Meta-research · 100 entries · Independence audit folded in · Gap research (precognition · panpsychism) folded in · Researched Nov 18, 2026

Is the intuition real?
The research on the research.

The question: a person's hypotheses keep landing on findings already published in literature they have never read. Is that a documented phenomenon, can bias across research sources be aggregated the way Ground News aggregates it for news, and where does a claim's ground truth actually come from? One hundred entries were pulled across four sub-questions. This page is what they say — including where the answer is “nobody has studied that.” An independence audit then asked whether the 100 were actually independent of each other. They weren't, fully — the honest count is 100 entries → 83 distinct contributions → ~55 independent evidence units → ~20–25 research programs, and the full audit is folded in directly below.

The verdict, up front

Convergence is real — at the level of communities, not lone minds. Multiple independent discovery is the statistical norm in science: Merton catalogued 264 multiples, Simonton 579, and the cases run from calculus to CRISPR. Aggregated human judgement also forecasts not-yet-run replication outcomes at 71–78% accuracy [published].

But the precise personal claim is untested. No published study was found that tests whether an individual can predict the specific findings of papers they have not read. The parsimonious account of “intuiting” a published result: you share the inputs, the problem is ripe, and the literature is large enough that the idea is probably already there — possibly dormant. That is the adjacent possible doing its work, not precognition.

And synchronicity has no empirical support. Jung's own statistical test did not confirm it [reported]. The structural explanation — shared prerequisites → same idea, repeatedly, measurably — is the one the evidence carries.

One correction, from this page's own audit: the 100 entries were called “independent” because they were independent of our prior work — not of each other. Audited against each other, they cluster into ~20–25 research programs. The verdicts above stand, but they rest on fewer independent legs than the row count suggests. Read the audit ↓

And the excluded territory has now been searched too. The first run deliberately left out parapsychology and panpsychism. A follow-up run (~50 entries: precognition meta-analyses, panpsychism, and the mundane “bridge” literatures) covered exactly that ground, with full critical appraisal — and the core gap stands: nothing there tests individuals predicting specific findings of papers they have not read. Read the gap research ↓

100
entries, 25 per sub-question — the audit below recounts what “independent” really comes to
1,370
unique source URLs compiled across all 26 prior research runs
12,642
sources indexed in the Mycelium, across 40 types
0
studies found testing an individual predicting unread findings — the core gap

The independence audit: are the 100 actually independent?

No. “Independent” in the original mission meant independent of our prior work — externally sourced, not from our own data. It did not mean independent of each other. Audited Nov 18, 2026, each entry was tested against the other 99 on five axes: same author or program, same dataset, same historical event, same source document, and same secondary link in the research notes.

The “100 independent” label is wrong at every grain except the loosest (100 distinct entries). The rough estimate given before the audit — “~15–25 programs” — held up: the audit says ~20–25. The honest label is the four numbers above.

The sharpest findings

The 17 removals — how 100 becomes 83

A24
Duplicate of C6 — the same SSRP paper, listed twice.
D17
Same event as A8 (Newton–Leibniz).
D20
Same event as A9 (Darwin–Wallace).
D19
Same event as A13 (CRISPR).
D24
The experiment inside D1's 1952 monograph.
D2
Pauli–Jung correspondence — same authors, same argument, same 1952 volume as D1.
B20
Hindawi counts are a slice of B19's Retraction Watch database.
A6
Johnson (2010) popularises A5 (Kauffman) — diffusion, not evidence.
C23
repliCATS is one arm of C22 (SCORE).
D9
Stigler's law is Merton's D7 idea restated — Stigler credits Merton.
C24
National Academies report summarises C2/C3 — a review, not new evidence.
C13
Munafò manifesto synthesises the replication literature — a proposal, not evidence.
D25
Garfield's essay reviews Merton/Simonton — a review of the multiples cluster.
A19
Systematic review of the literature-based-discovery field A17/A18 founded.
D22
Primordiality theory synthesises Merton/Kuhn/zeitgeist — a synthesis of the multiples cluster.
D23
Cambray reviews Jung/Pauli — a review of the synchronicity cluster.
D4
Simonton (2004) synthesises his own A3/A4/D5 work — a substantive synthesis, but no new data.

