Art & Evolution
The 73,000-Year Habit.
Why a Species Under Pressure Kept Making Art.
Someone sat in a cave on the southern coast of Africa and drew nine lines on a piece of stone. They were not decorating anything. They were not signalling to a predator. Whatever they were doing, we have not stopped doing it since, and nobody can agree on why we started.
For as long as I can remember, I have drawn things that do not exist. Invented worlds, other realms, technical sketches of machines nobody has built, figures out of a daydream. Nobody asked me to. There was no audience. I was a child with paper and the compulsion arrived fully formed, which is a strange thing to say about a behaviour and a strange thing to notice about yourself.
In part one I found out what happens in the body during that. The reward circuits, the cortisol drop, the network that comes off suppression when a work stops being an object and starts being about you. Good. Mechanism established.
But mechanism is not explanation. Why does a body have that machinery at all?
Because here is what bothers me about it. Evolution is not sentimental. A species living at the edge of survival does not casually develop an elaborate reward response to marks on a wall. Every calorie spent grinding ochre is a calorie not spent on food, shelter or a child. Whatever art is, it was expensive, and it persisted anyway, across every human population that has ever been studied, for a length of time that is genuinely difficult to hold in your head.
So this post is the why. What the archaeological record actually contains, how far back it goes, and the four serious explanations researchers have offered for why any of it exists. One of those four starts somewhere you will not expect, and it is the one I have found the least discussed. I will get to it.
Nina, if you want one more personal beat before the science, this is where it would sit. Two prompts, pick either or neither:
1. A specific memory of making one of the invented worlds as a child. What it was, where you were, what you were avoiding or working out by drawing it.
2. The women's faces on your wall. When you started that series, what you were painting through at the time, and whether you knew what they were for when you began.
Delete this whole block if the intro works as it stands. It does work as it stands.
The Record
Every securely dated milestone in human mark-making. Drag back through it.
Dates are minimum ages from uranium-series or stratigraphic dating, which means each one is a floor, not a birthday. The behaviour is older than the oldest surviving example, always. [1][2][4][5][6]
Two things about that timeline need saying before we go near the theories, because most popular writing about cave art gets both of them wrong.
The first is that every date is a minimum. Uranium-series dating of the carbonate crust that formed over a painting tells you the art is at least that old. It cannot tell you how much older. So when a headline says the oldest art is 51,200 years old, the accurate sentence is that the oldest art we have found, that survived, in a cave that happened to be excavated, and that happened to have datable mineral growth over it, is at least 51,200 years old.
The second is preservation bias, and it is brutal. Pigment on a rock face inside a cave survives. Pigment on skin does not. Body paint, scarification, hairdressing, woven pattern, song, dance, story, arrangement of objects: none of it fossilises. The archaeological record of art is a record of the tiny fraction of human making that happened to be done on stone, inside shelter, in a climate that did not destroy it.
The oldest drawing we have is nine lines. Christopher Henshilwood and Francesco d'Errico, working at Blombos Cave on the South African coast, found a small silcrete flake in levels dated to around 73,000 years ago. On it, a crosshatch pattern drawn deliberately with an ochre crayon. They reconstructed the technique experimentally to confirm it was drawn rather than stained or scratched. It is abstract. It represents nothing. Whoever made it was doing something for which we have no functional account. [2]
Ornament is older than drawing by a very long way. At Bizmoune Cave in Morocco, Steven Kuhn and colleagues recovered perforated shell beads from levels dated to at least 142,000 years ago, and possibly 150,000. Beads are not decoration in the trivial sense. A bead is a portable signal, worn on a body, made to be seen by other people. Which makes personal ornament the earliest surviving evidence of humans deliberately manipulating how they appear to each other. [1]
The oldest known picture that tells a story is 51,200 years old and it is not in Europe. Adhi Agus Oktaviana and colleagues used laser-ablation uranium-series dating on a panel at Leang Karampuang on Sulawesi, showing a pig and three human-like figures interacting with it. A scene, not a catalogue of animals. Earlier work by Adam Brumm's group had dated a warty pig at nearby Leang Tedongnge to at least 45,500 years. Both predate anything comparable in Europe, which quietly demolishes a century of framing in which European cave art was the origin of the human symbolic mind. [4][5]
So the question is not whether art is old. The question is what it was for, and this is where the field stops agreeing.
Four Explanations
What each one claims, what supports it, what would kill it, and how it is actually doing.
