Synthesis — Psychology

The evolving thesis. Spun out of the hub _inbox on 2026-07-28 from three sources parked under two tags across three weeks: jung-on-creativity (psychology-creativity, 2026-07-23), human-mouse-perceptual-maps (neuroscience, 2026-07-12), and kahneman-focusing-illusion — the trigger.

Current thesis

The mind runs on structure its owner didn't choose and can't inspect. Three sources from three incompatible traditions make the same claim about different layers:

• Attention. What you're looking at inflates its own importance, so a judgment partly reports the shape of the question rather than the world (focusing-illusion).
• Perception. A listener sorts sounds using an internal map that was largely in place before any training, and shows up in mice too (perceptual-learning).
• Creation. The artist is an instrument of material arriving from a layer they don't author (analytical-psychology).

Stated plainly, that's the discipline's oldest idea and the reason it needed to become empirical: introspection is being used to study the thing that introspection can't see.

The founding tension — three standards of evidence

The corpus is a method spectrum, and it was assembled by accident, which makes it useful:

The claims get more interesting and less checkable in the same direction. That is the spoke's standing caution — not that Jung is worthless or that psychophysics is the only real psychology, but that a claim's reach and its evidence tend to trade against each other here, and a source's prominence tracks neither.

Amended 2026-07-29 — "clinical" is not a position on this spectrum. Four therapy books arrived together and broke the ordering. Hollis and Yalom sit where the table predicts, at the interpretive end with Jung: rich frameworks, no measurement attached, and none possible without changing the method. Hayes sits at the opposite end of the whole corpus — ACT carries over 900 randomized controlled trials, WHO distribution, and APA and VA recognition, which is a larger evidence base than human-mouse-perceptual-maps, the spoke's only experiment. All three are clinicians treating patients.

So the spectrum's third row was mislabelled. It was never measuring clinical work; it was measuring a particular tradition's relationship to testing. Depth psychology declines the question, behaviour analysis builds its programme around it, and both are practised in a consulting room. The corpus assumed one axis and had two: what a source is about, and whether its tradition thinks evidence is the point.

The caution survives in better form. burnout-stress-cycle is the demonstration: a trade paperback making confident physiological claims with thin citation, shelved next to a trade paperback whose therapy has 900 trials behind it and which mentions none of them. Reach and evidence still trade against each other — but not at the level of the discipline, at the level of the book.

Sharpened again the same day (dopamine-nation). Lembke is a Stanford professor of psychiatry, medical director of an addiction medicine programme, author of 100+ peer-reviewed papers on the literature her bestseller compresses. She sits at no untested end of anything. The criticism — documented, unlike for the other four books — is that the trade book outran the work: the pleasure-pain seesaw is tidier than the neuroscience, "dopamine fasting" names a thing you cannot do to a neurotransmitter, the implied recovery timescale is days where the literature says weeks, and the anti-dopamine framing carried into territory (ADHD medication, abstinence versus medication-assisted treatment) with clinical consequences its author then walked back in a newspaper.

So the variable is register, not discipline, tradition, or credential. The same author can produce a peer-reviewed finding and a bestseller whose claims their own field disputes, and it is the second that a million people act on. This is the wiki's founding note — psychology has a replication problem and popular writing about it compounds one — with both steps visible inside one career. It also sets the corpus's sharpest evidential contrast: human-mouse-perceptual-maps, one experiment nobody outside the field has heard of, against dopamine-nation, a household phrase whose author has conceded its central simplification.

Both sides counted (the-body-keeps-the-score, 2026-07-29). Van der Kolk's book is the register problem with numbers attached in place of impressions. Reach: nearly 300 weeks on the New York Times bestseller list, 36 languages. Evidence: a 2025 BJPsych Bulletin review that worked through 122 claims and found the two load-bearing ones — trauma-induced brain damage, and the unique efficacy of body-based treatment — unsupported by current research. Every earlier version of this finding on the spoke had a vague quantity on one side; this one has a figure on each.

Two things it adds beyond that arithmetic. The sequence doesn't close. Lembke conceded her central simplification in a newspaper, and the corpus quietly read that as how the process works: claim, criticism, correction. Van der Kolk received McNally's critique in 2003, eleven years before the book, and wrote the book anyway; the criticism since has been peer-reviewed and has changed nothing about the readership. Correction was a temperament, not a mechanism. And the criticism names a harm rather than an inaccuracy: readers are steered away from CBT, which has the trials the somatic treatments lack, and Emi Nietfeld's Mother Jones piece argues the framing stigmatises the survivors it addresses. On burnout-stress-cycle and dopamine-nation the stakes were what a reader might waste effort on. Here they are what a traumatised reader declines.

Determinism, and why the argument is really about blame

Six sources arrived the day after founding, and three of them make the founding thesis explicit rather than implicit. If the mind runs on structure its owner didn't choose, the obvious next question is whether the owner chooses anything — and free-will is where that lands.

The useful move is splitting the question three ways: whether free will exists (metaphysics, and ../philosophy-wiki's), what actually produced a given behaviour (Sapolsky's biology), and what happens to people who stop believing in it (free-will-belief-effects, the only one this corpus can measure — less hatred, more forgiveness, and across 44 countries, weaker free-will belief tracking milder punishment).

What all three are really about is blame. Not agency in the abstract but how harshly to treat people, which is why Sapolsky spends his second half on criminal justice and why Walton measures punitiveness and shame. Determinism enters this wiki as an ethics of treatment.

And the sources arrive already arguing about who owns them. Two of determined's four reviewers say a neurobiologist has answered a philosophical question — Riskin calls it scientism, Setiya faults the ignored literature. The routing rule sends them here because the methods are biology and experiment; that decision is contested by the sources themselves, and the contest is recorded rather than settled.

Evidence is a subject here, not just a caution

The founding synthesis treated evidence quality as a caution to apply. Two of the new sources make it a topic: thinking-fast-and-slow's own citations fail to replicate (Chapter 4's replicability index of 14; daniel-kahneman: "I placed too much faith in underpowered studies"), and free-will-belief-effects uses non-replication as an argument, clearing away the older findings that free-will scepticism increases cheating.

That second use deserves the scrutiny. "It didn't replicate" is both a legitimate reason to discount a result and a convenient one, and here it is being applied to the opposing literature by an author whose own evidence is two experiments of ~260 participants. replication-crisis holds the general version. The corpus's method spectrum now has a companion: not just how strong a source's evidence is, but how it treats everyone else's.

A third form arrived the same day: the citation chain. why-writing-feels-hard cites Eisenberger, Baumeister, Chen, Raichle, Csikszentmihalyi and Dietrich by name — real work, cited accurately enough that spot-checking any one reference finds a genuine paper. The argument lives in the joins. Eisenberger's overlap between social and physical pain becomes "your brain… is genuinely convinced that putting words on a page might kill you"; Csikszentmihalyi's much-repeated 120-bits-per-second is used as a constant; transient hypofrontality is treated as mechanism rather than hypothesis. And the last join is the widest — no cited study is evidence that the practice being sold, ten minutes of insight meditation before writing, does anything for writing.

That's a harder failure to catch than "studies show," precisely because the references check out individually. Worth holding as a named pattern: a citation chain terminating in a practice, where each link is real and the load-bearing welds between them are untested. Note also that this spoke's standing caveat applies from inside the citation list — Baumeister is a central figure in the replication-crisis, which the essay doesn't mention.

A fourth form, and it is not about weak evidence at all (2026-08-05). dunning-kruger-misread brings an objection the three above have no room for. The 2022 Frontiers in Psychology case against the dunning-kruger-effect is that noisy self-judgments on a bounded scale reproduce the effect's pattern in data with no effect in it. The result is not fragile. It shows up every time, including when nothing is there.

That breaks the implicit ordering this section had been using, in which evidence gets better as it gets more replicable. Here replication is beside the point: a floor effect plus measurement noise generates a systematic-looking error with no mechanism behind it, and running the study on more people sharpens the artifact rather than exposing it. replication-crisis now carries both — the findings that vanish on re-testing, and the finding that never vanishes and may still be empty.

It also lands on heuristics-and-biases where it hurts most. Systematic rather than random was the claim that turned the programme into a science; a bounded scale supplies systematic direction for free.

A fifth form, and the most ordinary one (2026-08-12). handwriting-vs-typing-notes brings the outcome the four above leave no room for: the finding neither held nor vanished nor turned out to be an artifact. It shrank. Mueller & Oppenheimer's handwriting advantage survives a 2024 meta-analysis of 24 studies at Hedges' g = 0.248, while a 2021 direct replication (74 laptop, 68 longhand) and a 2026 EEG study (n=33) each found nothing on immediate testing. Something is there. It is small, and it is conditional on what the test asks for.

This is worth a slot of its own because it is almost certainly the most common fate of a famous result, and the one the corpus is least equipped to notice. A vanished finding makes news. An artifact makes an argument. A finding that comes back at a quarter of a standard deviation makes neither, so the 2014 version keeps circulating unopposed — not because anyone defends it, but because g = 0.248 cannot be said in a headline. Put beside feeling-good's cover copy below, the two are the same fault from opposite directions: the register a result gets sold in has no way to express "true and small."

There is a methodological payoff too. The nulls are only nulls against the popular framing. Read through the mechanism — the constraint on writing speed forces compression, and compression is what the later test rewards — a study measuring immediate recall is measuring before the effect has anywhere to show up. The three studies stop contradicting each other and start describing a boundary condition (note-taking-and-learning).

And the thread has one case running the other way, which is what keeps it from collapsing into "popular psychology is unsupported."

The one bestseller that was tested

feeling-good breaks the pattern this spoke had started treating as a law. The corpus's bestsellers arrive with enormous reach and an audit that goes badly afterwards: the-body-keeps-the-score against a 122-claim review, dopamine-nation against its own field, thinking-fast-and-slow against a chapter-level replicability index of 14. Burns's book sold four million US copies and the studies point the same way — a meta-analysis of six studies finding a large helpful effect for self-directed bibliotherapy, eleven more supporting the guided version, and Scogin et al. (1989) finding it effective for older adults with mild-to-moderate depression.

Note what was tested. Not the theory behind the book, and not a therapy the book describes: reading the book was the intervention. That is the only clean case in this wiki of a popular psychology title being evaluated as the thing it actually is — an object people buy and use alone — rather than being assessed on claims its author made in a different register.

Two limits keep this from being a clean win. The trials are small, several attach therapist contact to the "self-help," and the professional-recommendation survey dates from 1994. And the jacket outruns all of it: "cured without drugs," "immediately." david-d-burns argues that antidepressant research carries drug-favouring biases, which makes the cover copy on his own product the corpus's sharpest instance of the register problem — a modest true claim sold immodestly, which is a different fault from bessel-van-der-kolk's unsupported one and should not be filed as the same thing.

The distortion with no author behind it

Every overreach recorded above has a person at the end of it. Van der Kolk's unsupported claims, Lembke's seesaw, Burns's cover copy, Dávila's meditation practice — in each case someone stretched a result and put their name on the stretch. The spoke had begun treating that as the shape of the problem.

dunning-kruger-misread is the other shape. The 1999 paper measured self-assessment against performance at one sitting, and the popular version has a time axis, a rise, a collapse and a chart — the "Mount Stupid" curve, which does not appear in the paper. Nobody in the study was followed as they learned; the study everyone describes was not run. The distortion accumulated downstream, in the retelling, and it is now the form in which the finding is known.

This is a distinct failure mode and worth a name of its own: the finding that got away from its authors. The register problem asks whether a claim's presentation matches its evidence, and assumes a presenter. Here the evidence and the original presentation agree with each other fine, and neither resembles what the public holds. Any popular source in this corpus may be relaying a version like that, and — unlike a jacket promise — there is no author's overreach to catch it by.

And the first popular source caught relaying the corrected version

The paragraph above closes on a worry: any popular source here may be passing on a runaway version of a finding, with no author's overreach to catch it by. epictetus-learning-spacedaily (2026-08-09) is the first case in the corpus where that did not happen.

The genre is the one most exposed to it — a daily inspirational column, opening on a quotation, whose whole job is to make an idea feel portable. It states the 1999 result, then gives the correction, citing Eric Gaze in Scientific American: the defensible finding is mostly that most people rate themselves as better than average. No "Mount Stupid," no confidence curve, no claim about intelligence. That objection is already held here as Krueger and Mueller (2002), so the piece adds a fourth voice rather than new evidence — but it establishes that the corrected version now travels in the same channels the myth does.

One data point, and it does not retire the worry. It does mean this spoke should stop treating genre as a proxy for accuracy, which is what it had begun doing: does-language-determine-thought sits at T4 for citing nobody, and a column of the same weight cites four researchers and a critique. The tier tracks what a piece rests on, not where it appeared.

It also brought a mechanism the corpus was missing. illusion-of-explanatory-depth — via Fisher, Goddu and Keil's nine Yale experiments (2015), in which searching the internet inflated people's sense of understanding without changing what they knew. That belongs beside dunning-kruger-effect and is routinely merged with it, wrongly: Dunning-Kruger is about rank, measured once, with nothing done to the participant; this is about depth, and it is inducible on demand. A bias you arrive with and a distortion you can be handed are different objects, and the remedy differs too — the second has a test (explain it without looking anything up), the first does not. Held at one second-hand report; neither Rozenblit and Keil's original nor the 2015 paper is in the corpus.

The interpretive traditions, and the one instruction they share

The wing of this spoke that does not measure things has turned out to be more coherent than its individual sources suggested. Three threads run through it: a lineage, a shared clinical stance, and a theory of creation that collides head-on with the experimental account.

sigmund-freud and erich-fromm complete a line the founding corpus started mid-way with carl-jung. Freud is the origin of the corpus's central predicament — enormous explanatory reach on a clinical method that cannot test its own claims — and Jung and Fromm are two different departures from him: Jung deepening the unconscious into a shared layer, Fromm socializing it, using Marx to ask what an economic order does to the psyche it forms.

Fromm adds a genuinely new object: social character, the psyche as something a political order shapes. That is neither Jung's individual depths nor Kahneman's laboratory, and it stretches the spoke further than any other source so far.

Welcome the symptom — three schools, one instruction

in-the-absence-of-the-ordinary makes the third independent statement of what is now the corpus's most repeated clinical instruction. james-hollis on midlife symptoms: they "should be welcomed, for they represent… a powerful imperative for renewal." irvin-yalom on distress: a response to the givens of existence rather than a malfunction (existential-psychotherapy). francis-weller on grief and uncertainty: "welcoming all of who we are into the story. Nothing excluded" (grief).

