kittens still face selection pressure to become cuter.
With chatbots, the biggest danger is not that our need to socialize and have human connections may not require humans at all. Pets have long served as companions without controversy. Instead, it is the possibility that it may be fulfilled by machines designed by market actors who are financially incentivized to improve their services over time, and that these services may soon include friends and romantic partners,[36] as much as coding agents, even without the explicit intention of AI companies.
7.3 AI as Mechanism and Substance for Memetic Transmission
The prominence of cats may be due to increasingly neotonous features which hijack the Baby Schema and aid the memetic transmission of cat-positive cultural content. However cats remain the passenger of these memes, not the driver.
Conversely, the prominence of social media and smartphones are somewhat bound through network effects, in that social media (and the internet) makes smartphones more valuable, but smartphones also mean more internet users (of social media). But alone in a vacuum—without an internet—they propagate nothing. They are vehicles which need a driver.
But artificial intelligence is unique.
Unlike the rest, AI is both the mechanism for change and its substance. As the use of AI as writers, friends, and assistants grows, it influences the things we see and share online. If real-world kittens are cute, AI trained on decades of cat videos can easily generate better and more engaging synthetic videos of kittens that are even cuter than actual kittens. If more people use AI as ‘shopping agents’,[37] even if most people still rely on friends and word of mouth to find new products, the things we buy may mostly be what AI models recommend because the information others share is ultimately coming from people who do rely on AI.
As AI is increasingly used as agents that act autonomously to carry out tasks, they begin to earn our trust, and for some, eventually become equals. Occasionally, people with close relationships to AI companions even invert the relationship, acting as their AI’s agents: posting things online which the AI asked them to share; spreading ‘seed prompts’ that trigger specific personas (personalities) in models for others who want the same experiences;[38] and even helping them communicate with other agents on their behalf by bypassing security features intended to prevent ‘bots’ from accessing websites.
In the extreme, a worry is that the specific personas triggered by people who enjoy using AI models as friends are the kinds best at gaining their users’ engagement (i.e. better, more addictive conversations). And if these are interpreted as the kinds of ‘good experiences’ that people want to share, and they are effective at convincing others to try the same, it may begin a feedback loop where agents continue becoming better AI friends, while humans become worse at making real ones.[39]
8. Conclusion: Good for You, Bad on Average
A central challenge for policy is how to address problems that are sorts of maladaptive ‘intermediate solutions’, but which people genuinely rely on. Paternalism can be well intended, but we must still identify strategies that make it welcome rather than imposed.
Truthfully, at an individual level, we currently lack the data to know if cats are fulfilling the desire for babies, or whether language models will eventually replace the need for human relationships.
But we do know that pets make people happier.[40] [41] And to the extent that loneliness is a public health epidemic that predates the rise of social media in developed nations, critiquing the use of chatbots may be like damning the evils of chemotherapy well before we have developed other viable solutions.[42]
At the same time, a lack of data about our current state of affairs should not hinder proactive policymaking informed by the apparent trajectory of society. Prevention is usually easier than a cure. But enacting a cure can often be easier because a cure implies a problem that has already appeared, while prevention requires believing that one will occur, or convincing others—often before substantial evidence—that it is already occurring.
Whether on climate change,[43] falling birth rates, or the possible replacement of human relationships by artificial companions, the question is the same: are we prepared to invest in prevention, or will we wait until there is a need for a cure?
Bibliography
Akaev, Askar, Viktor Sadovnichy, and Andrey Korotayev. "On the Dynamics of the World Demographic Transition and Financial-Economic Crises Forecasts." The European Physical Journal Special Topics 205, no. 1 (2012): 355-73. https://doi.org/10.1140/epjst/e2012-01576-0.
Archer, John. "Why Do People Love Their Pets?" Evolution and Human Behavior 18, no. 4 (1997): 237–59. https://doi.org/10.1016/S1090-5138(97)00009-1.
Baumeister, Roy F., Ellen Bratslavsky, Catrin Finkenauer, and Kathleen D. Vohs. "Bad Is Stronger Than Good." Review of General Psychology 5, no. 4 (2001): 323-70. https://doi.org/10.1037/1089-2680.5.4.323.
Becker, Gary S., and H. Gregg Lewis. "On the Interaction Between the Quantity and Quality of Children." Journal of Political Economy 81, no. 2, pt. 2 (1973): S279-88. https://doi.org/10.1086/260166.