Two of these come from books rather than papers, which is marked. Evolutionary explanations of behaviour are unusually easy to tell and unusually hard to test, so the status line on each card is the part worth reading.
Of those four, exactly one has generated a body of experimental evidence in living humans rather than arguments about the fossil record. It is the social bonding account, and its results are worth sitting with.
Performing music raises pain threshold. Listening to it does not. Robin Dunbar and colleagues at Oxford used pain threshold as a proxy for endorphin activity, because endorphins cannot be measured directly in an intact human brain without invasive procedures. Across several samples, actively performing music raised pain threshold and lifted positive affect. Passively listening did not produce the same effect. Their reading: the endorphin system responds to the doing, and endorphins are the same system that underlies social bonding in primates through grooming. [7]
Singing bonds strangers faster than other group activities. Eiluned Pearce, Jacques Launay and Dunbar followed adult education classes over seven months, comparing singing classes with creative writing and craft classes. Everyone bonded eventually. The singers bonded much faster. The authors called it the ice-breaker effect, and argued that music is an efficient mechanism for bonding groups too large for one-to-one grooming to hold together. [8]
The proposed mechanism is synchrony plus endorphins. Bronwyn Tarr, Launay and Dunbar reviewed the pathways: moving in time with other people blurs the perceptual boundary between self and other, and synchronised exertion releases endorphins. Together these produce the feeling of merging with a group that anyone who has been in a choir, a crowd or a dance floor will recognise. [9]
That is a real mechanism with real experiments behind it. Note what it explains well and what it does not. Music, dance, ritual, anything synchronous and performed: strong fit. A woman alone in a cave drawing crosshatch on a stone with nobody watching: no fit at all.
Every explanation on offer covers part of the behaviour and leaves the rest standing there unaccounted for. Which is usually a sign that the thing being explained is not one thing.
Ellen Dissanayake has spent four decades arguing that everyone is looking in the wrong place. Not at courtship, not at group ritual, not at leftover circuitry. At mothers and babies.
Her argument runs like this. What art does, formally, is take something ordinary and make it special. Dissanayake calls the behaviour artification: a set of operations applied to ordinary material to mark it as significant. Formalisation, repetition, exaggeration, elaboration, and manipulation of expectation. Those five operations appear in every art form in every culture she surveyed. [10]
And they are the same five operations a mother uses on an infant. In her chapter in The Origins of Music, Dissanayake laid out the parallel in detail. The exchange between a caregiver and a pre-verbal baby is rhythmic, multimodal, repetitive, exaggerated in pitch and expression, and structured around building and releasing expectation. It is universal, it is not taught, and it regulates the emotional state of both participants. She argues this interaction is the behavioural template that ritual, and then art, was built on. [11]
Whether or not she is right, notice the shape of the claim. Three of the four theories locate the origin of art in display, competition or spare cognitive capacity. The fourth locates it in the oldest and most universal caregiving relationship in the species, and treats emotional regulation rather than signalling as the core function.
Now count how often you have heard the first three, and how often you have heard the fourth.
Whose Hands Were On the Wall?
A claim, a method, and what happened when somebody checked. Step through it in order.
This is the clearest example I know of a finding that was reported everywhere in its strong form and corrected almost nowhere. [3][12][13]
One of these assumptions was asked for evidence. The other never was.
Track what just happened with the hand stencils. Dean Snow proposed that the makers were mostly women. That claim was tested by at least two independent groups, found to rest on a method that works at around 60 percent accuracy in modern hands and fails on male hands in some populations, and substantially corrected: at El Castillo, with a local Iberian reference sample, the split came out 11 female to 10 male. [3][12][13]
Good. That is science working. Claim, test, correction.
Now ask when the previous assumption was tested. For most of the twentieth century the standard account had Upper Palaeolithic cave art made by men: hunting magic, painted by hunters, for hunters. It appeared in textbooks, museum captions and documentaries. It was not a finding. Nobody had measured anything. It was an inference from the assumption that the important activities of prehistoric life were male activities, and it was accepted for decades without anyone requiring a discriminant analysis.
The same pattern shows up elsewhere in the field. In 2020, Randall Haas and colleagues reported a 9,000-year-old burial in the Peruvian Andes of a young woman interred with a big-game hunting toolkit, and then reviewed the late Pleistocene and early Holocene record of the Americas. Between 30 and 50 percent of the big-game hunters they could sex were female. That finding was necessary because the opposite had been assumed, unmeasured, for a century. [14]
This is why the correction to Snow matters and why the strong headline was wrong. Not because women were not in those caves. Because the honest position is that we do not know who was in those caves, and only one of the two available answers was ever made to prove itself.