Three schools with different lineages, converging on one instruction, none of them reaching it by measuring anything. Record the convergence as what it is: not corroboration in the sense the rest of this wiki uses the word. It shows that the interpretive traditions share a stance toward suffering. Whether the stance helps anyone is a question none of these three sources is built to answer. acceptance-and-commitment-therapy is the interesting comparison: same instruction, roughly (accept rather than eliminate the aversive experience), and 900+ trials behind it. If a single finding connects the poetic corner of this spoke to its evidential one, it is that.

Weller's addition is scale. Everything else here locates the structure inside a person — family conditioning, perceptual maps, one judging mind's biases. Weller claims a collective initiation: "We are tumbling through a rough initiation. Radical alterations are occurring in our inner and outer landscapes." The only prior move in that direction is erich-fromm's social character, where an economic order forms the psyche. Fromm named the mechanism; Weller does not, and a period is not a patient.

On evidence, this is the far edge. T3 rates Popova's transmission, not what is transmitted: statements like "soul navigates the twining trail between sovereignty and intimacy" have no observation that would bear on them. the-middle-passage at least carried an untestable mechanism; this carries an image. It is also the first source here on a subject with a substantial clinical literature that it never mentions — grief lists what is missing, including a formal diagnosis the "grief is not pathology" framing would have to argue with.

Creativity has two accounts, and they disagree about what could settle it

flow gives the corpus a second explanation of making things, and it sits directly against the one already here. jung-on-creativity has the visionary work arriving from the collective unconscious with the artist as its instrument — knowable through interpretation. Flow has a describable state with proposed neural correlates (Default Mode Network, transient hypofrontality) that a person can be trained into.

Both are accounting for the same fact: good creative work doesn't feel deliberately authored. They disagree about what that's evidence of, and — the part that matters here — about what could ever settle it. So the founding method spectrum now appears within a single topic rather than only across different ones. That's a better test of the spectrum than the founding corpus offered, since the subject is held constant and only the method varies.

Both accounts are thin. Each arrived through a second-hand source (an essayist on Jung, a book excerpt on flow), and the spoke still holds no primary literature on either.

The corpus's own weak spot

Broken 2026-08-05, for the first time. kruger-dunning-1999 and magnus-peresetsky-2022 were fetched and read in full — the spoke's first two T1s and its first primary research of any kind. Everything below stood for eight days and describes the corpus that existed until then; it is kept because the pattern it names is still true of every other subject here, and because what the two papers changed is best read against it. The section that follows this one is what they changed.

Still true after the 2026-07-28 batch, and in some ways sharper. Nothing here has been read in the original. The founding three were second-hand; of the six that followed, three are Wikipedia articles about books and people, one is a Google Books catalogue entry (the-trouble-with-testosterone, the spoke's only T4 and its weakest page), and one is an author summarizing his own unread experiments. Encyclopedic sources are reliable about what a work says and how it was received and are not substitutes for the work.

Concretely: this wiki can now describe thinking-fast-and-slow's argument and its replication problems accurately, and has read neither the book nor a single one of the studies underneath it. The highest-value next sources remain primary — the 2006 Science income-happiness paper first, since the remembering/experiencing split does more work here than any other single result.

A second T4, and a new subject arriving at the weakest possible strength (2026-08-04). does-language-determine-thought opens language and cognition for the spoke — sapir-whorf-hypothesis, a subject with sixty years of experimental work behind it — and cites not one item of that work. It names Sapir and Whorf and no other researcher; its evidence is that languages carve up colour differently and differ in emotion vocabulary.

Worth separating what that costs from what it doesn't. The source is accurate on the two things it actually does: distinguishing determinism from relativity, and reporting which of them the field rejected. Those are the shape of a claim and the state of a consensus, and a popular overview is competent to relay both. What it cannot support is any specific finding — so the spoke now has a concept page whose central claim rests on nobody's data.

That makes it the cleanest instance yet of this section's point, and the first where a whole subject entered at that strength rather than a single book. The pattern is no longer "the corpus reads about primary work instead of reading it" but "the corpus can acquire an entire topic without touching its evidence base at all."

What reading two primaries actually bought

The spoke's founding complaint was that it described primary work instead of reading it. One subject now runs the other way, and the difference is worth recording precisely, because "get the primary sources" has been an article of faith here rather than a demonstrated gain.

Three things the secondary source could not have given, in ascending order of importance.

Detail. kruger-dunning-1999 is four studies with four stated predictions, not one correlation. Study 3's second phase — bottom- and top-quartile performers grading five peers' papers — is where the top-performer half of the effect gets its explanation, the false-consensus effect, and the finding that gives the whole paper its shape: the incompetent are wrong about themselves, the competent are wrong about other people. No popular account of this effect carries that sentence.

A correction to this wiki. Working from the secondary, this spoke wrote that the 2022 analysis showed the pattern arising "in data with no effect in it." magnus-peresetsky-2022 fits a censored model to 665 real exam predictions and argues the fit needs no psychological parameter. Close, and not the same claim. Corrected on the concept page with the change marked.

An argument neither source knows it is in. Kruger and Dunning anticipated the regression objection and built Study 4 against it: ten minutes of logic training moved the trained bottom quartile from an estimated 55th percentile to the 44th while an untrained control group did not move. A floor-and- ceiling artifact cannot produce that. Magnus and Peresetsky model the cross-sectional shape and never mention the experiment — and, in a smaller slip pointing the same way, state that top-performer underestimation was "not discussed in Kruger and Dunning (1999)" when it has a titled section there.

The rebuttal exists, in print, twenty-three years before the critique, and the critique does not engage it. No secondary source in this corpus reports that, and neither paper shows it alone — it becomes legible only when both sit on the same shelf, which is the wiki's whole method doing the one thing it claims to do.

What it cost: two fetches and one afternoon. Against eight days of describing this material at one remove. The founding weak spot was correctly diagnosed and the fix is cheaper than the section above implies — which is itself a finding about the corpus, and an argument for closing edges 1 and 2 on the remaining subjects rather than treating primary access as aspirational.

The second weak spot is not about strength of evidence but about who chose the reading list.

One essayist supplies four of the sources

maria-popova and the-marginalian were paged on the third ingest that came through them (jung-on-creativity, the-middle-passage, jung-on-death), because at three it stops being a coincidence and becomes a property of the corpus. in-the-absence-of-the-ordinary made it four the next day.

Her essays are the better end of T3 — she quotes at length, names her anthologies and dates the material, which is why this wiki has Jung's own sentences rather than a summary of them. The care needed is that the connective argument between the quotes is hers and is literary: in jung-on-death she reads Jung's spaceless, timeless unconscious through Einstein and the curvature of spacetime, and none of the Jung she quotes mentions spacetime. Her outlet has no editor and no fact-checker by design, so nothing sits between the framing and the page.

The subtler cost is selection, and the fourth source makes it hard to keep calling subtle. Four Popova pieces means four subjects Popova finds interesting — Jung twice, a Jungian analyst, a grief therapist — and every one of them lands on the humanistic, depth-psychological end of this spoke. None comes near the experimental end where replication-crisis lives. The corpus's shape is partly one reader's taste, which is worth knowing before reading the balance of these pages as a balance of the field. The correction is not to stop ingesting her; it is that the next few sources should come from somewhere else.

Two have, since: does-language-determine-thought (2026-08-04) and dunning-kruger-misread (2026-08-05). The second is the one that counts for balance — a tech newsroom arguing about a specific published result, which is the experimental end the Popova run never approaches.

Writing, twice, from two different disciplines

The corpus now holds two sources on why writing is hard, and they share no machinery at all.

why-writing-feels-hard says the brain reads exposure as social threat, and reaches for Eisenberger, Baumeister and transient hypofrontality. dysgraphia-via-dyslexia says writing is the mapping of spoken language onto marks, and that difficulty is inefficiency somewhere in that mapping. One account is affective and universal; the other is linguistic and clinical.

They are not in conflict, and the reason is worth stating because it is easy to miss: they are about different populations. Dávila addresses everyone who finds a blank page frightening; Hoh addresses children whose written output falls persistently short of their ability. A spoke that collected only one of them would have a tidier and worse picture of the subject.

What dysgraphia adds structurally is the first developmental disorder here, and with it the first case where the corpus's interest is in an intervention rather than an explanation.

The citation-chain pattern, third instance, with the weld in a new place. The named pattern above is a chain of real references terminating in a practice. This source has the chain — four cited works with volumes and pages, better sourcing than most popular writing here — and it terminates in a practice too: correct the foundational language processes rather than accommodate the child. But the untested weld is not the mechanism this time. Berninger's remediation research is real and predates the author. The weld is comparative: nothing cited establishes that correction beats accommodation for a given child, which is exactly the claim the piece is written to make.

And there is a feature the earlier instances did not have. Hoh is CEO of a remediation company and the article carries no disclosure. Accommodation is something a school gives away; correction is something a company sells. That does not make her wrong — but it means the corpus has now seen the clinician-with-a-service structure twice (anna-lembke the sharper case), and once at a venue, psychology-today, whose whole model is recruiting practising clinicians to write for the public. The pattern belongs to the venue as much as to the author.

A load-bearing premise nobody here has read. The argument is a transfer — use dyslexia's intervention evidence for dysgraphia — and this spoke holds nothing at all on dyslexia. That the dyslexia approach works is asserted by the interested party and unexamined here. A source on dyslexia remediation outcomes is now worth more to this corner than a second source on dysgraphia.

Jung, finally in his own words

carl-jung, analytical-psychology, second-half-of-life and jung-on-creativity were built entirely from essayists and encyclopedias — the spoke's largest body of second-hand claim. psychology-of-the-unconscious (1912, tr. Hinkle 1916, public domain) is the primary text, and it supplies three things the paraphrases had blurred.

The break with Freud is a definitional move, not an empirical one. Jung redefines libido as general life energy, "a hypothetical energy of life, comparable to Bergson's élan vital," with sexuality one channel among many. No observation separates the two positions; what separates them is how to individuate a theoretical quantity. That reframes the most-told story in depth psychology as a disagreement neither party could have settled by looking.

The physics analogy oversells. Jung compares libido to energy in physics, which is conserved and measurable; his is neither. The comparison imports a precision the concept lacks, and it is the clearest instance in this spoke of a metaphor doing argumentative work.

Myth and dream are one process. "The myths a mass-dream of the people: the dream the myth of the individual," resting on the split between directed thinking (verbal, logical, effortful) and fantasy thinking (archaic, image-based). Myth is not failed proto-science; it is dreaming at the scale of a culture.

And the method is now on the record rather than inferred. The book's central case analyses the published fantasies of Miss Frank Miller, a woman Jung never met, read for correspondences with myth, with nothing in the procedure that could have shown a reading wrong. Put next to kahneman-income-happiness-2006 and the measurement quarrels elsewhere here, the spoke now holds both evidentiary standards in primary form and can compare them directly instead of asserting that Jung's support is thin.

The second primary text, and what it does to the archetype

psychological-types (1921, tr. Baynes 1923, public domain) was read for its chapter of definitions — 57 of them, the lexicon Jung fixed his own vocabulary in. Three findings, and none of them is what the essayists set up.

The archetype is a minor term in 1921. It is not among the 57 headwords. It appears once, as a parenthetical alias inside the definition of Image: "The primordial image (elsewhere also termed the 'archetype') is always collective." The collective unconscious has no entry either; it is defined by implication, as the source primordial images draw on. The two concepts this spoke attributes to Jung as his central ones are, in the reference chapter of his most systematic book, a synonym and a background.

The inherited-symbol claim has a stated mechanism, and it is biology. "The primordial image is a mnemic deposit, an imprint ('engramm' — Semon)", and "the psychic expression of an anatomically and physiologically determined disposition". Richard Semon's mneme held that repeated experience lays down traces that are inherited; Jung takes that and adds an argument that the deposit cannot be a copy of the environment alone. The wiki has been calling this the untestable part of Jung. It is not untestable — it is a heredity claim resting on a mechanism biology afterwards abandoned, which is a harder verdict than vagueness. What Jung's method cannot test is not the existence of the disposition but the reading of any particular myth as its expression.

The evidence he offers for universality is one clinical observation, phrased in the language of 1921: mythological motives "demonstrated ... in the dreams and phantasies of thoroughbred negroes suffering from mental disorders". No procedure, no comparison, cross-referenced to his own earlier book. The pattern from Miss Frank Miller repeats exactly — the universality thesis is supported by the proposer's interpretation of his own cases.

Individuation comes out of the same chapter with its definition intact but its timing absent: definition 29 makes it "a process of differentiation, having for its goal the development of the individual personality" and says nothing about a life's midpoint, so second-half-of-life still runs on essayists. The 1934–54 archetype essays and the 1930 stages-of-life essay remain the second-hand part.

The third primary text dates the archetype, and softens it (research pass, 2026-08-28)

analytical-psychology-and-poetic-art (Jung's 1922 Zurich lecture, tr. Baynes, first published in England 1928, T1, read in full) is the target coverage edge 2a named, and it moves three things.

The archetype acquires a definition, six years after Types declined to give it one. Where 1921 had "archetype" only as a parenthetical alias inside Image, 1928 has it as a term with a stated content — "a figure, whether it be a daemon, man, or process, that repeats itself in the course of history wherever creative phantasy is freely manifested… essentially a mythological figure" — and a stated derivation from "the psychic residua of numberless experiences of the same type." So the concept's arrival in English is now datable from primary texts rather than assumed from the secondary ones.

The inheritance claim is weaker and better than this wiki has been repeating. The spoke's verdict after Types was a discredited heredity mechanism (Semon's engram). That mechanism survives here, joined by "anatomical formations in the very structure of the brain" — and then bounded: "It does not yield innate ideas, but inborn possibilities of ideas… categories of phantasy-activity, ideas a priori as it were, the existence of which cannot be ascertained except by experience." Inherited disposition, content from experience, in a Kantian frame. Two of this corpus's own pages gloss Jung as claiming inherited images, and on this text that gloss is wrong.

The circularity objection turns out to be Jung's stated method. "Only through the conclusion derived a posteriori from the perfected work of art are we able to reconstruct the primitive foundation of the primordial image." The spoke has been treating post-hoc reconstruction as the flaw it detected; the primary text presents it as the procedure. That does not settle whether the inference supports the claim — it does mean the disagreement is about inferential warrant, not about concealment, which is a better-posed dispute and a more interesting one.

And a dating correction, not an error correction. jung-on-creativity (T3) presents the psychological versus visionary pair as Jung's theory of creativity. The word "visionary" does not occur in these twenty-four pages and no two-mode distinction is drawn; the pair belongs to the later Modern Man in Search of a Soul essay Popova is reading. Both accounts stay, and what the corpus gains is a date: the two modes are not in the 1922 lecture, so they entered around 1930, and "Jung on creativity" is at least two positions eight years apart.