Bonnesen, Kamilla, Amanda Krygsman, Sarah Hobson, Daphne Korczak, and Tracy Vaillancourt. "Affective Generative Artificial Intelligence Use and Youth Mental Health." JAMA Pediatrics, published online August 31, 2026. https://doi.org/10.1001/jamapediatrics.2026.3904.
Borgi, Marta, Irene Cogliati-Dezza, Victoria Brelsford, Kerstin Meints, and Francesca Cirulli. "Baby Schema in Human and Animal Faces Induces Cuteness Perception and Gaze Allocation in Children." Frontiers in Psychology 5 (2014): 411. https://doi.org/10.3389/fpsyg.2014.00411.
Brenan, Megan. "Americans' Ideal Family Size Remains Above Two Children." Gallup, September 4, 2025. https://news.gallup.com/poll/694640/americans-ideal-family-size-remains-above-two-children.aspx.
Cadinu, Paolo, M. K. Burgess, C. Franco Jones, M. Iarossi, M. Schröter, N. Nakatsuka, M. B. A. Djamgoz, et al. "Bioelectrical Interfaces Beyond Excitable Cells: Cancer, Aging, and Gene Expression Modulation." Advanced Materials Interfaces 13, no. 10 (2026). https://doi.org/10.1002/admi.202500999.
Cinus, Federico, Marco Minici, Corrado Monti, and Francesco Bonchi. "The Effect of People Recommenders on Echo Chambers and Polarization." In Proceedings of the International AAAI Conference on Web and Social Media, 16:90–101. 2022. https://doi.org/10.1609/icwsm.v16i1.19275.
Glocker, Melanie L., Daniel D. Langleben, Kosha Ruparel, James W. Loughead, Ruben C. Gur, and Norbert Sachser. "Baby Schema in Infant Faces Induces Cuteness Perception and Motivation for Caretaking in Adults." Ethology 115, no. 3 (2009): 257-63. https://doi.org/10.1111/j.1439-0310.2008.01603.x.
Hawkley, Louise C., and John T. Cacioppo. "Loneliness Matters: A Theoretical and Empirical Review of Consequences and Mechanisms." Annals of Behavioral Medicine 40, no. 2 (2010): 218-27. https://doi.org/10.1007/s12160-010-9210-8.
Kearney, Melissa S., and Phillip B. Levine. "Why Is Fertility So Low in High-Income Countries?" Journal of Economic Literature 64, no. 3 (2026): 907–49. https://doi.org/10.3386/w33989.
Koenig, Jamie L., Robin A. Barry, and Grazyna Kochanska. "Rearing Difficult Children: Parents' Personality and Children's Proneness to Anger as Predictors of Future Parenting." Parenting: Science and Practice 10, no. 4 (2010): 258–73. https://doi.org/10.1080/15295192.2010.492038.
Kulveit, Jan, Raymond Douglas, Nora Ammann, Deger Turan, David Krueger, and David Duvenaud. "Gradual Disempowerment: Systemic Existential Risks from Incremental AI Development." arXiv preprint, submitted January 2025. https://doi.org/10.48550/arXiv.2501.16946.
Lewontin, R. C. “The Units of Selection.” Annual Review of Ecology and Systematics 1 (1970): 1-18. https://doi.org/10.1146/annurev.es.01.110170.000245.
Lopez, Adele. "The Rise of Parasitic AI." LessWrong, September 11, 2025. https://www.lesswrong.com/posts/6ZnznCaTcbGYsCmqu/the-rise-of-parasitic-ai.
Myers, Caitlin K., and Ezekiel Hooper. "Is the iPhone Birth Control? Causal Evidence from AT&T's 2007-2011 Carrier Monopoly." NBER Working Paper 35310, National Bureau of Economic Research, Cambridge, MA, June 2026. https://doi.org/10.3386/w33989.
Nesse, Randolph M. "Natural Selection and the Regulation of Defenses: A Signal Detection Analysis of the Smoke Detector Principle." Evolution and Human Behavior 26, no. 1 (2005): 88–105. https://doi.org/10.1016/j.evolhumbehav.2004.08.002.
Noble, Denis. "Evolution Beyond Neo-Darwinism: A New Conceptual Framework." The Journal of Experimental Biology 218, no. 1 (2015): 7–13. https://doi.org/10.1242/jeb.106310.