And it connects to Dissanayake. The theory that puts the mother-infant bond at the origin of art is not obviously weaker than the theory that puts male courtship display there. It has better formal detail and it fits the ornament record at least as well. It gets a fraction of the citations. Sometimes a field's blind spot is not in what it concludes. It is in which hypotheses get taken seriously enough to be attacked.
Evolutionary explanations of behaviour are the easiest stories in science to tell badly
Start with the structural problem. Behaviour does not fossilise. Everything in the second half of this post is an attempt to explain a psychological disposition using a record made of stone, pigment and shell. That gap is not closable, and any account that sounds fully satisfying is probably hiding it from you.
Then falsifiability. A good evolutionary hypothesis specifies what would disprove it. Sexual selection and byproduct accounts of art are both notoriously slippery here: almost any observation can be absorbed into either. The bonding account is the strongest of the four precisely because it made predictions someone could go and test in a room with actual people, and the artification account is the weakest on this axis for the opposite reason, however elegant it is.
Dating carries its own disputes. The claim that Neanderthals made red marks in three Spanish caves at least 64,800 years ago, published by Dirk Hoffmann and colleagues in 2018, would push symbolic behaviour outside our own species. It has been formally contested: Ludovic Slimak and colleagues published a comment in Science arguing the uranium-series dates on the carbonate crusts are not reliable in that setting, and the exchange has continued in the literature since. I have marked it as disputed in the timeline rather than leaving it out, because a genuinely open argument is more informative than a tidy one. [6][15]
And a note on what I have not claimed. Nothing here says art evolved for one reason. The likeliest answer, and the one most researchers in the area now hold in some form, is that several of these pressures acted on overlapping capacities at different times, and that what we call art is a family of behaviours rather than a single trait with a single origin. That is a less satisfying sentence than any of the four theories. It is also the one best supported.
If you have ever treated your own need to make things as a quirk, a distraction from real work, or something you would grow out of, the record disagrees at a scale that is hard to argue with. No human population has ever been found that does not do this. The behaviour predates the wheel by roughly 137,000 years.
Dunbar's pain threshold results separated performing from listening and only performing moved the marker. Part one found that making art lowered cortisol regardless of skill. Two different labs, two different measures, same direction: the participation is doing something the consumption is not. Put that next to how most of us now encounter art, which is by scrolling past it.
Singing together bonded strangers faster than writing or crafting together. The proposed mechanism is moving in time with other people, which blurs the self-other boundary and releases endorphins. No equipment, no skill threshold, no cost. If you are building anything that needs a group to cohere, this is the most under-used finding in the social science of the last twenty years.
The hand stencil story is a small, clean case study in a pattern you can now spot elsewhere: a claim about women gets rigorously tested, while the default it displaced was never tested at all. Once you have the shape of it, you will see it in clinical trial populations, in exercise physiology, in pain research, and in most of what this blog is about.
I did not know, as a child drawing worlds that do not exist, that I was doing something with a 73,000 year record behind it. I thought I was avoiding homework. Some of the time I was.
What the deep record gave me is not a justification, because I do not think the habit needs one. It gave me a correction of scale. The compulsion to make a thing that has no function, that feeds nobody and shelters nobody, was already present in people who had no agriculture, no writing, no cities and no reliable food supply. They did it anyway, with ground mineral and a piece of stone, in conditions where the calorie cost was real. Whatever that behaviour is for, it was worth it to them under pressure I will never experience.
And the four explanations do not resolve. Display, byproduct, bonding, artification. Each accounts for a part and leaves a part exposed. I find I am not troubled by that, and part of the reason is Dissanayake. If the origin of art really is the rhythmic, exaggerated, repetitive exchange between a woman and a baby who cannot yet speak, then the first art was not a signal to a mate or a monument for a group. It was two people regulating each other. That would make art, at its root, a technology of care.
I cannot prove that. Neither can she, and she says so. But of the four available stories about why my hands have always needed to be doing this, it is the one I would want to be true, and I am telling you plainly that wanting it to be true is not evidence.
Two down, one to go.
- Art and longevity. What happens in the body, and what the cohorts found.