One sentence that belongs to a different corner of this spoke. Defending psychology's right to treat art, Jung turns the same knife on his own field: psychology's "specific value and essential quality is at once obliterated as soon as it is regarded as a mere brain activity, thus bringing it into line with other glandular activities as a mere subdivision of physiology. In actual fact, this depreciation has already occurred." Written in 1922, and it is the objection to Sapolsky's reduction that the free-will corner of this wiki has been making without a source. It is not an evidential objection and does not pretend to be.

The neuroscience corner gets its literature (2026-08-23)

Until today the behavioural neuroscience corner — the one this spoke's own CLAUDE.md describes as "what biology does to agency" — was the corpus's weakest by evidence, and it was weakest in a way the corpus had not said out loud. Sapolsky's thesis that behaviour has biological causes sat on determined (a Wikipedia summary) and the-trouble-with-testosterone (a bookseller's catalogue entry, the spoke's only T4). The claim was carried entirely by two accounts of books. The biology itself — a receptor, a gene, an enzyme, an experiment — appeared nowhere.

lewis-borrelli-reward-2021 is that biology: a peer-reviewed Springer chapter, read in full, on the dopaminergic reward system. It is the spoke's eighth T1 and its first from the life sciences, and it brings three pages with it (reward-system, dopamine, circadian-rhythm).

What it settles, and where

The chapter's value here is not that it proves Sapolsky right. It is that it lets the spoke stop describing a mechanism it had never read.

It puts a citation under the corpus's register finding. That finding — from dopamine-nation and anna-lembke — is that the same person can produce a peer-reviewed record and a bestseller their own field disputes, and the second is what a million readers act on. It was inferred from critics' reviews. Now both registers are held for the same subject in the same year:

The register difference is not a disagreement about facts. Both say social media engages the reward system the way gambling does. It is a difference in what gets dropped when the audience changes — and the dropped item is the hedge, every time.

And it inverts the corpus's usual complaint

The standing weak spot below says this wiki describes primary work instead of reading it. This source is the other failure mode, and worth naming so the spoke does not swap one for the other: a peer-reviewed review is still secondary. Nobody in this chapter measured anything. Its 141 citations make it checkable, which the essayists here are not, and checkable is not the same as independent — several of the D2R results run back through mouse lines that its senior author built.

Set against human-mouse-perceptual-maps, still the spoke's only experiment, the contrast is sharp. That paper tested humans and mice and reported the correspondence as its finding. This chapter tests nothing, states its confident claims from mouse knockouts, and states its claims about human life — phones, sleep, obesity — from five associational human studies. The confidence and the evidence run in opposite directions, which is the same shape as the reach-versus-evidence trade the founding tension names, now visible inside a single T1 source rather than between a paper and a paperback.

The subject is new to the whole hub

Nothing in any of the 36 spokes mentioned circadian rhythm or the suprachiasmatic nucleus before today. circadian-rhythm is a first, and it arrives with the claim worth keeping: reward is a zeitgeber. Feeding, exercise, sex, drugs and social contact do not merely happen on a schedule set by the clock; through dopamine they move the clock. If that holds, a large family of everyday advice — when to eat, when to look at a screen — is a claim about neuroscience and not about willpower.

Open questions

• How much of the heuristics-and-biases corpus survived replication? Partly answered (2026-07-28, thinking-fast-and-slow): the damage is documented at chapter granularity — Ch. 4 ("The Associative Machine") at a replicability index of 14, social priming worst hit — and Kahneman conceded it. What's still unanswered is the other side: which findings did hold. The corpus can now say what to distrust and not what to rely on, which is the less useful half. Prospect theory and the core judgment-under-uncertainty work are generally held to have survived; no source here establishes that. Advanced 2026-08-05: dunning-kruger-misread reports a 2021 Nature replication (~4,000 participants per study, two studies) coming down on the metacognitive side — this wiki's first named finding that held. Two caveats keep it from closing the question: the source names no author and no title for it, and the same source carries a statistical case that the dunning-kruger-effect would replicate whether or not it exists. So the corpus now has one survivor and a live argument about whether surviving proves anything. Advanced again 2026-08-12, and this time the verdict is fully documented. income-happiness-conflict-resolved (Killingsworth, Kahneman & Mellers, PNAS 2023) is an adversarial collaboration in which Kahneman took part in the reanalysis that revised his own satiation result: the plateau in experienced well-being holds only near the 15th percentile of happiness, up to about $100,000, and the original flat line came from 85% of high-income responses sitting at the scale maximum. Named authors, named method, a quantity that changed, and the disagreement published rather than settled by seniority. So the answer to this question now has three shapes in the corpus — a finding that broke (thinking-fast-and-slow Ch. 4), a finding that held (dunning-kruger-misread's 2021 replication), and a finding that was corrected to a narrower claim. The third is the most useful and the rarest.
• Does free-will scepticism change behaviour, and in which direction? free-will-belief-effects reports less hatred and more forgiveness; the older literature reported more cheating and aggression; the author's answer is that the older work didn't replicate. Both bodies of evidence are underpowered by current standards. Genuinely open, and consequential — determined argues for rebuilding criminal justice on the assumption.
• Does dopamine teach, or does it want? (added 2026-08-23) The two dominant accounts of what dopamine is for — Schultz's reward prediction error, and Berridge's split between wanting and liking — are absent from this corpus entirely. lewis-borrelli-reward-2021 cites Schultz once, for the accumbens, and Robinson & Berridge once, for sensitization, and engages neither theory; it is a chronobiology chapter and had no reason to. So reward-system and dopamine currently describe the anatomy and the molecule with no theory of the computation attached. This is the single largest gap the ingest opened. — needs: Schultz's primary work, or a review of the incentive-salience argument. Sharpened, not answered, 2026-09-19. tubbs-mind-after-midnight-2022 describes the nocturnal reward state as elevated dopamine anticipation with a blunted response on receipt. That is the shape of Berridge's wanting/liking split arriving from a paper that never names the theory — so the corpus now has a phenomenon the missing account would explain, and still no account. It also narrows the ask: of the two candidate theories, the one this evidence bears on is incentive salience, not reward prediction error.
• Which way does the obesity arrow point? (added 2026-08-23) Striatal D2R availability is lower in obese humans than controls, and a palatable diet lowers striatal D2R expression in mice. The chapter reports both and does not say whether reduced receptors follow the eating or precede it. The distinction decides whether "food addiction" names a cause or a consequence, and this corpus cannot currently tell.
• Is the mouse evidence transferable? (added 2026-08-23) Almost every mechanism in lewis-borrelli-reward-2021 is a knockout or a lesion in mice, and its human-facing conclusions are not. human-mouse-perceptual-maps is the corpus's one case where cross-species transfer was tested rather than assumed, and it held for a perceptual map. Whether it holds for reward circuitry is a separate question no source here addresses.
• Is being awake at the wrong hour different from being tired? (added 2026-09-19) The whole content of mind-after-midnight-hypothesis is that circadian phase and sleep loss contribute separately to the nocturnal risk state. No study in the literature the paper reviews separates them — most measured executive function during the day after a sleep-deprived night, which conflates the two by construction. mind-after-midnight-nimh-grant is designed to pull them apart, with prolonged wakefulness and abrupt 2am awakening as separate arms, and it will take five years. Until then this corpus cannot say whether the hypothesis names a distinct mechanism or a re-description of sleep deprivation. — needs: the result, or any study measuring at the circadian hour in question.
• Is "world model" one thing or two? perceptual-learning uses the phrase for a perceptual similarity map; the AI spokes in this hub use it for a learned generative model of an environment. Currently an analogy on both sides.
• What is in ACT's 900 trials? (added 2026-07-29) acceptance-and-commitment-therapy arrives with the largest evidence base in this wiki and the corpus cannot assess a single trial of it. The known failure modes of psychotherapy research are exactly the ones a raw count hides — researcher allegiance (the developer is frequently an author), small samples, and what the control condition was. Note the asymmetry in how this spoke has been treating numbers: it discounted thinking-fast-and-slow's citations on replication grounds and is currently crediting ACT's on trial-count grounds, from a Wikipedia summary in both cases. The highest-value next source is a meta-analysis or a critical review of ACT — and if it turns out the count is doing more work than the trials, that is the same finding as the Ch. 4 replicability index, one discipline over.
• Does the "complete the stress cycle" mechanism have a literature? (added 2026-07-29) burnout-stress-cycle asserts that physical activity discharges an incomplete physiological stress response and that this prevents burnout. The claim is specific enough to be testable and no source here names a study. Its structural half (Human Giver Syndrome, from Kate Manne) is a moral-economy argument and would not be settled by one.
• What happened to developmental trauma disorder? (added 2026-07-29) van der Kolk proposed it for children with chronic early adversity, and no source here says whether any diagnostic manual took it. That matters more than the usual missing fact: a rejected proposal and an accepted one imply very different readings of the whole book, and the 122-claim audit doesn't cover it either way.
• Does the borrowed vocabulary flow back? System 1/2 has travelled into AI system design (../defensive-security-wiki's soc-fast-slow-thinking). Whether anything learned by building those systems constrains the psychology, or the traffic is one-way, is an open and testable question.
• What does the clinical literature on grief say? (added 2026-08-03) grief entered the wiki through a poetic account that engages none of the research on its own subject. Three specific holes: prolonged grief disorder as a formal diagnosis (which the "grief is not pathology" framing would have to argue with), whether the famous stage model survived scrutiny, and the process models that describe what mourners actually do. The second is the same question this spoke already asks of thinking-fast-and-slow and the-body-keeps-the-score, on a framework with even wider public circulation.
• Does "welcome the symptom" work? (added 2026-08-03) Three schools give the instruction (the-middle-passage, existential-psychotherapy, in-the-absence-of-the-ordinary) and none offers evidence for it. acceptance-and-commitment-therapy is the nearest testable relative, with 900+ trials behind a similar stance. Whether that literature actually bears on the depth-psychological version, or merely shares its vocabulary, is the connection this corpus most needs and does not have. Sharpened the same day (feeling-good): the instruction now has an explicit rival inside the wiki rather than only an absence of evidence — CBT says correct the thought, not welcome it.
• CBT or ACT? (added 2026-08-03) The two testable therapies here prescribe opposite operations on the same object and both cite large trial literatures. Head-to-head comparisons and dismantling studies are what would settle it, and the wiki holds none — the same shape as the unassessed 900 trials, now doubled. Highest-value next source in this corner: a comparative meta-analysis, or Beck's own work rather than a book about a book about it (aaron-beck is paged from one encyclopedia article and deserves better).
• What is Jung's status now? Enormously influential, empirically unsupported. Whether contemporary psychology has salvaged anything specific from analytical psychology, or routed around it entirely, is unsourced.

A case where nothing failed to replicate (quality cycle, 2026-08-22)

replication-crisis has been carrying an argument with no worked example under it. The first one this spoke has acquired is debunking-the-stanford-prison-experiment, and it does not fit the shape the crisis usually has.

The standard story is statistical: an effect was real in one sample and vanishes in the next, and the fixes are pre-registration, power and open data. Le Texier's archives describe something else. The guards were briefed on the wanted result and were not told they were subjects; 11 of the 17 rules came from a class experiment three months earlier that went uncredited for decades; the conclusions were drafted in advance. There is no noisy finding here to shrink toward zero, which means the corpus now holds three distinct failure modes rather than one: a finding that shrinks under scrutiny (the Dunning-Kruger literature, magnus-peresetsky-2022), a finding that was misread by its later citers (dunning-kruger-misread), and a finding that was manufactured.

The second half of Le Texier's paper is the part that connects to this spoke's oldest theme. Fromm published ethical and methodological objections in 1973, two years after the study, and textbook authors went on teaching the uncriticised version for four more decades. Whatever the discipline's error-correction machinery is, in this case it was not the literature — the correction arrived when somebody read the boxes. That is a distribution failure, and it is the same failure the spoke keeps recording in a smaller form every time a claim here rests on an essay about a paper rather than the paper.

A fourth kind of replication (quality cycle, 2026-08-30)

The section above sorted replication failures into three shapes: an effect that shrinks, a finding misread by its citers, and a finding that was manufactured. dtd-field-trial-replication adds a shape that is not a failure at all and is still not what the word usually promises — a team replicating itself.

Ford, Spinazzola and van der Kolk proposed developmental trauma disorder, wrote its structured interview, ran the field trial, and then ran it again at six new sites with 271 children. The result holds: DTD without PTSD carries 5.2 comorbid diagnoses against 3.8 for PTSD without DTD, and three associations survive controlling for PTSD and everything else. That is real generalization across geography and treatment setting, and it is exactly the kind of work the spoke keeps complaining nobody does.

It is also the same people. The instrument, the criteria and the professional stake are constant across both runs, so what has been demonstrated is that the finding travels between sites, not that it travels between investigators. The distinction matters more here than usual because the diagnosis was rejected from DSM-5 for insufficient evidence — a self-replication is precisely the evidence that objection anticipates, and precisely the evidence it does not settle.

The reflexive note for this spoke: psychological-trauma had carried "no source here establishes its formal status" for a month. Closing it took one search. The sentence was not wrong when written; nobody had looked.

The argument had a second round, and both sides published it (quality cycle, 2026-09-01)

The spoke has carried the Dunning-Kruger dispute for a month on three sources: the 1999 paper, a 2022 censored-model critique, and a trade piece summarising the fight. The 2002 exchange — the round where the statistical objection was first made formally and first answered — was named in the backlog and absent from the corpus. Both halves are now here, and reading them changes the shape of the dispute rather than the score.

What the critics did. krueger-mueller-2002 is a successful replication used as a weapon. The pattern reappears; the explanation is what they attack. Seven of eight unbiased mediator tests move the correlation by .01 or less, and subtracting the better-than-average effect turns 1999's four asymmetries (33, 32, 32, 29 percentile points) into 1, 10, −4 and 7. The finding survives, its mechanism does not.

What the authors did. kruger-dunning-2002-reply concedes measurement error contributes and then makes an argument the corpus had not seen: an artifact cannot answer a manipulation. Ten minutes of logic training moved the trained bottom quartile from 36 percentile points of overestimation to 17, and untrained controls did not move. No regression slope changes because somebody read a packet.

Three things worth carrying beyond this topic.

A correlation is not an accuracy measure. The reply's weather forecasters are the cleanest statement of it in the corpus: Rob's forecasts correlate perfectly with the temperature and are wrong by 48 degrees on average; Laura's correlate worse and are wrong by 7. Krueger and Mueller measured metacognitive skill as a within-person correlation, so on this argument they measured Rob and called him skilled. That objection travels — measuring-annotator-agreement in ../machine-learning-wiki is the same complaint about agreement statistics, arrived at from a different field.