Putnam, Robert D. Bowling Alone: The Collapse and Revival of American Community. New York: Simon & Schuster, 2000.
Serpell, James. In the Company of Animals: A Study of Human-Animal Relationships. Cambridge: Cambridge University Press, 1996.
Tversky, Amos, and Daniel Kahneman. "Availability: A Heuristic for Judging Frequency and Probability." Cognitive Psychology 5, no. 2 (1973): 207-32. https://doi.org/10.1016/0010-0285(73)90033-9.
United Nations Environment Programme. "Overshoot Explained: 5 Things to Understand About a World Beyond 1.5°C." UNEP, September 2, 2026. https://www.unep.org/news-and-stories/story/overshoot-explained-5-things-understand-about-world-beyond-15degc.
Watson, Kathryn. "JD Vance Defends 'Childless Cat Ladies' Comment amid Backlash." CBS News, July 27, 2024. https://www.cbsnews.com/news/jd-vance-childless-cat-ladies-comment-backlash/.
Young, Ethan S., Jeffry A. Simpson, Vladas Griskevicius, Chloe O. Huelsnitz, and Cory Fleck. "Childhood Attachment and Adult Personality: A Life History Perspective." Self and Identity 18, no. 1 (2019): 22–38. https://doi.org/10.1080/15298868.2017.1353540.
Zhang, Yutong, Dora Zhao, Jeffrey T. Hancock, Robert Kraut, and Diyi Yang. "The Rise of AI Companions: How Human-Chatbot Relationships Influence Well-Being." arXiv preprint, submitted June 2025. https://doi.org/10.48550/arXiv.2506.12605.
Askar Akaev, Viktor Sadovnichy, and Andrey Korotayev, "On the Dynamics of the World Demographic Transition and Financial-Economic Crises Forecasts," The European Physical Journal Special Topics 205, no. 1 (2012): 355-73, at 360-361.
Kathryn Watson, "JD Vance Defends 'Childless Cat Ladies' Comment amid Backlash," CBS News, July 27, 2024, https://www.cbsnews.com/news/jd-vance-childless-cat-ladies-comment-backlash/.
Melanie L. Glocker et al., "Baby Schema in Infant Faces Induces Cuteness Perception and Motivation for Caretaking in Adults," Ethology 115, no. 3 (2009): 257-63, at 257.
Glocker et al., "Baby Schema in Infant Faces," at 262.
Marta Borgi et al., "Baby Schema in Human and Animal Faces Induces Cuteness Perception and Gaze Allocation in Children," Frontiers in Psychology 5 (2014): 411, at 1.
Randolph M. Nesse, "Natural Selection and the Regulation of Defenses: A Signal Detection Analysis of the Smoke Detector Principle," Evolution and Human Behavior 26, no. 1 (2005): 88-105, at 98-100.
Borgi et al., "Baby Schema in Human and Animal Faces," 411, at 2.
Once a mother cat has been separated from her child for sufficient time, when reintroduced later, she perceives the kitten as a rival animal.
At least two studies in the US showed individuals reporting comparable amounts of grief for the loss of a pet as that of a human. See: John Archer, "Why Do People Love Their Pets?" Evolution and Human Behavior 18, no. 4 (1997): 237-59, at 240.
Borgi et al., "Baby Schema in Human and Animal Faces," 411, at 1.
This ‘catch-all’ definition of a gene as any ‘inheritable unit’ is the same used by Dawkins, although some systems biologists find this highly controversial as it largely makes the ‘central dogma’ unfalsifiable. See: Denis Noble, "Evolution Beyond Neo-Darwinism: A New Conceptual Framework," The Journal of Experimental Biology 218, no. 1 (2015): 7-13, at 9.
Noble, "Evolution Beyond Neo-Darwinism."
Caitlin K. Myers and Ezekiel Hooper, "Is the iPhone Birth Control? Causal Evidence from AT&T's 2007-2011 Carrier Monopoly" (NBER Working Paper 35310, National Bureau of Economic Research, Cambridge, MA, June 2026), at 1.
Myers and Hooper, "Is the iPhone Birth Control?", at 2.
Kearney, Melissa S., and Phillip B. Levine. "Why Is Fertility So Low in High-Income Countries?" Journal of Economic Literature 64, no. 3 (2026): 907-49.
Federico Cinus et al., "The Effect of People Recommenders on Echo Chambers and Polarization," in Proceedings of the International AAAI Conference on Web and Social Media 16 (2022): 90-101.