Kept, with dependence noted: A23/A25 (forecasts of C2/C3 data — but the forecasts themselves are new measurements about human judgement); B21 (an analysis of B22's list — the analysis is new even though the list is shared); C21 (reviewer ratings of C20's format — new data about the format); B1 (the Ground News product — the integration of B2–B4 is itself the contribution). All 100 entries remain in the index below so the audit trail stays visible.

The 20 clusters

  1. Multiples sociology (Merton–Simonton lineage) — 14 entries; 1 program. Empirical core is 3 overlapping catalogues (A1, A2, A3).
  2. Historical multiple-discovery cases — 12 entries; 9 events (3 double-listed: calculus, natural selection, CRISPR).
  3. Adjacent possible / recombination — 4 entries; 3 programs (Kauffman, Arthur, Uzzi); A6 derivative.
  4. Delayed recognition — 2 entries (+2 shared); Stent's concept + Ke's 22M-paper quantification.
  5. Literature-based discovery — 3 entries; 1 program (Swanson) + 1 review.
  6. Intuitive expertise — 2 entries; 1 program (Klein; A21 is the Kahneman–Klein synthesis of it).
  7. Replication movement + forecasting — 11 entries; 7 datasets, 5 sub-programs, 1 movement; A24 = C6. Nosek leads C2 and C19/C25; Camerer leads C5/C6; Klein leads C3/C4.
  8. Industry replication — 2 entries; 2 programs (Amgen, Bayer) — genuinely independent.
  9. Funding bias — 3 entries; 3 teams, overlapping primary literature.
  10. Publication bias — 6 entries; 4 empirical efforts + 2 methods.
  11. Citation distortion / cartels — 5 entries; 5 teams — the best-diversified cluster.
  12. Predatory publishing / retractions — 5 entries; 3 sources (B20 ⊂ B19; B21 sampled from B22).
  13. Ground News rating stack — 4 entries; 1 system (B1 aggregates B2–B4 by design).
  14. Citation-intelligence tools — 2 entries; 2 products (scite, Consensus).
  15. Evidence-grading infrastructure — 5 entries; 3 lineages (GRADE, Cochrane, PRISMA) — tools, not evidence.
  16. Preregistration / Registered Reports — 4 entries; 1 movement (Nosek/OSF + Chambers).
  17. Synchronicity — 5 entries; 1 primary source (the 1952 Jung–Pauli volume) + 1 review.
  18. Discovery/invention philosophy — 4 entries; 1 debate (2 primary positions + 2 meta-analyses).
  19. Standalone empirical — 5 entries; 5 programs (Ioannidis, Baker, Freedman, Simmons, DENDRAL).
  20. Standalone syntheses — 2 entries; 2 reviews (C13, C24).

Concentration verdict — by this page's own rule

The framework below sets a citation-concentration rule: >50% of supporting citations from one cluster = authority concentrated, not distributed. The report proposed that rule but never ran it on its own corpus. The audit did:

What this means — and the audit's own limits

Four questions, four answers

Each sub-question got 25 studies. Ratings used throughout: direct = bears directly on the question, adjacent = same phenomenon at another scale or a mechanism an answer needs, distant = background only.

A · Intuition converging with literature

Documented for groups. Unstudied for individuals.

Multiples are the norm, not the exception — 264 in Merton's corpus (179 doublets up to 2 nonaries) [published], 579 in Simonton's [published]. Simonton's adjudication favours chance, then zeitgeist, over genius: given enough workers and a ripe problem, duplication is expected. Prediction markets forecast replication at 71% and 75% accuracy; market beliefs correlated 0.84 with outcomes [published]. Expert intuition itself is only trustworthy in high-validity environments with rapid feedback (Kahneman & Klein) — and science is mostly a low-validity one.

Plain gap: no study tests individuals predicting specific findings of papers they haven't read. A23–A25 forecast whether findings replicate, never what they say.

B · Bias aggregation for research sources

Every component exists. The product doesn't.

Funding bias is quantified: industry sponsorship → favourable results RR 1.27, favourable conclusions RR 1.34 (Cochrane, 75 papers) [published]. Publication bias nearly halves some effects after adjustment (0.67 → 0.42) [published]. Greenberg showed citation bias, amplification, and invention manufacturing authority across 242 papers and 220,553 citation paths [published]. Retractions are openly tracked (63,000+ records [reported]) and scite has classified 1.2B+ citation statements [reported].