An unreliable test hides a real effect and fakes a dead one. The critics' difficult test had a split-half reliability of .17, and their own re-test figures say a participant had a .32 chance of landing in the same extreme quartile twice. A "bottom quartile" that unstable is partly a group of people having a bad twenty minutes, which cuts both ways: it weakens the metacognitive reading of their own null and weakens their correction as evidence against Study 4, whose test ran at .93.

Neither side claimed the win. The reply ends by refusing to dismiss the regression account, saying more work is needed and that which explanation holds "remains to be seen" — and then notes the irony that both camps are confident and one of them must be wrong: "Whoever is wrong, they do not seem to know it." Twenty-four years later magnus-peresetsky-2022 is running the same argument with a better model, which is the strongest evidence available that the 2002 modesty was the correct posture.

This is the third distinct shape the spoke now holds for "is a famous result real": a finding that broke (thinking-fast-and-slow Ch. 4), one that was corrected to a narrower claim (income-happiness-conflict-resolved), and now one where the data are agreed and the explanation is not, with the disagreement conducted in public by name. That third shape is the healthy case, and it is the one the popular account of this effect makes invisible.

The floor of the evidence ladder, and what sits on it (2026-09-01)

This wiki has spent its life sorting sources by how far a claim is from the study behind it — dysgraphia-via-dyslexia with four references carrying volumes and pages, handwriting-vs-typing-notes naming a meta-analysis and its effect size, dunning-kruger-misread with three of six citations missing an author, does-language-determine-thought citing nobody past Sapir and Whorf. musical-expression-neuroscience establishes what the bottom of that ladder looks like: no author, no outlet, no date, and not one reference, with five appeals to an unnamed literature ("Cognitive neuroscience shows", "Neuroimaging studies demonstrate", "Physiological research proves", "Empirical data shows", "Controlled studies indicate") in roughly 500 words.

The instructive part is not that it is unsourced. It is which claims get more specific as the sourcing gets weaker. The document's vaguest attributions carry its most precise assertions: a named regional dissociation between mechanical and expressive playing (motor cortex and cerebellum against anterior cingulate and inferior frontal gyri), and a controlled result that AI feedback produces "significantly faster" development of expressivity than traditional teaching. Compare dunning-kruger-misread or handwriting-vs-typing-notes, where the citation layer constrains what the writer can assert — a named meta-analysis comes with g = 0.248 attached, and that number caps the claim. Remove the layer and nothing caps it. The corpus had been treating citation density as a measure of a source's reliability; this one suggests it also works as a brake on a source's ambition, and the brake is what actually goes missing first.

It also marks a genuine hole rather than filling one. The psychology of musical performance and expertise has no page here — not perceptual-motor expertise, not deliberate practice, not the performance-anxiety literature — and this document is a list of the places to go look, not a source to build on. Nothing in it should be cited without a primary underneath.

A fifth shape, and the spoke's first source about groups (2026-09-05)

polarization-connectivity-model arrived by Telegram as a trade write-up and turned into three pages. It is worth recording twice over: once for what it does to the is a famous result real thread above, and once because it is the first source this spoke holds about collective behaviour rather than individual minds.

The dispute shape, at speed. The section above records three shapes — a finding that broke, one corrected to a narrower claim, and one where the data are agreed and the explanation is not. This is a fourth and it is the fastest: a 2025 PNAS paper drew two independent critiques and two authors' replies in the same venue inside about a year (polarization-connectivity-dispute). The two critiques do not overlap. Górski et al. dispute the micro-mechanism and, as the reply notes, "do not question the central finding … that polarization undergoes a phase transition at critical connectivity." Völker, Corten and Mollenhorst dispute the empirical premise — whether close friendships doubled at all — which is the load-bearing input, not a detail of the model.

And then both sides say the premise is unestablished. The critics: such a rise "would be a remarkable finding, as no such trend has been documented in the extensive sociological literature on personal networks." The authors: "Yes, the state of evidence on the scale and direction of how friendships change is lamentable. The only way forward is to collect more evidence." A contested claim whose disputed input is named in the same words by both parties is the healthiest state this spoke has seen a live argument reach, and it is a different thing from agreement.

The tertiary layer failed again, on schedule. psypost-polarization-model, four days before this was written, reports the model as an explanation and mentions none of the exchange, and states a threshold ("four and a half close connections") that the paper's own continuous transition cannot support. That is the second instance of the defect first recorded for dunning-kruger-misread, now in a different decade and a different subfield: the secondary layer preserves the finding and drops the argument around it, which is precisely the information a reader needs to calibrate belief.

What the spoke should be honest about: this is a boundary case. The paper is statistical physics — a Metropolis algorithm on a small-world graph, published by a complexity-science group — and this spoke's own note says to decide by method, argument versus experiment or clinic. Simulation is neither. It routed here on subject: the model's two ingredients, homophily and Heider's social balance, are social-psychology results, and its output is a claim about how people's opinions form under social pressure. That is the closest fit among 36 spokes and it is not a comfortable one. If two more computational-social-science sources arrive, the question of whether the corpus needs a place for them should be reopened rather than settled by precedent from this one ingest.

A parameter worth watching, on the wiki's own terms. The model's fit to the Pew curve rests on β = 2.7, sensitivity to social stress. The paper cites prior work measuring β between 1 and 2 for car brands and investments, and sets it higher for politics on a stated hypothesis that people feel political belief more keenly. That hypothesis is a psychology claim, it is the hinge of the empirical result, and nobody in the exchange tests it. It is the most tractable open question the source leaves.

The citation layer that carries nothing (2026-09-07)

musical-expression-neuroscience set the floor by having no citations, and the lesson drawn from it was that the citation layer works as a brake on a writer's ambition — remove it and nothing caps the claim. generational-boundaries-yourtango is the same finding inverted, and it is the version a reader will meet far more often.

The links are all there. Six of them, under a piece with no survey, no sample and no named researcher: a Johns Hopkins domain, two Psychology Today posts, a "clinically reviewed" vendor blog. Three were chased.

• The Johns Hopkins link, attached to "millennials are the first to make setting boundaries a priority," is a student careers blog about workplace values that never mentions boundaries.
• The Psychology Today link, attached to the claim that Gen Z "control other people's behavior under the guise of boundaries," is a licensed therapist's generation-neutral argument that rigid boundaries isolate the person setting them, citing three peer-reviewed works. It contains no generational claim.
• The Truity link is a personality-test company's blog post with no data in it.

So the brake is not citation presence. It is whether a citation carries a measurement that the writer then has to live with — handwriting-vs-typing-notes cannot inflate past g = 0.248 while the number is on the page. Six links to sources that measure nothing constrain nothing, and they buy the appearance of exactly what is missing.

And the defect has a new degree. dunning-kruger-misread and psypost-polarization-model each show the secondary layer preserving a finding and dropping the argument around it. This one adds a claim its source does not make: a clinical caution addressed to everybody arrives as a fault belonging to one cohort. Preserving-and-flattening is a loss of information; this is manufacture, and it is harder to catch because the manufactured version is the more quotable one.

What the spoke should not do with it. Nothing here says whether generational differences in boundary-setting are real. The piece is a specimen of transmission, not evidence about cohorts, and the gap it exposes is registered as edge 12 rather than answered.

The clock arrives at the clinic (2026-09-19)

circadian-rhythm entered this wiki in August as molecular chronobiology — clock genes, the SCN, dopamine made and cleared on a schedule — and the note then was that no spoke in the hub had mentioned circadian rhythm before. tubbs-mind-after-midnight-2022 attaches suicide, overdose and impulsive decision-making to the same clock, and in doing so connects two corners of this spoke that had not touched: the neuroscience corner and the clinical one.

The claim. Being awake during the biological night is a risk state in its own right, not merely a tired one. Four rhythms stack: positive affect bottoming out at 1am–4am while negative affect climbs, attention narrowing onto negative stimuli, striatal reward activation distorting into high anticipation and blunted receipt, and behavioral inhibition falling through the night. Set out on mind-after-midnight-hypothesis.

What makes this worth the spoke's attention is the shape of the evidence, not the conclusion. The corpus has spent two months sorting sources by what they rest on, and this one splits cleanly down the middle:

• The epidemiology is real and adjusted. Threefold suicide risk midnight–6am, eightfold among veterans midnight–3am, 4.7-fold opioid overdose risk at night while heroin use falls. The authors insist on the adjustment and explain why it matters: unadjusted counts by hour mostly measure how many people are awake. That is the same correction this wiki applauded in income-happiness-conflict-resolved — a number that changes when you ask what the denominator was.
• The mechanism is assembled, not measured. Every piece of it comes from studies run at other hours for other purposes, and the authors say so: most executive-function results were collected during the day after a sleep-deprived night, which is the wrong time to measure a claim about the circadian night. Their own summary is that "all aspects of this hypothesis require empirical validation."

A fourth position on the evidence ladder. This spoke has recorded three shapes so far — a finding that broke, one that held, one corrected to a narrower claim. This is a fourth: a hypothesis whose authors name the experiments that would kill it, and state that none has been run. It is not a weak finding. It is a well-specified conjecture, which is a different object, and the corpus needed a slot for it.

And then the experiment got funded. mind-after-midnight-nimh-grant records a five-year, $3.8M NIMH study at the university-of-arizona, led by two of the paper's own authors, whose two arms are exactly the confound the paper complained about: subjects kept awake to 2am in one visit, allowed to sleep and woken at 2am in the other. The spoke has never before held a case where the stated gap and the protocol closing it are both on the record. Whether the result arrives, and whether it confirms anything, is the open part — but the hypothesis is currently in the rarest position in this corpus: specified precisely enough to fail, and being given the chance.

The popular layer behaved, mostly. mind-after-midnight-sciencealert is the best T3 this spoke has ingested. It links every claim to its primary — which is the only reason the paper could be read — carries the authors' central caveat instead of dropping it, and separates its subject from ordinary 3am waking. What it loses is the article type (a mini review, never mentioned), the venue's name (Network Psychology for Network Physiology), and the authors' tentativeness, under a "scientists warn" frame. Against dunning-kruger-misread, psypost-polarization-model and generational-boundaries-yourtango, the defect here is the mildest kind: framing inflated, argument intact.

The one thing it did quietly that is worth naming. The article is a re-dated republication — stamped 20 September 2026 over an image path from October 2025 — with no disclosure. The hub's rule says record the date the source states, so that is what the page carries, with the discrepancy written down beside it. A corpus that dates sources by what they claim needs to notice when an outlet re-stamps old copy, because staleness becomes invisible the moment republication looks like publication.

Effort, voice and the cost of outsourcing writing (2026-09-27)

the-joy-of-writing is a Psychology Today piece by a credentialed cognitive neuroscientist, james-danckert, arguing that handing writing to an LLM costs three things: homogenized output (regression toward a training-data mean), the reward of effort itself, and the development of a personal voice. It gave the spoke its first cognitive-miser page and its clearest instance yet of this corpus's citation-chain practice applied to a live, contested question rather than a settled one.

The chain checked out unevenly, which is the finding. Three citations, three different outcomes:

• Unverifiable. The homogenization claim rests on a Bieleke & Wolff PsyArXiv preprint (wzveh_v1) whose OSF page renders client-side and returned no text on fetch. Recorded as unverified, not false — maik-bieleke and wanja-wolff both have real, independently confirmed work adjacent to the claim (boredom and self-control, co-authored with Danckert elsewhere), so the claim is plausible on priors even though this spoke could not read the paper itself.
• Mis-cited but not fabricated. The "cognitive misers" link goes to Kool, McGuire, Rosen & Botvinick (2010) — a demand-avoidance experiment — not to Fiske & Taylor's 1984 paper that coined the term. The idea invoked is real; the citation attached to it is one step off. See cognitive-miser.
• Accurate and undersold. yakura-llm-human-speech-2024 is the strongest citation in the piece — a synthetic-control analysis of 737,083 hours of podcast speech plus a preregistered N=496 experiment, both showing chatbot vocabulary entering active human speech, not just passive exposure — and the article flattens it to "AI buzzwords," which undersells a result with an actual causal design behind it.

The tension the page had to hold rather than resolve. cognitive-miser and michael-inzlicht's effort paradox point in opposite directions on the same question: is effort something minds default to avoiding, or something that can itself be the reward? The Danckert piece leans on the first to argue against outsourcing writing, then leans on the second (Inzlicht's Trends in Cognitive Sciences work) to explain what is lost when the effort is skipped. Both traditions are real and both are cited correctly here; the corpus records the tension rather than picking a side, per the hub's contradiction rule.

A disclosed conflict of interest, recorded rather than hidden. Danckert states in the piece that his and john-eastwood's co-authored book, Out of My Skull, was used by anthropic to train models without permission, and that he expects to benefit from a settlement. That does not make the argument wrong — the citation-chain check above stands on its own — but it is relevant context for an argument that AI use degrades human writing, and the page records it plainly.

A normative counterpart to the biases literature (2026-10-04)

algorithms-to-live-by (Christian & Griffiths) is the corpus's first source that supplies a standard for decisions rather than a catalogue of departures from one. heuristics-and-biases asks where judgment errs; this book asks what a computer scientist would do under the same limits and answers with optimal-stopping, explore/exploit and Bayesian rules. Read only through a stub and a publisher site (T3), so the page claims no findings. The open question it adds: whether following such rules improves real human outcomes, which nothing held here tests. It leans toward cognitive-miser's reading that shortcuts can be rational, and against any reading of System 1 as plain error.

A new corner: the neuroscience of altered consciousness (2026-10-07)

lsd-cortical-dynamics-meg is the spoke's first source on psychedelics, and it opens rather than closes a question: a weaker brain rhythm and a faster one look identical under a standard band-power analysis, so a drug's effect on "desynchronization" can be measured wrong by the most common method in the field. The finding sits in the behavioural-neuroscience corner by method — direct measurement of brain activity, not argument — but names no open question this spoke already held. One source does not make a topic; it is recorded here so a second source on consciousness, psychedelics or EEG/MEG method has somewhere to land instead of starting a duplicate thread.

A Russian-language lineage beside the Kahneman tradition (2026-10-08)

professional-deformation (professional-deformation-wikipedia-ru, T2) is the spoke's first concept from the Russian and Soviet psychological tradition: Sorokin, then Gellerstein, Rogov, Markova, Zeer, Druzhilov. It is filed as a cognitive bias, but it does not belong with heuristics-and-biases. That programme explains single judgments by a mental shortcut and tests them in experiments; this one describes a trait-level change built by years in one occupation and is supported here by classification and example only. It is a cross-tradition adjacency to track, not a duplicate. Two seams are worth holding: it is not burnout-stress-cycle (shaping vs depletion, though the source lists emotional burnout as one of its forms), and it is not "professional destruction," which the source keeps apart. Gap-relevance: it extends the judgment-and-bias corner sideways, toward organizational psychology. It does not advance any open question; the source holds no study, and no Russian-language primary has been read. A second source from this lineage, ideally a Rogov or Zeer text, would let the taxonomy be checked rather than relayed.

