Sunlike
Did they find the same thing? We asked sixty strangers to read passages from traditions that never met. Here is what they saw, and what it might take to see more.
The night the numbers came back, I was asking a narrow question. The Map pairs teachings from different traditions that say, as far as I can tell, the same thing about how we experience existence. Some of those traditions met each other. Buddhism walked into China and married the Tao. Plato’s heirs taught the Christian mystics and the Sufis. Agreement between traditions that met could be borrowing, and borrowing proves nothing except that a good idea travels. So I asked how much of the Map’s agreement borrowing could explain.
The answer was less than half. Of the connections between historical traditions on the Map, more than half run between traditions with no known contact at the time their teachings were written. Eckhart never read the Upanishads. The authors of the Gospels never read Lao Tzu.
Here is what I think that means. These traditions arose apart. Whether what they found is a truth about the world, a feature of consciousness, or the wiring every human nervous system shares is an open question, and I don’t think it matters as much as it seems to. All three would be remarkable. All three come down to the same thing said about how we experience existence.
There is one answer that would matter, though. If the agreement was manufactured, it isn’t agreement at all. Someone chose which passages to set side by side. Someone translated them into English. If the choosing and the translating did the work, then the Map shows its maker’s eye and nothing more.
That answer can be tested, so we tested it. Strangers who had never seen the Map read pairs of old passages, some that the Map connects and some drawn at random from the same books, without knowing which was which, and rated how much each pair made the same point. The rules were written down and frozen before anyone answered.
Here is what they found. The strangers, reading cold through difficult old translations, still saw a small but real extra resemblance in the pairs the Map connects. Small is the honest word: about half a point on a five-point scale. Whether someone who knows these texts would see more, because they have learned to see, or because they have been taught what to find, is the question this test can’t answer. The rest of this piece is how we got here, and why that last question is the right one to be left holding.
I.Forty times
In 1977 two zoologists at Harvard, Luitfried von Salvini-Plawen and Ernst Mayr, went through the animal kingdom phylum by phylum and asked a simple thing of every eye they found: what is it made of? Whether the light-catching membrane grows from a cilium or from a fringe of microvilli. Whether the eye forms from skin or from nerve tissue. These details turn out to be stable within a group of animals and different between groups, the way a building’s foundation tells you more about who built it than its paint. By their count, eyes had evolved independently at least forty times, and perhaps sixty-five.
The squid is the famous case. It has a camera eye like ours: a lens, an iris, a retina. Look closer and the two are built backward from each other. Our retina is wired from the front, so the nerve fibers cross over the light-catching cells and leave a blind spot where they exit. The squid’s is wired from behind and has none. The cells respond to light with opposite electrical signs. Two lineages arrived at the same instrument by different roads.
It doesn’t even take long. In 1994 two Swedish biologists, Dan-Eric Nilsson and Susanne Pelger, modeled a flat patch of light-sensitive cells improving one tiny step at a time, choosing every assumption to make the process slow. Their pessimistic answer was about 364,000 generations to a focused lens eye. For a small animal that is less than half a million years, a blink of geological time.
Then, the same year, Walter Gehring’s laboratory in Basel published something that seemed to undo all of it. A gene the fruit fly needs to build its eyes, called eyeless, turned out to be the same gene a mouse needs to build its eyes, called Pax-6. Gehring’s group switched the gene on in the wrong places and grew extra eyes on a fly’s legs, wings and antennae. When they used the mouse version of the gene, the fly still grew fly eyes. Gehring argued that the old picture of many independent eyes had to be reconsidered. One master switch, one ancestral eye, forty descendants.
Mayr didn’t move. Interviewed in 1997, he was firm that the gene did not overturn what he and Salvini-Plawen had found. The likelier story, he and others suspected, was that the gene had first belonged to building the nervous system, and that lineage after lineage later put it to work on eyes. Two years later, giving a lecture in Mayr’s name, the zoologist Michael Land offered the detail that settles the intuition: there are animals with Pax-6 and no eyes at all, corals among them. The switch is older than the thing it switches on.
Biologists have stopped treating this as a fight. Every animal light-detecting protein descends from one ancestral protein, so the parts list is shared. The eyes built from those parts arose many times. How many “times” you count depends on what you decide counts as an eye, and whether you look at molecules, at cells or at whole organs.