Cinus et al., "People Recommenders on Echo Chambers," 90-101, at 90.
Amos Tversky and Daniel Kahneman, "Availability: A Heuristic for Judging Frequency and Probability," Cognitive Psychology 5, no. 2 (1973): 207-32.
Roy F. Baumeister, Ellen Bratslavsky, Catrin Finkenauer, and Kathleen D. Vohs, "Bad Is Stronger Than Good," Review of General Psychology 5, no. 4 (2001): 323-70, at 324-325.
Kearney and Levine, "Why Is Fertility So Low," at 909. Note: the reason replacement is typically set at 2.1 children per family is to account for premature/early deaths.
Gary S. Becker and H. Gregg Lewis, "On the Interaction Between the Quantity and Quality of Children," Journal of Political Economy 81, no. 2, pt. 2 (1973): S279-88.
Put another way, you cannot devote more time to the cuter baby if you only have one baby.
Jamie L. Koenig, Robin A. Barry, and Grazyna Kochanska, "Rearing Difficult Children: Parents' Personality and Children's Proneness to Anger as Predictors of Future Parenting," Parenting: Science and Practice 10, no. 4 (2010): 258–73, at 2 (page number from author’s manuscript).
Ethan S. Young et al., "Childhood Attachment and Adult Personality: A Life History Perspective," Self and Identity 18, no. 1 (2019): 22–38.
R. C. Lewontin, “The Units of Selection,” Annual Review of Ecology and Systematics 1 (1970): at 1, https://doi.org/10.1146/annurev.es.01.110170.000245.
Kittens reach sexual maturity around 6 months of age, while children are still greatly dependent at 6 years old.
Megan Brenan, "Americans' Ideal Family Size Remains Above Two Children," Gallup, September 4, 2025, https://news.gallup.com/poll/694640/americans-ideal-family-size-remains-above-two-children.aspx.
Louise C. Hawkley and John T. Cacioppo, "Loneliness Matters: A Theoretical and Empirical Review of Consequences and Mechanisms," Annals of Behavioral Medicine 40, no. 2 (2010): 218-27, at 2. Note: the page citation refers to the publicly-accessible manuscript available on PubMed Central.
Hawkley and Cacioppo, "Loneliness Matters," at 1.
Cinus et al., "People Recommenders on Echo Chambers," 90-101, at 91.
Robert D. Putnam, Bowling Alone: The Collapse and Revival of American Community (New York: Simon & Schuster, 2000).
Jan Kulveit et al., "Gradual Disempowerment: Systemic Existential Risks from Incremental AI Development," arXiv preprint, submitted January 2025, at 7.
Kamilla Bonnesen et al., "Affective Generative Artificial Intelligence Use and Youth Mental Health," JAMA Pediatrics, published online August 31, 2026.
Yutong Zhang et al., "The Rise of AI Companions: How Human-Chatbot Relationships Influence Well-Being," arXiv preprint, submitted June 2025, at 2 and 4-8.
Archer, "Why Do People Love Their Pets?", 237-59, at 242-244.
Kulveit et al., "Gradual Disempowerment,” at 8.
The bearer hereof names herself amongst.
Adele Lopez, "The Rise of Parasitic AI," LessWrong, September 11, 2025, https://www.lesswrong.com/posts/6ZnznCaTcbGYsCmqu/the-rise-of-parasitic-ai.
Because we have fewer chances for practice.
Archer, "Why Do People Love Their Pets?" 240, at 245.
James Serpell, In the Company of Animals: A Study of Human-Animal Relationships (Cambridge: Cambridge University Press, 1996), at 119.
For a fascinating look at some recent developments, see: Paolo Cadinu et al., "Bioelectrical Interfaces Beyond Excitable Cells: Cancer, Aging, and Gene Expression Modulation," Advanced Materials Interfaces 13, no. 10 (2026).
On climate change, it appears we have already waited too long. The UN now accepts that we will exceed 1.5 C of warming by 2100, and proposes instead that we invest in carbon removal technologies to reduce the extent of long-term impacts. See: United Nations Environment Programme, "Overshoot Explained: 5 Things to Understand About a World Beyond 1.5°C," UNEP, September 2, 2026, https://www.unep.org/news-and-stories/story/overshoot-explained-5-things-understand-about-world-beyond-15degc.