- Art and evolution. Why we started, and why nobody can agree. You just read it.
- Art and science. The four hundred year argument between them. Vesalius and the anatomical plate, the history of who was permitted to draw the female body, and what each discipline still refuses to take from the other.
A recurring argument in The Art of Female Health is that the default assumptions about a female body are usually older than the evidence for them, and that finding out which is which is most of the work. This post is that argument applied to a cave wall instead of a clinic. nammu.academy/book
References
- Sehasseh, E. M., Fernandez, P., Kuhn, S., et al. (2021). Early Middle Stone Age personal ornaments from Bizmoune Cave, Essaouira, Morocco. Science Advances, 7(39), eabi8620. https://doi.org/10.1126/sciadv.abi8620
- Henshilwood, C. S., d'Errico, F., van Niekerk, K. L., et al. (2018). An abstract drawing from the 73,000-year-old levels at Blombos Cave, South Africa. Nature, 562(7725), 115–118. https://doi.org/10.1038/s41586-018-0514-3
- Snow, D. R. (2013). Sexual dimorphism in European Upper Paleolithic cave art. American Antiquity, 78(4), 746–761. https://doi.org/10.7183/0002-7316.78.4.746
- Oktaviana, A. A., Joannes-Boyau, R., Hakim, B., et al. (2024). Narrative cave art in Indonesia by 51,200 years ago. Nature, 631(8022), 814–818. https://doi.org/10.1038/s41586-024-07541-7
- Brumm, A., Oktaviana, A. A., Burhan, B., et al. (2021). Oldest cave art found in Sulawesi. Science Advances, 7(3), eabd4648. https://doi.org/10.1126/sciadv.abd4648
- Hoffmann, D. L., Standish, C. D., García-Diez, M., et al. (2018). U-Th dating of carbonate crusts reveals Neandertal origin of Iberian cave art. Science, 359(6378), 912–915. https://doi.org/10.1126/science.aap7778
- Dunbar, R. I. M., Kaskatis, K., MacDonald, I., & Barra, V. (2012). Performance of music elevates pain threshold and positive affect: Implications for the evolutionary function of music. Evolutionary Psychology, 10(4), 688–702. https://doi.org/10.1177/147470491201000403
- Pearce, E., Launay, J., & Dunbar, R. I. M. (2015). The ice-breaker effect: Singing mediates fast social bonding. Royal Society Open Science, 2(10), 150221. https://doi.org/10.1098/rsos.150221
- Tarr, B., Launay, J., & Dunbar, R. I. M. (2014). Music and social bonding: "Self-other" merging and neurohormonal mechanisms. Frontiers in Psychology, 5, 1096. https://doi.org/10.3389/fpsyg.2014.01096
- Dissanayake, E. (2013). The deep structure of Pleistocene rock art: The "artification hypothesis". Palethnologie, 5. https://doi.org/10.4000/palethnologie.5177
- Dissanayake, E. (1999). Antecedents of the temporal arts in early mother-infant interaction. In N. L. Wallin, B. Merker, & S. Brown (Eds.), The Origins of Music. MIT Press. https://doi.org/10.7551/mitpress/5190.003.0027
- Galeta, P., Bruzek, J., & Lázničková-Galetová, M. (2014). Is sex estimation from handprints in prehistoric cave art reliable? A view from biological and forensic anthropology. Journal of Archaeological Science, 45, 141–149. https://doi.org/10.1016/j.jas.2014.01.028
- Rabazo-Rodríguez, A. M., Modesto-Mata, M., Bermejo, L., & García-Dez, M. (2017). New data on the sexual dimorphism of the hand stencils in El Castillo Cave (Cantabria, Spain). Journal of Archaeological Science: Reports, 14, 374–381. https://doi.org/10.1016/j.jasrep.2017.06.022
- Haas, R., Watson, J., Buonasera, T., et al. (2020). Female hunters of the early Americas. Science Advances, 6(45), eabd0310. https://doi.org/10.1126/sciadv.abd0310
- Slimak, L., Fietzke, J., Geneste, J.-M., & Ontanón, R. (2018). Comment on "U-Th dating of carbonate crusts reveals Neandertal origin of Iberian cave art". Science, 361(6408), eaau1371. https://doi.org/10.1126/science.aau1371
Two further sources are books rather than papers and are cited as such in the interactive: Miller, G. (2000). The Mating Mind. Doubleday. Pinker, S. (1997). How the Mind Works. Norton.