What makes eyes keep happening? Biology gives four answers, and it doesn’t make you choose between them.
- There is light. Eyes evolve because there is something out there to see, and seeing it pays.
- There is a shared toolkit. Every lineage that builds an eye starts from the same inherited parts.
- The problem has a shape. Optics allows only a few good designs, so separate lineages find the same ones.
- There is a choice of what to count. The number of origins depends partly on the definition of an eye.
All four are true at once.
Those four answers are the same four this Inquiry is asking about. The traditions might agree because there is something real they are all responding to, the way eyes respond to light. They might agree because every human brings the same equipment, the same machinery the site’s dispatch The Same Machinery described. They might agree because any creature that suffers and dies will ask the same few questions, and there are only so many good answers. Or the agreement might be partly an artifact of how we count, of who chose the passages and who put them into English.
For eyes, the four can be pulled apart, because we can measure light without using an eye. For the traditions, we can’t measure the thing they say they found without a human being doing the measuring. That’s where this Inquiry will leave its question. But the fourth answer, the one about counting, can be checked, and the case against goes first.
II.Copies are not confirmations
On November 13, 1888, the anthropologist Edward Tylor stood before the Anthropological Institute in London and presented something new: a statistical comparison of customs across hundreds of societies. He was looking for customs that tended to appear together, and he found them. When he finished, the chairman stood up. It was Francis Galton, and his objection is still with us. How many of these societies, Galton asked, had really arrived at these customs on their own? Some had surely copied each other, or inherited them from a common ancestor. If they had, they were copies, and copies can’t be counted as separate confirmations. The problem now carries his name.
The Map has already done some of what Galton asked. The site’s Who Met Whom grades every pair of traditions by whether they had contact, and the Inquiry’s opening numbers come from it. But Galton’s problem has relatives, and some of them land on claims this site has already made.
The instrument found what it was built from. The site’s dispatch The Same Machinery said that a structure in mystical experience “holds from the United States to Iran.” The claim rested on a 2001 study by the psychologist Ralph Hood and colleagues. They gave a questionnaire to American and Iranian samples and found the same three-part structure in both. The questionnaire, though, was built from one philosopher’s theory of what mystical experience is: W. T. Stace’s, from 1960. The Iranian sample was 185 people answering a translated version of it. Critics, the Dutch psychologist Jacob van Belzen among them, point out that a questionnaire built from a theory will tend to find that theory wherever it is given. The study shows the questions travel well. It shows less about whether the experiences were alike before anyone asked. That sentence in The Same Machinery claimed more than its evidence holds, and it now carries a footnote saying so.
The blind men had read the same books. The largest new study of the experience mystics call pure awareness comes from the German philosopher Thomas Metzinger, whose 2024 book takes its title from the Map’s own parable: The Elephant and the Blind. His team’s survey drew 1,403 usable responses from meditators. About two-thirds answered in German and nearly a third in English. Most practiced Vipassana or Zen. These are serious practitioners describing a state they could reach again, and the study shows that state is real and describable. What it can’t show is that strangers found it separately. Most of its blind men had studied with the same teachers.
The structure that shrank. For decades the near-death experience had a famous core: the tunnel, the light, the life passing before one’s eyes. When the sociologist Allan Kellehear gathered accounts from outside the West, the tunnel largely disappeared, and the life review was mostly a Western feature. A study of accounts recorded before and after the 1975 bestseller Life After Life found that only one of its fifteen classic features had become more common after the book: the tunnel. The parts that held up across cultures are humbler: leaving, a boundary, a presence, a return, a changed life. The danger is plain. If “structure” just means whatever survives each new comparison, the claim can never lose.
A universal that was borrowed. In 2015 a team led by Joseph Watts placed 96 Austronesian cultures on their family tree, a way of correcting for Galton’s problem with a family history in hand. Belief in supernatural punishment showed up early, before complex societies. Belief in a moralizing high god came after them, and the tree suggests it spread between them. Control for descent, and a universal can turn out to be a loan.