Extended 2026-10-08 by professional-deformation-wikipedia-en (T2), which brings in a second, Western sociological path to the same concept: Warnotte, then Merton's 1957 grouping with Veblen's "trained incapacity" and Dewey's "occupational psychosis," plus Carrel and Pullum. The two articles agree on the core idea and differ in vocabulary and lineage, so they are held apart on the page, as with the philosophy traditions elsewhere in the hub. Flagged provenance conflict: the Russian article says pitirim-sorokin coined the term; the English one says Warnotte "probably" did, Sorokin "possibly," and marks the Sorokin claim citation needed. Both are kept, unresolved; neither article cites a primary. This is the sharper reason a primary (Warnotte or Sorokin) is the next source worth finding. The English article's related concepts (law of the instrument, bias blind spot, marketing myopia) are link-only and were not paged.

Carrel quotation upgraded to primary, 2026-10-08. The English article's one illustration, Alexis Carrel's line about specialists, was checked against the book (man-the-unknown-carrel-1935, T1, the spoke's Wikipedia-to-primary move as with Kant in philosophy-wiki). The wording holds. The context did not travel: Carrel wrote it in a 1930s argument that physics and chemistry had outrun physiology, psychology and sociology, leaving a society "built at random," and "professional bias" is a passing phrase in an inventory of man's many aspects, not a named concept. So the quote shows the idea was in circulation in 1935; it does not date the term or settle the Sorokin-Warnotte conflict. The same book argues for eugenics, which is why alexis-carrel is held as a person page with that noted: the quote is sound, the authority behind it is limited and period-bound.

Growth edges

Ranked; each names the kind of source that would close it (see ../QUALITY.md → Growth edges).

1. First T1 — closed 2026-08-05 by kruger-dunning-1999 and magnus-peresetsky-2022, both read in full. The spoke went from zero peer-reviewed sources to two in one batch. What replaces it: primary sources are now the norm to hold this corpus to, not an aspiration. Every subject below is the same edge with a different topic attached.
2. Second-hand accounts dominate everywhere else — half closed 2026-08-12. The named next source was read: kahneman-income-happiness-2006 (T1, in full), and with it income-happiness-conflict-resolved (T1), which revises it. The Kahneman/Tversky corner now has two primaries under it and focusing-illusion no longer rests on an essay about the paper. Jung half closed 2026-08-13 (research pass) by psychology-of-the-unconscious (Jung 1912, tr. Hinkle 1916, T1, public domain) — the corpus can now quote Jung on libido, on directed vs fantasy thinking, and on the myth/dream identity from Jung, and the Freud break has a text and a date. Jung half advanced again 2026-08-21 (research pass) by psychological-types (1921, tr. Baynes 1923, T1, public domain): the archetype, the primordial image, individuation, anima and projection are now quotable from Jung's own lexicon, and the reading changed — "archetype" is not one of the 57 definitions, the inherited-image claim rests on Semon's engram, and definition 56 denies the metaphysical reading this spoke had given it. Not closed. The 1920s–30s work proper is still absent, so second-half-of-life and jung-on-creativity remain essayist-sourced. — needs: the archetype essays (CW 9i, 1934–54) or a historian of psychology working from the primary texts. 2a. The two Jung pages the primary texts did not reach — new 2026-08-21, the residue of edge 2. second-half-of-life rests on the 1930 stages-of-life essay and jung-on-creativity on Modern Man in Search of a Soul (1933); neither is in the spoke and both are in copyright. One target is identified and cheaper than either: "On the Relation of Analytical Psychology to Poetry", in Contributions to Analytical Psychology (tr. Baynes, 1928) — the earlier printing of the creativity argument. HALF CLOSED 2026-08-28 (research pass). That volume is in, read in full: analytical-psychology-and-poetic-art (T1, Baynes tr., first published in England 1928; the translators title it "Poetic Art", not "Poetry"). It supplies the creativity mechanism the paraphrase omitted — the autonomous complex, defined operationally — the archetype's first real definition in English, and a materially weaker inheritance claim than this wiki had been repeating ("not innate ideas, but inborn possibilities of ideas"). It also dates the psychological/visionary pair out of the 1922 lecture, so jung-on-creativity is describing a later position. Successor, unmet: second-half-of-life is untouched — the stages-of-life timing claim is not in this volume, and individuation is still defined here without a timing claim, exactly as in Types. Two Jung texts now agree on that silence, which makes the 1930 essay the specific thing missing rather than a general gap. — needs: the 1930 stages-of-life essay, or a scholarly history quoting it at length. Not eligible again before 2026-09-11.
3. Clinical psychology is in the domain and thin. — needs: T1/T2 clinical source. The ACT-meta-analysis question under Open questions is the sharpest version.
4. The rest of the Dunning-Kruger literature — narrowed 2026-08-05 now that the two ends are held. HALF CLOSED 2026-09-01 (quality cycle): the 2002 exchange this edge asked for first is in, both halves read from the PDFs — krueger-mueller-2002 and kruger-dunning-2002-reply, T1, printed four pages apart in the same issue of JPSP. The answer to the question the edge posed: the earlier round did not ignore Study 4 the way magnus-peresetsky-2022 does. Krueger and Mueller left the experiments alone and attacked the mediation; Kruger and Dunning replied on the experiments and on the reliability of the tests used against them (Study 4 at Spearman-Brown .93, still ~40 percentile points of bottom-quartile overestimation after the critics' own correction). Residue, unmet: Burson/Larrick/Klayman (2006) and Gignac & Zajenkowski (2020) are still cited here at second hand, and the 2021 Nature replication dunning-kruger-misread rests on is still unnamed — the strongest positive evidence in the corpus for a finding that held, and the one item a reader cannot go and check. — needs: the Nature replication identified by author and title, then Burson (2006). Not eligible again before 2026-09-15.
5. Experimental evidence for linguistic relativity — new 2026-08-04. The spoke acquired sapir-whorf-hypothesis from a source citing no study at all, so the whole subject sits on a consensus report with nothing under it. This is edge 1 in miniature and the easiest instance to close, because the literature is large and the specific claims are well known. — needs: primary experimental work on language and categorization (colour perception, spatial frames of reference, grammatical gender), T1 preferred; a review or meta-analysis would close it in one.

Coverage edges (added 2026-08-08, at the curator's request for a wider backlog). These widen what the spoke covers instead of answering an open question above; one ordinary solid source closes any.

6. The Big Five. The best-validated model of personality is absent while analytical-psychology and carl-jung carry several pages between them, which leaves the spoke's personality coverage resting on the interpretive tradition alone. — needs: a T1 review of the five-factor evidence.

7. Attachment and development. Bowlby and Ainsworth are assumed by the trauma pages (the-body-keeps-the-score, psychological-trauma) and stated by none. — needs: a review article or the original longitudinal work.

8. The classic social experiments, with their replication status attached. PARTLY CLOSED 2026-08-22 — debunking-the-stanford-prison-experiment (Le Texier, American Psychologist 74(7) 2019, T1, read in full) supplies the Stanford prison case from the archives, and it turns out not to be a replication story at all: the original was staged, so there is nothing to fail to reproduce. Milgram and Asch are still absent, and they are the more interesting pair now, because Milgram has been partially replicated (Burger 2009) and Asch has a conformity-rate literature. — needs: Milgram (1963) and Asch (1956) in the original, plus the modern re-runs. 8a. The other side of the Stanford dispute — new 2026-08-22. This wiki holds Le Texier's account and notes that Zimbardo published a June 2018 response which Le Texier says does not address most of the findings. Nobody here has read the response. — needs: Zimbardo's reply read directly, or a third party who has weighed both.

9. Memory. The spoke covers thinking, bias and therapy, and not remembering — encoding, retrieval, or Loftus on false memory. — needs: a T1 review or the original eyewitness studies. Partly addressed 2026-10-07 by ai-cognitive-science-origin (T3): the connective claim that AI hallucination is the same reconstructive-memory failure mode Loftus documented in people, cited by name only. The primary Loftus literature and a review of encoding/retrieval are still absent.

10. Is political belief held more strongly than other belief, and by how much? New 2026-09-05, from polarization-connectivity-model. That model's fit to two decades of US polarization data turns on β = 2.7, a sensitivity-to-social-stress parameter set above the 1 to 2 range prior work measured for car brands and investments, on the authors' stated hypothesis that political opinions are felt more keenly. Nobody in the PNAS exchange tests it — the critics went after the connectivity data instead. This is a psychology question holding up a physics result, and it is measurable: the same online group-experiment paradigm that produced the 1–2 estimates could be run on political items. — needs: an experimental estimate of opinion-updating sensitivity for political versus consumer beliefs, T1/T2. The cheapest genuinely open question the spoke has acquired this month, and the only one where this wiki's own domain is the missing input to somebody else's model.

11. Social psychology, which this spoke does not have. New 2026-09-05, and it is a scope question as much as a sourcing one. Three pages on polarization arrived this week and they are the spoke's first sources about groups rather than individual minds — the six corners in the domain header (judgment, depth psychology, perception, clinical, neuroscience, psycholinguistics) contain no social corner, and 85 pages contained no conformity, no group dynamics, no intergroup contact. Either the spoke owns collective behaviour and should say so, or these three pages are a boundary case that belongs elsewhere. — needs: two more sources, then a decision. Do not settle it by precedent from one Telegram arrival.

12. Do generational categories measure anything? New 2026-09-07, from generational-boundaries-yourtango. Boomer / Gen X / millennial / Gen Z framing is one of the most-used explanatory devices in popular psychology and this corpus holds not one source on whether birth-cohort labels survive measurement — whether apparent generational differences separate from age and period effects at all, and what the design would have to be to tell them apart. The spoke cannot currently say either that the four sketches in that listicle are wrong or that they are right. — needs: a peer-reviewed treatment of the generations construct (the age–period–cohort identification problem, or a critique of generational research in organizational psychology), T1/T2. Cheap to close and it would arm the spoke against a whole genre rather than one article.

13. Sleep, and the wiki's first source on it is about being awake. New 2026-09-19, from tubbs-mind-after-midnight-2022. The spoke now holds a hypothesis about nocturnal wakefulness and nothing at all about sleep itself — no sleep architecture, no memory consolidation, no insomnia literature, and no evidence on what sleep deprivation does to cognition measured at the right hour. That last one is load-bearing rather than merely absent: the Mind After Midnight argument turns on separating circadian phase from sleep loss, and this corpus cannot currently characterise either term. note-taking-and-learning and perceptual-learning both sit next to a consolidation literature the wiki does not have. — needs: a T1 review of sleep and cognition, or of sleep deprivation's neurobehavioral effects; the insomnia/CBT-I evidence base would also connect it to cognitive-behavioral-therapy.

14. Suicide risk as a subject, not a statistic. New 2026-09-19. Three numbers about suicide timing entered the corpus with this ingest and there is no page behind them — no epidemiology, no prevention literature, nothing on what the field takes to be modifiable. A spoke holding psychological-trauma, grief and existential-psychotherapy has a clinical corner where this belongs, and fabian-xose-fernandez's claim that nocturnal risk is a modifiable target is a testable proposition the wiki cannot currently situate. Handle with the care the subject deserves. — needs: a T1 review of suicide epidemiology or of prevention-intervention evidence.

Contradictions / tensions

The founding note stood for a while: no factual conflicts, because the sources addressed different layers, and the live tension was methodological — a corpus whose members disagree not about findings but about what would count as one. That is still the deepest one.

Jung says the unconscious is not a metaphysical concept; this wiki says it is

carl-jung and analytical-psychology both describe the collective unconscious as a metaphysics-of-mind claim, and use that to place Jung on the seam with ../philosophy-wiki. The primary text refuses it. Definition 56 of psychological-types opens: "The concept of the unconscious is for me an exclusively psychological concept, and not a philosophical concept in the metaphysical sense" — a "psychological boundary-concept", justified "purely and solely from experience, and in particular from psychopathological experience", with speculation about its physical substrate waved off as "phantasies".

Both readings stay. They are not the same claim: Jung is stating what he intended the term to do, the wiki is stating what the evidence behind it can carry. A boundary-concept justified by recovered amnesias is a modest thing; inherited motives common to all times and races is not, and the second does not follow from the first by anything Jung sets out here. The tension is exactly where the spoke's routing rule bites — method decides, and his own method claim is now on the record instead of inferred from summaries.

The one dispute in this corpus that got settled

Two sources here disagree about whether experienced happiness stops rising with income, and unusually for this wiki, the disagreement has an ending. kahneman-income-happiness-2006 reports a correlation of 0.06 between household income and duration-weighted momentary happiness and reads it as independence; the satiation version of that claim put the ceiling near $75,000. Killingsworth's experience sampling found no ceiling at all.

income-happiness-conflict-resolved holds both. The flattening is real near the 15th percentile of happiness and up to about $100,000; above that percentile happiness keeps rising with log(income) and accelerates in the happiest group; the original flat line was a ceiling in the instrument, with about 85% of high-income responses at the top of the scale. Below $100,000 the slope is steeper for the unhappy, above it flatter, both at P < 0.00001 — so the two results were measuring different parts of one distribution.

What makes it worth its own entry is not the finding but the procedure. Both parties reanalysed one dataset together with a third author as arbiter, and the senior party's result was the one that narrowed. Set that against replication-crisis as this corpus usually invokes it — a story about incentives that reward neither correction nor being corrected. Kept as a tension rather than filed as resolved, because the reconciliation is correlational by the authors' own statement, and Rohrer & Wenz's objection that an "unhappy group" cannot be cleanly identified was conceded, not answered.

Two therapies, one fact, opposite consolations

jung-on-death gives the spoke its first substantive disagreement between two sources that are talking about the same thing.

irvin-yalom makes mortality one of four givens of existence and treats death anxiety as what therapy works on: look directly at it, with no promise about what follows (existential-psychotherapy, staring-at-the-sun). carl-jung treats death as the goal of a directed life-process, calls refusing the ending a refusal to live, and — after the heart attack he was not expected to survive — describes what follows it as "unspeakably glorious."

Both are clinicians writing about the same fact for the same purpose, and they offer incompatible consolations. Neither source cites the other and no evidence here can adjudicate, which is the point worth recording: the disagreement is not empirical, so nothing in this corpus could settle it. It sits exactly on the founding methodological tension, one layer down — two traditions differing about what a therapy is for rather than about what is true.

A finding and its refutation, both held, neither winning

Added 2026-08-05, and it is the spoke's first contradiction between two peer-reviewed sources — every earlier one was essayist against essayist or clinician against clinician.

kruger-dunning-1999: the bottom quartile scored at the 12th percentile and claimed the 62nd, because incompetence removes the means of noticing incompetence. magnus-peresetsky-2022: a three-parameter censored model with a random floor and ceiling, carrying no psychological term at all, fits 665 students' self-predictions almost perfectly — "there is an effect, but it does not reflect human nature."

What makes this different from the other tensions on this page is that it is decidable and has not been decided, and the reason it hasn't is legible from the two papers side by side. Kruger and Dunning saw the objection coming: Study 4 trains the bottom quartile for ten minutes and their self-estimates drop sharply while an untrained control's do not, which a boundary artifact cannot produce. Magnus and Peresetsky never address that experiment. The critique refutes the chart and leaves the experiment standing.

Recorded per the spoke's edge rule as both claims kept and the conflict flagged, rather than as a debunking. The defensible position is narrow: the percentile-crossing figure everyone reproduces is explained by censoring, and the training result is not explained by anything except metacognition yet.

Two second-order notes worth keeping. The critique comes from econometricians, not psychologists (jan-magnus, anatoly-peresetsky) — the sharpest live challenge in this wiki was brought from outside the field, with tools the field does not routinely use, which is an argument for the discipline's porousness and also for why the rebuttal went unread. And the 1999 evidence base is narrower than its fame: every participant across four studies is a cornell-university undergraduate taking a short test for credit, in quartile cells of eleven to thirty-seven.

Correct the thought, or accept it

The spoke now holds both halves of therapy's central disagreement, and they contradict each other directly.

cognitive-behavioral-therapy, through feeling-good: the distressing thought is an error — a cognitive distortion — and the work is to catch it, test it against evidence, and replace it. acceptance-and-commitment-therapy: disputing the thought for accuracy is itself a way of staying entangled, so the work is to change the person's relationship to it. Same object, opposite operation, and both traditions descend from aaron-beck's break with psychoanalysis.

What makes this the wiki's most useful contradiction so far is that, unlike Jung against Yalom, it is not beyond evidence. Both sides are testable by construction and both arrive with large trial literatures — 900+ RCTs for ACT, an older and larger one for CBT. The disagreement is empirical and the corpus simply doesn't hold the comparison.

It also runs the other way through everything else here. Three interpretive traditions tell the reader to welcome the symptom (the-middle-passage, existential-psychotherapy, in-the-absence-of-the-ordinary); ACT is the testable relative of that instruction; CBT is its flat opposite and the one with the mass-market reach. Yesterday's reading — that the poetic corner and the evidential corner might be saying the same thing in different registers — survives only for the acceptance side of the split. It has a rival, and the rival is the one four million people bought.

A weaker second note. second-half-of-life holds two accounts that agree, Jung's and james-hollis's, and the agreement establishes less than it appears to: Hollis is a Jungian analyst writing in a Jungian series, and both reach this wiki through the same essayist (maria-popova). A school agreeing with its founder, relayed by one reader, is one source wearing two hats.

Does the mind avoid effort, or seek it out?

cognitive-miser (Fiske & Taylor, 1984): thought defaults to the cheapest adequate strategy; effort is a cost the mind minimizes whenever a shortcut will do. michael-inzlicht's effort paradox (Inzlicht, Shenhav & Olivola, 2018): effort is aversive by default but can become rewarding in itself once it signals competence, progress or meaning — which is why people voluntarily choose harder puzzles, harder workouts, harder books. the-joy-of-writing cites both traditions in the same short piece, one to explain why AI-shortcut writing is tempting and the other to explain what is lost by taking the shortcut, without noticing they are in tension. Both are real, evidenced traditions; the corpus records the disagreement rather than resolving it — the likely reconciliation (effort aversion is the default, reward is a learned override under the right conditions) is not yet backed by a source this spoke has read in full.

Cross-spoke adjacency

• ../philosophy-wiki — philosophy of mind versus empirical psychology. Jung is the boundary case (he read Kant, Schopenhauer and Nietzsche; he practised psychiatry). Decide by method, not vocabulary: "the unconscious," "perception," "judgment" and "the self" are live terms in both.
• ../defensive-security-wiki — soc-fast-slow-thinking builds a security-operations design on System 1/2. The borrowed metaphor stays there; the psychology lives here.
• ../research-wiki — tools built to augment cognition, and the mechanizing-reasoning lineage. Human cognition as subject is here; the tool is there. Runner-up spoke for the-joy-of-writing (2026-09-27): yakura-llm-human-speech-2024 also supports an AI-buzzword-diffusion angle — how chatbot vocabulary spreads through a population's speech — that leans toward that spoke's human/AI-interaction-tooling territory. Noted here as cross-spoke context only; not ingested there.
• ../speech-audio-wiki — audio AI models; the neuroscience of hearing is here.
• ../optimization-algorithms-wiki, ../machine-learning-wiki, ../mathematics-wiki — the algorithms in algorithms-to-live-by (explore/exploit, Bayes's rule, Markov processes) live there as mathematics; the question of whether people do or should use them lives here. No runner-up spoke was strong; noted as context only.

Index — Psychology Wiki

Catalog of every page, grouped by schema.org @type. Spine: synthesis (thesis), log.md (history), this file (catalog). Spun out of the hub _inbox on 2026-07-28 (3 sources). Psychology has a replication problem and popular writing about it compounds one — record the study behind a claim, or say that a source gave none.

DefinedTerm (concepts / traditions)

• focusing-illusion — attending to something inflates its importance; the frame manufactures part of the answer, so a question shapes its own result · cognitive-bias
• heuristics-and-biases — the Kahneman/Tversky programme: shortcuts producing systematic error; System 1/2 and how far that vocabulary has travelled · research-programme
• analytical-psychology — Jung's school: the collective unconscious, archetypes, the artist as instrument; clinical and interpretive method, and what that method can't establish · tradition
• perceptual-learning — internal perceptual maps: largely stimulus-driven, stable across learning, similar in humans and mice; the corpus's most tractable version of "structure you didn't choose" · mechanism
• free-will — the question split three ways (metaphysics / what caused this behaviour / what belief in it does); the practical stake is blame, not agency; the disciplinary fight over who owns it is part of the record · problem
• flow — Csikszentmihalyi's absorbed-engagement state (DMN, transient hypofrontality, the circulated 120-bits figure); creativity's empirical tradition, set against Jung's depth account of the same fact · concept
• existential-psychotherapy — Yalom's school: distress as a response to four givens (isolation, meaninglessness, mortality, freedom), not a malfunction; the spoke's cleanest method-not-vocabulary boundary case with ../philosophy-wiki · tradition
• second-half-of-life — Jung's parabola reversing at life's midday and Hollis's midlife passage; the corpus's clearest teacher-to-school lineage, and its weakest kind of agreement · concept
• replication-crisis — psychology's finding that much of its literature doesn't reproduce; here it is both a caveat and a move used inside arguments — cited against the opposing literature more readily than one's own · methodological-problem
• sapir-whorf-hypothesis — linguistic relativity: does the language you speak shape how you think? Determinism (strong, rejected) vs relativity (weak, broadly accepted); opens language and cognition, and the corpus holds no experimental evidence for any of it · hypothesis
• dunning-kruger-effect — poor performers overrate their standing (12th percentile → 62nd); the paper has no time axis and no "Mount Stupid" curve. Written from both primaries: a censored-regression model explains the famous chart with no psychology in it, and explains none of Study 4's training result — the critique refutes the figure and leaves the experiment standing · cognitive-bias
• illusion-of-explanatory-depth — believing you understand a mechanism in more detail than you do, revealed the moment you try to explain it; inducible — Fisher/Goddu/Keil (Yale 2015) show searching the internet inflates felt understanding without changing knowledge. The corpus's cleanest contrast with dunning-kruger-effect: depth vs rank, given vs arrived-with · cognitive-bias
• note-taking-and-learning — does how you record a lecture change what you retain? The popular claim (paper beats keyboard), the measured one (Hedges' g = 0.248 for handwriting on tests, 0.919 for typing on volume, two nulls), and the mechanism that reconciles them — the constraint is the active ingredient, not the medium; every figure arrives through a T3 secondary · mechanism
• circadian-rhythm — the ~24h clock, and the finding that reward is a zeitgeber: feeding, sex, drugs and social contact set the clock through dopamine rather than merely obeying it. SCN, clock genes, the unlocated "food clock"; new to the whole hub — no spoke mentioned circadian rhythm before 2026-08-23 · mechanism
• mind-after-midnight-hypothesis — being awake during the biological night as a risk state in its own right, not merely a tired one: positive affect bottoming out 1am–4am, attention narrowing onto the negative, reward distorting into high anticipation and blunted receipt, inhibition falling. Adjusted epidemiology underneath (threefold suicide risk midnight–6am; 4.7-fold opioid overdose while use falls) and an untested middle its own authors name — the spoke's first source that is a well-specified conjecture rather than a finding · hypothesis
• grief — the response to loss, held here only in its depth-psychological form (apprenticeship, descent, "nothing excluded"); the page exists mainly to name the clinical literature this corpus lacks — prolonged grief disorder, the unsupported stage model, the process models · clinical-subject
• cognitive-miser — Fiske & Taylor's 1984 model of thought as effort-averse, absorbing Kahneman/Tversky's heuristics directly; a Psychology Today citation to it turned out to point at a 2010 demand-avoidance study instead, and michael-inzlicht's effort paradox argues the opposite half-truth — effort can itself be the reward · concept
• professional-deformation — a job slowly reshaping habits, judgment and manner, carried outside work (Sorokin's term per the Russian source, disputed in the English one, which names Warnotte; Rogov's four-part taxonomy; Merton's cluster with trained incapacity). Filed as a cognitive bias but a trait-level, occupation-built change, not a single-judgment error; not burnout and not 'destruction'. A Russian/Soviet lineage, set beside the Kahneman/Tversky tradition · concept

MedicalCondition

• dysgraphia — a persistent gap between the writing produced and the writing predicted; a language disorder the name points away from — messy handwriting can be the downstream cost of not retaining the word. Overlaps dyslexia in over a third of cases; production is harder than comprehension. The corrective-vs-accommodation question is its live one, and this corpus holds no outcome comparison · clinical-subject
• dyslexia — deliberately thin: no source here is about it. It arrived as the comparison case in dysgraphia-via-dyslexia, whose whole argument transfers dyslexia's intervention evidence — making it the spoke's clearest load-bearing unread premise · clinical-subject
• psychological-trauma — trauma as an injury to body and nervous system rather than memory alone; attachment as protection, developmental trauma disorder, somatic treatment over talking — and the 122-claim audit that found the core of it unsupported · clinical-subject

MedicalTherapy

• cognitive-behavioral-therapy — thoughts cause feelings, so correct the thought: cognitive distortions caught, tested and replaced; built by aaron-beck out of the objection that psychoanalysis couldn't be tested, and the flat opposite of the corpus's repeated "welcome the symptom" instruction · therapy
• acceptance-and-commitment-therapy — ACT: mindfulness/acceptance/values, built on Hayes's Relational Frame Theory; 900+ RCTs, WHO/APA/VA recognition — the largest evidence base in this wiki, and the counterexample to "clinical = untestable" · therapy

BrainStructure

• reward-system — the mesolimbic system: VTA → nucleus accumbens, the direct/indirect medium-spiny-neuron split (90–95% of striatal neurons), the 1–2% cholinergic interneurons that track salience. Its described job is valence and priority, not pleasure — and the corpus holds no theory of the computation (no reward prediction error, no wanting/liking) · anatomy

Substance

• dopamine — made by tyrosine hydroxylase, cleared by MAOA, both on a daily clock; D1R and D2R with opposite effects and both required. Not the pleasure molecule — and not something you can fast from: mice without TH in VTA neurons die of hypophagia · neurotransmitter

Chapter (sources)

• lewis-borrelli-reward-2021 — Lewis, Florio, Punzo & Borrelli, Adv Exp Med Biol 1344:57–69, read in full: the spoke's first life-sciences primary. Chapter 4 of a circadian volume, so the title promises more than it delivers — no reward prediction error, no Parkinson's, no treatment. What it does deliver is the mechanism under dopamine-nation, and the hedge Lembke had to concede sitting in its opening paragraph. 141 references, zero new data, and mouse knockouts carrying the confident claims while five associational human studies carry the ones about phones and obesity · source · T1 · pmc.ncbi.nlm.nih.gov

ScholarlyArticle (sources)

• lsd-cortical-dynamics-meg — Subramani et al., Communications Biology 9:1244 (2026): re-analysis of 17-participant MEG data (LSD vs. placebo, with/without music). LSD both genuinely weakens alpha/beta oscillatory power AND shifts peak frequencies faster, two separable effects a standard band-power analysis conflates; aperiodic slope flattens, signal fractality/complexity rise; music does not amplify the signature. n=17, secondary analysis · source · T1 · doi.org

• polarization-connectivity-model — Thurner, Hofer & Korbel, PNAS 2025: the spoke's first source about groups rather than individual minds. A homophily + Heider-social-balance model on a Poissonian small-world (N=10⁵) where minimising cognitive dissonance flips opinions and friendships. Polarization ψ jumps discontinuously at a critical degree ⟨k⁺⟩ ≈ 5, splitting the population into two groups of ~50% each; 2% radical influencers turn the jump continuous and bring polarization on at lower connectivity; hysteresis means cutting ties back below the threshold does not undo it. Fits the US curve — Pew polarization 0.11 pre-2010 → 0.19 by 2017 against connectivity 2.2 (2004) → 4.1 (2024) — with only connectivity varied. The hinge is β = 2.7, set above the measured 1–2 range on a hypothesis that political belief is felt more keenly · source · T1 · pnas.org

• polarization-connectivity-dispute — the four PNAS letters about it, read in full: two independent critiques and two authors' replies inside about a year. Völker, Corten & Mollenhorst attack the empirical premise — a doubling of close friendships "would be a remarkable finding, as no such trend has been documented in the extensive sociological literature" — with five specifics including GSS series that reported a decline. Górski et al. attack the mechanism, and the reply's counterexample is the sharp one: their metric scores a textbook polarized society at P_LP ≈ 0.05 where ψ = 1, so it "conflates fragmentation with polarization." What both sides then agree on is the finding: the authors call the friendship evidence base "lamentable" and ask for more · source · T1 · pnas.org

• dtd-field-trial-replication — Ford, Spinazzola & van der Kolk, Eur. J. Psychotraumatology 12(1):1929028 (2021): the DTD field trial re-run at six US sites, 271 children — DTD 27%, PTSD 39%, and DTD-without-PTSD carrying 5.2 comorbid diagnoses against PTSD-without-DTD's 3.8. Also the source for DTD's status: rejected from DSM-5, absent from ICD-11, where complex PTSD was admitted instead · source · T1 · tandfonline.com