The translators chose the words. Most of what Western readers know of the Eastern classics, they know from Victorian translators. One of the most important was James Legge, a Scottish missionary who spent decades in Hong Kong and became Oxford’s first professor of Chinese. Legge believed the ancient Chinese had known the true God. When he translated the old Chinese name for the highest power, Shang-ti, he wrote “God.” When those translations appeared in the Sacred Books of the East in 1879, a group of missionaries wrote to the series editor, Max Müller, to protest the choice. Müller defended it. Both sides understood the stakes: a translator’s word can make two traditions sound more alike than they are, or less. A reader of Legge meets “God” where a reader of the Chinese meets something else.
Put together, the case against is strong. The literature can’t rule out that the Map’s agreement is partly manufactured. The scholar Samuel Sandmel gave the danger a name, and the next section starts with him.
III.The test we ran
In December 1961 Samuel Sandmel gave his presidential address to the Society of Biblical Literature and called it “Parallelomania.” His target was scholars who line up passages from two bodies of writing, find them similar, and then conclude that one borrowed from the other. Two passages can sound alike out of context and mean quite different things in it. His remedy was to read whole texts, not excerpts set side by side.
The Map is excerpts set side by side. It has always admitted that arranging is an act. So we built a way to ask whether the arranging did the work. The test, its rules and its results are on their own page.
What we built. We took four sets of books from traditions that never met: the Gospels and Epistles, the Tao Te Ching, the Bhagavad Gita and two of the Upanishads, and the Roman Stoics. From them we built three piles of passage pairs, 39 in each:
- The Map’s pairs: passages the Map already connects.
- Random pairs: passages drawn by a computer from the same books, matched in length.
- Furniture pairs: random pairs whose passages both mention the furniture of religion: heaven, sacrifice, temples, the dead.
The last pile was there to answer a sharper worry: do readers say “same point” whenever two old religious texts are set side by side?
Every passage in every pile came from one public-domain translation per book, chosen to be close to the original and plain to read. We set aside translations that paraphrase, including the popular modern Tao Te Ching the Map itself quotes. A translator who smooths texts toward each other would flatter the Map, and we wanted the test to be hard to flatter.
Each reader saw twelve pairs, four from each pile, in random order, with no labels and no sources. For each pair they answered one question on a scale of 1 to 5: Do these two passages make the same point about how we experience existence? They also said whether they had read either passage before. Before anyone answered, we wrote down what would count as a win for the Map, what would count as a small effect, and what would count as a loss, and froze it.
The first readers. Three members of my family took a trial run, and their answers were never counted. It took them 16, 23 and 39 minutes, and all three said the passages were hard going. The old translations are dense. We cut each reader’s load from eighteen pairs to twelve, added one line to the instructions (read for the gist), and left the passages alone.
Finding strangers. I posted the test to two Reddit communities that exist for surveys and bought a small ad. The ad drew 88 visitors to the test, who stayed an average of eighteen seconds and produced one completed response. People scrolling a feed don’t sign up to read old scripture for ten minutes. Then we turned to Prolific, a platform that pays people to take part in research. Fifty-eight readers there, from six English-speaking countries, took the test for $3 each. With the two who came through Reddit, that made the sixty the rules asked for. We had expected to wait three weeks. Prolific filled the places in an afternoon, and the test closed early, on reaching its sample, before anyone looked at a single answer. The decision and the reason were written down first.
Two readers finished in under four minutes and were set aside, as the rules said in advance. That left 58.
What they saw. On average, readers rated the Map’s pairs 2.88 out of 5. Random pairs from the same books got 2.23, and furniture pairs got 2.26.
For each reader, we took the gap between how they rated the Map’s pairs and how they rated the random ones, then averaged across readers. The gap was 0.56 points. Allowing for chance, the true gap is likely between 0.20 and 1.03. It is clearly above zero: the Map’s pairs were not just what a random draw from these books produces. It is also well short of the full point the rules set as the mark of strong support. By the rules written in advance, the result is a small effect.
Three other findings matter.
- Furniture didn’t fool anyone. Pairs that shared religious vocabulary (heaven, sacrifice, the dead) were rated no higher than random pairs. The extra resemblance readers saw belonged to the Map’s pairs alone.
- Readers disagreed with each other a lot. On a measure where 0 means readers agree no more than chance and 1 means they agree completely, these readers scored 0.13. The same pair could look like a match to one person and like nothing to the next.
- Recognition changed little. Keeping in the pairs readers said they had read before moved the result only slightly, to 0.60.