• debunking-the-stanford-prison-experiment — Le Texier, American Psychologist 74(7) 2019, read in full: the SPE archives, 15 participant interviews, seven findings — 11 of 17 rules copied from a class experiment, guards briefed on the wanted result and not told they were subjects, conclusions written in advance · source · T1 · gwern.net (PDF)

• kruger-dunning-1999 — the paper itself, read in full: JPSP 77(6), four studies (humour/logic/grammar), the 12th→62nd percentile figure, the four stated predictions, The Burden of Expertise and the false-consensus account of top-performer error, and Study 4's training manipulation — built to answer the regression objection · source · T1 · Cornell undergraduates throughout

• krueger-mueller-2002 — Krueger & Mueller, JPSP 82(2) 180–188 (T1), read from the PDF: the statistical case at its source. A successful replication used against its own explanation — seven of eight unbiased mediator tests move the correlation by ≤.01, and subtracting the better-than-average effect turns 1999's asymmetries of 33/32/32/29 percentile points into 1/10/−4/7. Their difficult test's split-half reliability is .17 · source · T1 · Brown University

• kruger-dunning-2002-reply — Kruger & Dunning's answer, JPSP 82(2) 189–192 (T1), four pages later in the same issue: reliability .17 guaranteed the critics' result, Study 4 at .93 still leaves ~40 points of bottom-quartile overestimation after their correction, a within-person correlation is the wrong instrument (the two weather forecasters), and no artifact answers a training manipulation — 36 percentile points of overestimation down to 17. Ends by refusing to claim the win · source · T1 · Illinois / Cornell

• kahneman-income-happiness-2006 — Kahneman/Krueger/Schkade/Schwarz/Stone, Science 312:1908, read in full: household income correlates 0.32 with life satisfaction and 0.06 with duration-weighted happiness; the predicted bad-mood gap between income bands is 32 pp against an actual 12.2; the time-use mechanism (passive leisure 34.7% → 19.9%, work+commute up) · source · T1 · a Perspective, part review and part first presentation of the authors' data

• income-happiness-conflict-resolved — Killingsworth, Kahneman & Mellers, PNAS 120(10) 2023: the adversarial collaboration that revised Kahneman's own satiation result — flattening only near the 15th happiness percentile and only to ~$100,000, the original plateau traced to 85% of high-income answers at the scale maximum; with the authors' 2024 reply to critics · source · T1 · article paywalled (403), held from the abstract plus the open-access reply

• magnus-peresetsky-2022 — Magnus & Peresetsky, Frontiers in Psychology 13:840180: a doubly-censored tobit model with no psychological parameter fits 665 students' self-predicted exam grades almost perfectly — "there is an effect, but it does not reflect human nature." Never engages Study 4, and misstates that K&D didn't discuss top-performer underestimation · source · T1 · econometricians, not psychologists

• tubbs-mind-after-midnight-2022 — Tubbs, Fernandez, Grandner, Perlis & Klerman, Frontiers in Network Physiology vol. 1 (pub. 2022-03-03), read in full: the Mind After Midnight paper. A mini review — no new data — stacking four circadian curves into a risk state, with properly adjusted behavioural epidemiology under it and an assembled mechanism over it. Its own stated gaps are the page's best content: "all aspects of this hypothesis require empirical validation," and no study has tested sleep deprivation × circadian timing on reward processing · source · T1 · frontiersin.org

• yakura-llm-human-speech-2024 — Yakura et al., arXiv:2409.01754 (2024, rev. 2026): a synthetic-control analysis of 737,083 hours of podcast speech plus a preregistered experiment (N=496) showing people adopt a chatbot's vocabulary into their own active speech after one conversation — read in full for this ingest, and the strongest citation in the-joy-of-writing, which undersells it as "buzzwords" · source · T1 · arxiv.org

Book (sources)

• psychology-of-the-unconscious — C. G. Jung, 1912/1916, tr. Beatrice M. Hinkle (Gutenberg, public domain): the spoke's first primary Jung text. Libido redefined as general life energy "comparable to Bergson's élan vital"; directed vs fantasy thinking; "the myths a mass-dream of the people: the dream the myth of the individual"; and the Miss Frank Miller analysis — a central case built on documents by a woman Jung never met. The book that broke with Freud · T1 · gutenberg.org
• analytical-psychology-and-poetic-art — Jung 1922, tr. Baynes, first published in England 1928 (pp. 225–249 of Contributions to Analytical Psychology), read in full: the spoke's third primary Jung text. The archetype's first real definition in English — "a figure… that repeats itself in the course of history wherever creative phantasy is freely manifested" — a weaker inheritance claim than the paraphrases carry ("not innate ideas, but inborn possibilities of ideas"), creativity as an autonomous complex defined operationally, and Jung stating the post-hoc reconstruction as his method rather than conceding it. Dates the psychological/visionary pair out of this lecture · source · T1 · archive.org
• psychological-types — C. G. Jung, 1921, tr. H. Godwyn Baynes 1923 (Kegan Paul / Harcourt, Brace; UVic scan, public domain): the second primary Jung text here, read for its 57-entry chapter of definitions. "Archetype" is not one of them — it appears once as a parenthetical alias for primordial image; the image's stated basis is Semon's inherited engram; definition 56 calls the unconscious "an exclusively psychological concept, and not a philosophical concept in the metaphysical sense". Individuation defined without a timing claim · T1 · archive.org
• algorithms-to-live-by — Christian & Griffiths (Henry Holt, 2016-04-19, via Wikipedia stub + the book's site): computer-science answers (the 37% rule, explore/exploit, caching, Bayes) offered as the standard that heuristics-and-biases measures people against; read through descriptions, no chapters seen, no human-subjects evidence in hand · source · T3 · en.wikipedia.org
• thinking-fast-and-slow — Kahneman 2011 (via Wikipedia): System 1/2, prospect theory, WYSIATI, the experiencing vs remembering self and the peak-end rule; and its own Ch. 4 citations at a replicability index of 14, which Kahneman conceded · source · T2 · en.wikipedia.org
• determined — Sapolsky 2023 (via Wikipedia): no free will, therefore rebuild blame and punishment; reception splits on whether a neurobiologist can answer a philosophical question (Riskin: "scientism") · source · T2 · en.wikipedia.org
• staring-at-the-sun — Yalom 2008 (via Wikipedia): death anxiety as a pervasive influence dissolved by looking at it directly; the mortality given of existential-psychotherapy taken alone · source · T2 · en.wikipedia.org
• the-middle-passage — Hollis 1993 (via The Marginalian): first adulthood runs on a provisional personality assembled from family conditioning; midlife distress is the authentic self pressing against it, and should be "welcomed" — the corpus's most direct statement of its own thesis, and the only one claiming the structure can be dismantled · source · T3 · themarginalian.org
• a-liberated-mind — Hayes 2019 (via Wikipedia on Hayes): the trade statement of acceptance-and-commitment-therapy; the entry confirms the book exists and not what it argues — recorded for the 900-trial programme behind it · source · T3 · en.wikipedia.org
• burnout-stress-cycle — Nagoski & Nagoski 2019 (via listings/reviews): the stressor is not the stress — removing the cause doesn't close the physiological loop; plus Human Giver Syndrome (Kate Manne) as a structural account of women's burnout; confident physiology, thin citation · source · T3 · burnoutbook.net
• dopamine-nation — Lembke 2021 (via Wikipedia + critical reviews): pleasure and pain on one seesaw, tipped by digital overstimulation; abstinence to recalibrate. The mechanism holds and the book outran it — the seesaw oversimplifies, "dopamine fasting" misnames the object, recovery runs weeks not days, and Lembke conceded the central simplification in the NYT · source · T2 · en.wikipedia.org
• the-body-keeps-the-score — van der Kolk 2014 (via Wikipedia): trauma lives in the body, so treat the body (EMDR, yoga, neurofeedback) — ~300 weeks on the NYT bestseller list against a 2025 BJPsych Bulletin audit of 122 claims finding the brain-damage and body-treatment core unsupported; the corpus's first case where reach and evidence are both counted, and the first with no author concession · source · T2 · en.wikipedia.org
• feeling-good — Burns 1980 (routed as an Amazon listing, content via Wikipedia): the mass-market statement of cognitive-behavioral-therapy, 4M US copies — and the only book here that was itself trialled as the treatment (meta-analysis of six bibliotherapy studies, Scogin et al. 1989), which inverts the spoke's bestseller-then-audit pattern; the jacket still outruns the trials · source · T2 · en.wikipedia.org + amazon.com
• in-the-absence-of-the-ordinary — Weller (via The Marginalian, 2026-08-02): uncertainty as initiation, grief as apprenticeship, "welcoming all of who we are… Nothing excluded" — the third statement of the corpus's "welcome the symptom" instruction and the first to claim a collective initiation; the least testable source here (T3 rates Popova's transmission, not the content) · source · T3 · themarginalian.org
• the-trouble-with-testosterone — Sapolsky 1997 essays (via a Google Books listing): the early statement of the biology-of-behaviour position; the spoke's weakest source — establishes what the book covers, nothing about what it argues · source · T4 · books.google.com

Article (sources)

• man-the-unknown-carrel-1935 — Carrel, Man, The Unknown (1935): primary text behind the specialist-bias quote on professional-deformation, read in context (science-of-man lag, 'built at random'); also argues eugenics. T1 · source · T1 · 1935
• professional-deformation-wikipedia-en — English Wikipedia on déformation professionnelle: medical origin, Warnotte as probable coiner (Sorokin 'citation needed'), Merton's cluster, Carrel, 'nerdview'. T2 encyclopedic, short, link-only related concepts · source · T2 · en.wikipedia.org
• professional-deformation-wikipedia-ru — Russian Wikipedia on professional deformation: definition, Sorokin as coiner, two classifications, deformation vs destruction. T2 encyclopedic, but flagged for missing footnotes since 2012 and its first classification uncited; no studies behind any example · source · T2 · ru.wikipedia.org
• ai-cognitive-science-origin — Hout & Robbins (The Conversation, 2026-10-07): AI's learn-from-examples architecture descends from cognitive science (Rosenblatt's Perceptron, Hebb, 1980s connectionists), which is why it hallucinates like reconstructive human memory rather than erring like a lookup table; cites Loftus on false memory by name only. Names/dates checkable, no primary citations · source · T3 · theconversation.com
• mind-after-midnight-sciencealert — ScienceAlert (Carly Cassella, stamped 2026-09-20), the source that arrived by Telegram. The best T3 this spoke has ingested: every claim linked to its primary (which is the only reason tubbs-mind-after-midnight-2022 could be read), the authors' central caveat carried rather than dropped, and its subject separated from ordinary 3am waking. What it loses is the article type (mini review, never mentioned), the venue's name (Network Psychology for Physiology) and the authors' tentativeness, under a "scientists warn" frame. A re-dated republication — 2026 stamp over a 2025 image path — with no disclosure · source · T3 · sciencealert.com
• mind-after-midnight-nimh-grant — a university-of-arizona press release reprinted by a local paper, undated, found while reading the paper: $3.8M from NIMH, five years, Grandner and Fernandez leading. The design is the content — two lab visits, kept awake to 2am in one and woken at 2am in the other, which is the paper's own confound turned into two arms. A funded protocol is not a result, and the page says so · source · T3 · eacourier.com
• psypost-polarization-model — PsyPost (Karina Petrova, 2026-09-04), the source that arrived and the reason two others exist. Reports polarization-connectivity-model accurately in outline and as an explanation — and mentions none of the published exchange over it, including the critique aimed at the premise its headline rests on. Also states a threshold ("four and a half close connections") that the paper's own continuous transition at γ=0.02 cannot support. The second instance in this spoke, after dunning-kruger-misread, of the tertiary layer keeping the finding and dropping the argument · source · T3 · psypost.org
• free-will-belief-effects — The Conversation (Aaron Walton, Melbourne, 2026-07-26): two experiments (260/262) find free-will scepticism raises forgiveness and lowers hatred; a 44-country survey (n=8,917) ties stronger free-will belief to harsher punishment; discounts the contrary literature as non-replicating · source · T3 · theconversation.com
• kahneman-focusing-illusion — Space Daily (Mal James, 2026-07-23): Kahneman's "fortune-cookie maxim," the California/weather study with Schkade, and the 2006 Science income-happiness split; the trigger source · source · T3 · spacedaily.com
• jung-on-creativity — The Marginalian (Maria Popova, 2026-07-22) on Jung's Modern Man in Search of a Soul: psychological vs visionary creation, the artist as instrument, the collective unconscious as Jung's break with Freud · source · T3 · themarginalian.org
• jung-on-death — The Marginalian (Maria Popova, 2026-07-31) on Jung's "death is not an end but a goal": life as an irreversible energy-process, an unconscious in "relative spacelessness and timelessness," the parabola reversing at midlife, and the 1944 near-death letters. Quotes are solid; the Einstein/spacetime framing is the essayist's, not Jung's · source · T3 · themarginalian.org
• why-writing-feels-hard — Lit Hub (April Dávila, from Sit Write Here, 2026): writing feels dangerous because the brain reads exposure as social threat; cites Eisenberger/Baumeister/Chen/Raichle/Csikszentmihalyi/Dietrich by name, then welds them into a case for ten minutes of pre-writing meditation — the corpus's clearest citation chain terminating in a practice · source · T3 · lithub.com
• does-language-determine-thought — TheCollector (Giulia Villa, 2026-08-04): a clean separation of linguistic determinism from linguistic relativity and an accurate report of the consensus — citing no study and no researcher past Sapir and Whorf, with colour terms and emotion vocabulary as its whole evidence; omits both Hopi time and Inuit snow · source · T4 · thecollector.com
• dunning-kruger-misread — Silicon Canals (Justin Brown, 2026-08-05): what the 1999 study measured (12th → 62nd percentile, single sitting) against what the internet made of it; four studies for and against, incl. the first named successful replication in this wiki (2021 Nature, ~4,000/study) and a 2022 Frontiers artifact case — three of its six citations name no author · source · T3 · siliconcanals.com
• epictetus-learning-spacedaily — Space Daily (mal-james, 2026-08-08) on Discourses 2.17.1: adds Fisher/Goddu/Keil (Yale 2015, nine experiments) on search-inflated understanding, two unheld david-dunning quotes, and a self-test ("explain it without looking anything up"). Reports the Dunning-Kruger reassessment rather than the myth, citing Eric Gaze in Scientific American — the genre most likely to get it wrong, getting it right · source · T3 · spacedaily.com
• dysgraphia-via-dyslexia — coral-ps-hoh, psychology-today (2026-08-09): dysgraphia as a language disorder, not penmanship; heavy overlap with dyslexia; correct the foundational processes rather than accommodate. Four real references with volumes and pages — the best-cited popular source in this wiki — and no disclosure that the author's company sells remediation. The 20% prevalence figure is cited but unverified, and is the same number quoted for dyslexia · source · T3 · psychologytoday.com
• handwriting-vs-typing-notes — silicon-canals (2026-08-12): the handwriting-beats-typing claim cut down to size — a 2024 Educational Psychology Review meta-analysis (24 studies, g = 0.248), a 2021 direct replication of Mueller & Oppenheimer that found no difference (74 laptop / 68 longhand), and a 2026 Frontiers in Education EEG study (n=33) that found none either. The corpus's first replication failure that shrank a finding instead of deleting it — and the outlet's second study reported with no authors · source · T3 · siliconcanals.com
• human-mouse-perceptual-maps — MedicalXpress on Seiler & Rumpel (Mainz, Communications Psychology 2026): humans and mice build similar internal maps of pulsed-sound sequences, stable through learning; the corpus's only experiment · source · T2 · medicalxpress.com
• generational-boundaries-yourtango — YourTango (Luke Aliga, 2026-09-06): boomers keep distance, Gen X won't set boundaries, millennials were first to, Gen Z overdo it — 813 words, no survey, no sample, no named researcher. Six links do the authority work and three were chased: the Johns Hopkins one is a careers blog that never mentions boundaries, the Psychology Today one is a therapist's generation-neutral argument reworked into an accusation against Gen Z, the Truity one is a personality-test vendor with no data. Not the citation-free floor (musical-expression-neuroscience) but its inverse — a citation layer that constrains nothing · source · T4 · yourtango.com
• musical-expression-neuroscience — unattributed text pasted to the curator's channel (2026-09-01, held in raw/): how pianists encode emotion — music borrowing speech pathways, an ACC/inferior-frontal shift from mechanical to expressive playing, autonomic and kinematic correlates, and AI analytics teaching expressivity "significantly faster". No author, no outlet, no date, no reference; five appeals to an unnamed literature in ~500 words. The corpus's zero-citation floor, and a map of a real gap (performance psychology) rather than a finding · source · T4 · no URL
• the-joy-of-writing — Psychology Today Canada, Modern Boredom blog (james-danckert, 2026-09-24): handing writing to an LLM costs three things — homogenized output, the reward of effort, and a writer's own voice. Two of three load-bearing citations check out (michael-inzlicht's effort paradox; Yakura et al.), one (a Bieleke & Wolff preprint) could not be read at all. States its own conflict of interest: the author expects to benefit from a settlement against anthropic over unauthorized use of his book · source · T3 · psychologytoday.com