Readers also told us how they lean: toward the empirical, the contemplative, or neither. All three groups saw the same direction of gap. Those who leaned neither way saw the largest (0.82, from fourteen readers). The empirical group was too small to say much (0.69, nine readers). The contemplative group, the largest, saw the smallest (0.41, thirty-two readers). None of those groups is large enough to say they really differ.
What the test doesn’t settle. The limits are part of the result.
- Nearly every reader was a paid participant on one platform, all from English-speaking countries. A different crowd might read differently.
- Many read quickly. A full set runs about 1,800 words, which takes most people seven to nine minutes just to read, and many of these readers finished in four to seven. The only speed rule fixed in advance was under four minutes, so that is the only one we applied. Changing it after seeing the times would have meant changing the rules to suit the result.
- The test closed early. That was decided before any answer was opened, and it is recorded with its reasons.
- Some surface features differ between piles. The Map’s Bible passages are more often addressed to “thee,” and more often name God, than the random ones. Both come from what those teachings are about, and neither can be removed without changing the text. We noticed the second only after the result was in, and report it here for that reason.
- The passages are hard, and hard reading adds noise. The low agreement between readers says the same.
That last point cuts in a particular direction. Noise makes real differences look smaller, not larger. If careful readers with clearer texts would see more, this result is closer to a floor than a ceiling. That is an argument, not a finding, and the next test can check it.
IV.What survives
Writing was invented at least three or four times: in Mesopotamia, probably in Egypt, in China, and in Mesoamerica, by people who never met. Farming was invented in several places. Nobody concludes from this that writing is an illusion, and nobody infers a god from hunger. Independent invention shows that a problem was real and that people solved it, more than once, in similar ways.
Here is the rule this Inquiry ends up with. When traditions arose apart, and their agreement survives a test against random passages from the same books, and they disagree on the furniture, the convergence is real. On this evidence the Map’s agreement passes, modestly. Strangers who had never seen the site, reading old and difficult English, rated the pairs it connects as more alike than chance pairs, and rated religious vocabulary as no more alike than chance. Selection and translation didn’t produce all of it.
What remains open is what the agreement is about. The test can’t tell a truth about the world from a feature of the mind from the shape of the questions every human life asks. Nothing in this piece can, and the rest of it is no longer evidence.
From here the piece leaves what can be tested.
My view is that the first two may be closer than they look. For the eye, the light and the toolkit stopped being rivals long ago. The toolkit was shaped by the light. If minds were shaped by whatever the traditions met, then “a feature of the wiring we share” and “a truth about the world” may be the same fact, seen from either end. That is a possibility, not a finding.
V.Sunlike
In Rome in the mid-250s, the philosopher Plotinus wrote the first treatise of his life, a short one on beauty. Near its end he turns to the reader who wants to see the highest beauty, and tells them what it will cost. The eye has to be made ready first, made like the thing it means to see. In Stephen MacKenna’s translation:
“Never did eye see the sun unless it had first become sunlike.”
More than fifteen centuries later, in 1810, Goethe opened his book on color with the same thought, turned toward biology. In Charles Eastlake’s English of 1840:
“The eye may be said to owe its existence to light, which calls forth, as it were, a sense that is akin to itself.”
He goes on: the eye is formed with reference to light, “the light it contains corresponding with the light without.” Then he reaches back to an old mystic for the rhyme he had in mind: “If the eye were not sunny, how could we perceive light?” Half a century before Darwin, a poet stated the argument from convergence: the seer is shaped by the seen. And unlike most of the agreements on the Map, this one certainly traveled, from Plotinus to Goethe, who credits it to an unnamed mystic.
The test’s result sits right on Plotinus’s line, and the line cuts both ways.
The promise. I expected the passages to be hard. That was the safeguard. What I didn’t expect was for strangers reading them for the first time to see what someone who has spent years with this material sees. They didn’t, quite. They saw something, and it was real, but faint. I think it takes a trained eye. If an eye has to become sunlike to see the sun, then a reader who has lived with these texts may see in them what a stranger, reading quickly through Victorian English, can only glimpse.
The danger. That is exactly what Sandmel warned about. A trained eye sees more. It may see more because it has learned to see, the way a radiologist sees a shadow on a scan that the rest of us miss. Or it may see more because it was taught what to find, the way every parallelomaniac found borrowing everywhere. From inside, the two feel the same. That is why I didn’t rate the test, and why its readers were strangers.