Person

• alexis-carrel — surgeon-biologist, Nobel 1912, author of Man, The Unknown; held for the specialist-bias quote, with his eugenics advocacy recorded · thin node

• pitirim-sorokin — Russian-American sociologist; credited by the Russian source (disputed in the English one) with coining professional-deformation · thin node

• karim-jerbi — Université de Montréal (CoCo Lab) / Mila; corresponding author of lsd-cortical-dynamics-meg · thin node

• robin-carhart-harris — UCSF; co-author of lsd-cortical-dynamics-meg, collected the original MEG data it re-analyzes · thin node

• julian-d-ford — UConn School of Medicine; first author of dtd-field-trial-replication and co-proposer of developmental trauma disorder · thin node

• joseph-spinazzola — Foundation Trust, Melrose MA; co-author of both DTD field trials and of the 2002–03 survey that argued for the diagnosis · thin node

• sigmund-freud — founder of psychoanalysis; origin of the corpus's central predicament — enormous explanatory reach on a method that can't test its own claims

• robert-sapolsky — neuroendocrinologist and primatologist; two books here, twenty-six years apart, hardening one claim (behaviour has biological causes) into a thesis about free will and punishment

• erich-fromm — Frankfurt School psychoanalyst; social character — the psyche as something a political and economic order shapes; the spoke's hardest philosophy boundary case

• daniel-kahneman — founder (with Tversky) of heuristics-and-biases; 2002 Nobel in economics for psychology; publicly walked back his confidence in social priming

• irvin-yalom — Stanford psychiatrist emeritus, built the best-known model of existential psychotherapy; wrote a novel about Nietzsche and files here anyway, on method

• mal-james — writer of Space Daily's "Mind & Meaning" essays; two sources here, and one method: a compressed saying, two named findings, and a test the reader can run rather than an exhortation. Two for two on accuracy; every study still arrives at second hand

• james-hollis — Zurich-trained Jungian analyst; Jung's individuation applied to one life stage, the midlife passage from provisional self to second adulthood

• steven-hayes — clinical psychologist (Nevada, Reno); built both a theory of cognition (RFT) and the therapy applying it (ACT), then trialled it — the spoke's counterexample on clinical testability

• anna-lembke — Stanford addiction psychiatrist, 100+ papers; the spoke's clearest case of one person at both ends of its evidence problem — the peer-reviewed record and the bestseller her field disputes

• emiliana-borrelli — Chancellor's Professor at UC Irvine, INSERM research director; built the first dopamine D2 receptor knockout mouse, so much of the D2R literature her chapter reviews runs back through her own lab's tool — the register problem's mirror image, a specialist keeping the hedge she has no book to sell

• carl-jung — founder of analytical psychology; enormous influence, thin empirical support; the spoke's clearest boundary case with ../philosophy-wiki

• maria-popova — sole author of the-marginalian; the route by which four of this spoke's sources arrived, and therefore a selection filter on it as well as a quoter of record — all four land on the depth-psychological end

• bessel-van-der-kolk — psychiatrist, founded the Trauma Center (1982), DSM PTSD committees; the register problem without the concession — recovered-memory testimony in the 1990s, a 2003 critique never answered, and a bestseller in between

• aaron-beck — father of cognitive therapy; built it out of the finding that psychoanalysis had no empirical support for depression, and both of this spoke's testable therapies descend from that break; paged thin from one article about someone else's book

• david-d-burns — Beck-trained psychiatrist who put CBT into four million households; the register problem in its cleanest form — a rigour critic of antidepressant research whose own jacket promises a cure, over a real but modest evidence base

• david-dunning — social psychologist, co-author of kruger-dunning-1999; paged on the ingest that made him an author rather than a subject, which is the ENTITIES rule working

• justin-kruger — first author of kruger-dunning-1999; on the record here for the four-prediction structure and the Study 4 training manipulation, the part of the paper most often left out

• jan-magnus — econometrician (VU Amsterdam), first author of magnus-peresetsky-2022; the wiki's sharpest live challenge to a psychological finding, brought from outside psychology

• matthew-killingsworth — happiness researcher whose experience-sampling app supplied the data; contradicted Kahneman on income satiation and was brought into the reanalysis that settled it, which is the corpus's one counter-instance to its own story about incentives

• anatoly-peresetsky — econometrician (HSE Moscow), co-author of magnus-peresetsky-2022; the 665-student exam dataset behind it is his institution's

• justin-brown — Co-Publisher of silicon-canals and byline on dunning-kruger-misread; paged thin on purpose — no author page, no stated credential, no biography on the piece

• coral-ps-hoh — clinical linguist (PhD linguistics), CEO of EduNational LLC; a linguist writing clinically where this corpus's clinicians are psychiatrists — and the milder second instance of the clinician-with-a-service structure after anna-lembke · author

• luke-aliga — byline on generational-boundaries-yourtango; a technical-writing degree and no psychology credential, writing clinical and generational claims in the present tense — the justin-brown case again, one rung lower · author

• francis-weller — psychotherapist working on grief in its ritual and communal forms; the third clinician here to call distress an invitation rather than a malfunction, and the first to say it of a period rather than a person; makes no claim an audit could check

• elizabeth-klerman — Professor of Neurology, MGH / Harvard Medical School; senior author of tubbs-mind-after-midnight-2022 and the only one of its five authors not at Arizona. A biomathematical modeller of sleep and circadian rhythms, which is where the paper's insistence on separating circadian phase from sleep loss comes from. ORCID confirmed by lookup, not inferred · author

• michael-grandner — director of the Sleep and Health Research Program, university-of-arizona; third author of the paper and lead investigator on the NIMH study built to test it — the person carrying the hypothesis from review to protocol. On the record undercutting his own press release: the folk claim "isn't groundbreaking," the neuroscience is what's missing · author / PI

• fabian-xose-fernandez — associate professor of psychology and neurology, university-of-arizona; second author and co-lead investigator, whose claim for the study is the testable one — that nocturnal suicide risk is modifiable, which would turn a description into a treatment target · author / PI

• andrew-tubbs — MD/PhD, first author of tubbs-mind-after-midnight-2022; trained in CBT-I under two of his own co-authors. No same-as recorded — every candidate identifier 404'd or would not resolve, and the hub rule is to leave the field off rather than guess · author · thin node

• carly-cassella — science journalist at ScienceAlert; paged for one editorial decision — she kept the paper's central caveat and separated her subject from an adjacent one, where this corpus's popular layer usually keeps the finding and loses the argument · author

Organization

• yourtango — NewsMediaOrganization: US relationships-and-self listicle publisher; graded T4, a step below this spoke's T3 cap on popular psychology, because there is no credential, no review line and no measurement anywhere in the chain. Wikipedia knows it only as a section of Tango (American magazine), the 2005 print title it grew from · publisher

• the-marginalian — NewsMediaOrganization: Popova's reader-funded, ad-free, one-person essay site since 2006; no editor and no fact-checker by design, which is why its pieces cap at T3 here · publisher

• cornell-university — CollegeOrUniversity: the entire empirical base of kruger-dunning-1999 — both authors, and every participant across four studies, in cells of 11–37 · research site

• space-daily — NewsMediaOrganization: space-and-astronomy publisher (est. 1995) running a daily "Mind & Meaning" psychology column; two sources here, both T3 for the genre's habit of reporting studies without titles or venues · publisher

• silicon-canals — NewsMediaOrganization: Amsterdam tech publication (2014, Brown Brothers Media since Nov 2025) whose three editorial pillars include Mind; discloses AI-assisted drafting with human editorial judgment, and caps at T3 anyway — the review is of the writing, not of the science · publisher

• uc-irvine — CollegeOrUniversity: where lewis-borrelli-reward-2021's authors and mouse genetics come from, including INSERM U1233 on a US campus. The counterpart to cornell-university — that node is where a paper's participants came from, this one where its authors did · research site

• psychology-today — NewsMediaOrganization: the US's best-known popular psychology title, whose blog platform recruits practising clinicians to write under their own credentials with an editorial reviewer line; the clinician-with-a-service pattern is built into the venue, not just its authors. Caps at T3 like the rest of the genre · publisher

• university-of-arizona — CollegeOrUniversity: institutional home of the whole Mind After Midnight programme — four of five authors, and the NIMH study testing it. Two units carry it across psychiatry, psychology and neurology, which is not administrative detail: the hypothesis needs a clinical outcome, a behavioural measure and an EEG in one protocol, and the authorship follows that shape · research site

• science-alert — NewsMediaOrganization: Australian science-news site (2004). The top of this spoke's T3 band — it links every claim to the primary and names its fact-checker, where space-daily does neither. Held at T3 for framing, not error, and for re-dating republished copy without saying so. Hard-blocks automated fetching (403); read through Firecrawl · publisher

CreativeWorkSeries

• advances-in-experimental-medicine-and-biology — CreativeWorkSeries: Springer's themed-volume biomedical series; lewis-borrelli-reward-2021 is chapter 4 of vol. 1344, Circadian Clock in Brain Health and Disease. The spoke's first peer-reviewed venue node among four news publishers — and the page that records what they share: neither publishes new data · venue

Periodical

• frontiers-in-network-physiology — Periodical: open-access Frontiers journal; tubbs-mind-after-midnight-2022 is in its founding volume 1. The spoke's first journal node. Two facts it exists to hold: the venue publishes "mini review" as an article type and applies the label accurately — every retelling downstream drops it — and Frontiers' peer review is contested, so a T1 here is a claim about this paper, not about the venue · venue

Synthesis

• synthesis — the thesis: the mind runs on structure its owner didn't choose and can't inspect, plus the founding method spectrum (psychophysics → survey → clinical interpretation)

Notes

• Amazon-listing precedent (2026-08-03): feeling-good was routed as a bookseller page, which is T4 and yields edition facts and jacket copy only (cf. the-trouble-with-testosterone). The argument and the bibliotherapy studies were taken from Wikipedia and the page is tiered on that, with both URLs recorded. A listing alone is not enough to write from; a listing plus an encyclopedia is the same standard as the rest of the Book shelf here.
• Entity nodes deferred — Amos Tversky, David Schkade, Maria Popova, Mal James, Aaron Walton, Jason Zweig, the determined reviewers (Fischer, Riskin, Setiya, Crumey) and the Seiler/Rumpel authors are named on their pages but not paged. Tversky remains the likeliest first — he is half of heuristics-and-biases and recurs across thinking-fast-and-slow. Michael L. Perlis (Penn) joins the list on 2026-09-19: fifth author of tubbs-mind-after-midnight-2022, named on three pages, with no evidence in either source beyond an affiliation and a training mention. Andrew Tubbs was paged without a same-as the same day — the other half of the rule, where the node is warranted and the identifier is not. Dunning and Kruger were deferred earlier the same day and paged hours later — the rule behaving exactly as written, recorded because it is a clean worked example. Against dunning-kruger-misread they were the subject of a source, which ../ENTITIES.md excludes from discovery. Against kruger-dunning-1999 they are its authors, and the paper supplied the affiliation and the substance the popular piece hadn't. Deferral was the right call on the evidence then available, not a backlog item.
• Standing weakness — broken once, on 2026-08-05. For its first eight days nothing in this wiki had been read in the original. kruger-dunning-1999 and magnus-peresetsky-2022 are the exception and the spoke's first two T1s. The weakness holds for every other subject: three sources are Wikipedia articles about books and people, one is a bookseller's catalogue entry, and encyclopedic sources are reliable about what a work says and how it was received without being the work. Next highest value is still the 2006 Science income-happiness paper.
• What the primaries bought (2026-08-05): reading both ends of one dispute surfaced a fact no secondary source in the corpus reports — the 2022 critique never engages the 1999 paper's Study 4, the experiment built to answer exactly that objection. Argument in synthesis.md. The cost was two fetches, which is the case for closing the remaining primary-source edges rather than filing them as aspirations.
• First life-sciences primary (2026-08-23). lewis-borrelli-reward-2021 closed the neuroscience corner's evidence gap — until then determined (Wikipedia) and the-trouble-with-testosterone (a bookseller listing, the only T4) carried "behaviour has biological causes" with no biology under it. It also opened the corpus's largest current gap: no theory of what dopamine computes — Schultz's reward prediction error and Berridge's wanting/liking are absent from every page here.
• Bridge node: soc-fast-slow-thinking (../defensive-security-wiki) borrows System 1/2 for SOC design. The metaphor lives there; the psychology lives here.