The test can’t tell those two apart. A better one could. It would use plain, modern, literal translations in every pile, so that reading is easy without any translator smoothing the texts toward each other. And it would ask each reader how much of these traditions they have read. If familiar readers see a larger gap than strangers, and the gap holds against random pairs, the trained eye is seeing something. If familiar readers see resemblance everywhere, in random pairs as much as in the Map’s, the eye was trained to find it. That is Field Test 002, and the site will run it.
Until then, the honest place to stand is where the result put us. Sixty strangers, reading cold through hard translations, saw a little more agreement where the Map says there is some than where there is none. Not much more. Enough to say the arranging didn’t do all of it.
The eye doesn’t need to know whether light is a wave or a particle to see by it. It needs only to have become, a little, like what it sees. Whatever the traditions found, they were shaped by it, and they said so in nearly the same words in places that never met. Everything else is still open.
What this Inquiry does not establish
- That “arose apart” means no contact ever happened. It means no known contact when the teachings were written. Influence can travel in ways history doesn’t record.
- That the agreement points to an object rather than to shared wiring or to the shape of the problem. The test was built to rule out one explanation, the manufactured one, and it rules it out only partly.
- That the result would hold for other readers or other translations. Almost every reader came through one paid platform, from six English-speaking countries, and many read quickly. The translations are hard.
- That a trained eye sees more truly. That is the author’s view, and Field Test 002 is designed to test it.
- That the eye is evidence. It is a frame. Light can be measured without an eye. What the traditions describe can’t be measured without a person.
- That the effect is large. It isn’t. About half a point on a five-point scale, likely between 0.2 and 1.0.
A note on process: the ideas, the direction and the lived experience here are mine. I write these with Claude, an AI model, for research, structure and drafting, and I read and sign off on every piece. More on the About page.
- Salvini-Plawen, L. v., & Mayr, E. (1977). On the evolution of photoreceptors and eyes. Evolutionary Biology 10, 207–263.
- Nilsson, D.-E., & Pelger, S. (1994). A pessimistic estimate of the time required for an eye to evolve. Proc. R. Soc. B 256, 53–58.
- Quiring, R., Walldorf, U., Kloter, U., & Gehring, W. J. (1994). Science 265, 785–789.
- Halder, G., Callaerts, P., & Gehring, W. J. (1995). Induction of ectopic eyes by targeted expression of the eyeless gene in Drosophila. Science 267, 1788–1792.
- “Eye-opening gene: How many times did eyes arise?” Science News, 10 May 1997.
- Land, M. F. (2000). The evolution of eyes (Ernst Mayr Lecture, 9 Nov 1999). Berlin-Brandenburgische Akademie der Wissenschaften.
- Tylor, E. B. (1889). On a method of investigating the development of institutions. Journal of the Anthropological Institute 18, with Galton’s discussion.
- Hood, R. W., Ghorbani, N., Watson, P. J., et al. (2001). Journal for the Scientific Study of Religion 40(4), 691–705.
- Gamma, A., & Metzinger, T. (2021). The Minimal Phenomenal Experience questionnaire. PLOS ONE 16(7), e0253694.
- Metzinger, T. (2024). The Elephant and the Blind. MIT Press. Open access.
- Kellehear, A. (2008). Census of non-Western near-death experiences to 2005. Journal of Near-Death Studies 26(4).
- Athappilly, G. K., Greyson, B., & Stevenson, I. (2006). Journal of Nervous and Mental Disease 194, 218–222.
- Watts, J., et al. (2015). Broad supernatural punishment but not moralizing high gods precede the evolution of political complexity in Austronesia. Proc. R. Soc. B 282, 20142556.
- Müller, F. M. Introduction to the Science of Religion, appendix “On the Chinese name for God.”
- Sandmel, S. (1962). Parallelomania. Journal of Biblical Literature 81, 1–13.
- Field Test 001: the results page, with the rules as fixed in advance, the answer key, the analysis script and the anonymous data.
- Plotinus, Enneads I.6.9, trans. Stephen MacKenna and B. S. Page.
- Goethe, J. W. von (1840). Theory of Colours, trans. C. L. Eastlake. John Murray. Introduction. Project Gutenberg.