Plain gap: no integrated “Ground News for research” tool or study was found. The pieces sit in separate silos.

C · Navigating to ground truth

Locatable — but only probabilistically.

Depending on field and criterion, roughly a third to three-fifths of tested findings replicate: psychology 36%, Many Labs 2 54%, social science 62%, preclinical cancer as low as 11% [published]. GRADE is the best certainty-grading machinery available. Preregistration demonstrably changes the evidence base: first-hypothesis support runs 96% in standard literature vs 44% in Registered Reports [published]. DARPA's SCORE: 49% of 164 papers replicated; human forecasters hit 76–78%, three ML systems failed [reported].

Working rule: a single study's truth probability starts at its field's replication rate — never at 50/50, never at the author's confidence.

D · Discovery vs invention, synchronicity

A live philosophy question. An empty empirical one.

Popper: theories are invented conjectures, tested by falsification. Bacon: laws are in nature, discovered by induction. Ropolyi (2024) dissolves the dichotomy — invented artefacts must then be discovered. Priority disputes are structural (Merton): credit is science's sole property right, so convergence reliably produces conflict — Newton wrote the Royal Society report that declared Newton first. The Matthew effect then hands the credit to the famous name.

Plain gap: synchronicity has no peer-reviewed empirical support. Jung's own astrological experiment did not confirm it, and the concept is widely regarded as unfalsifiable.

The numbers that carry the argument

Every figure below is drawn from the studies in the index, tagged the way the research tagged it.

Replication rates by field

Share of published findings that survived direct replication. This is your prior for any single-study claim. All [published] except SCORE [reported].

Merton's multiples: how many minds, same idea

264 independent multiple discoveries by team size [published]. Singletons — one mind alone — are the case that needs explaining.

Can judgement forecast unseen results?

Accuracy of forecasts of replication outcomes made before the replications ran [published]. Base rate sits near 50% — the edge is real but modest.

The funding-bias signal

Industry-sponsored vs independent studies, Cochrane review of 75 papers [published]. RR above 1.0 = more favourable to the sponsor. Conclusions agreed with the studies' own results less often (RR 0.83).

The 100 entries

The full index — all 100 rows kept, including the 17 the audit folds into other entries, so the audit trail stays visible. Filter by question or by how closely an entry relates to the meta-concept, or search by name. Each entry links to the source the research recorded.

The navigation framework: qualia & quantifiers

The bounds and rules for finding where a claim's ground truth comes from. This framework is a synthesis built from the corpus above — a proposal, not a published instrument — and its thresholds are calibrated to the figures in this research. Note, per the audit: the report proposed the citation-concentration index below but never ran it on its own corpus until the audit did.

Qualia — the judgements a reader must make

Provenance depth

How many citation hops separate a claim from primary data. A chain that dead-ends in reviews is a different animal from one traceable to data (Greenberg, B14).

Prerequisite saturation

Are the claim's necessary inputs — methods, data, theory — all in the public record? If not, the claim is premature, not wrong (A5, A14).

Independence of confirmation

Do confirming studies share authors, funders, data, or labs? Shared lineage is not independent evidence (B7, B14–B16).

Adversarial exposure

Has the claim survived attempts to refute it — contradicting citations, failed replications, critical reviews — not just accumulated supportive ones?

Registration integrity

Was the plan timestamped before data collection, and do the reported outcomes match the plan? (C19–C20)

Quantifiers — named measures, bounds, rules

Try it: the ground-truth dial

Pick a field. The dial returns that field's replication prior and effect-shrinkage factor from the studies above — the two numbers that should discipline any single-study claim before you read a word of it.


Replication prior
—
Effect on replication
—
Rule that follows
—

Seven rules of navigation

  1. Trace before trusting. Follow a claim to primary data; if the chain passes only through reviews, downgrade certainty one level. (B14)
  2. Count independent confirmations, not citations. Citations from one cluster count once. (B14–B16)
  3. Start from the base rate. A claim's prior is its field's replication rate — not 50/50, not the author's confidence. (C1–C9)
  4. Discount the first study. Apply the field's shrinkage factor to any unreplicated effect before acting on it. (C2–C3, C7)
  5. Check the registry. No preregistration plus a flexible design means Low certainty by default. (C12, C19–C20)
  6. Weight by funding and by contradiction. Industry-favourable conclusions and high contradiction ratios each cost one certainty level until independently confirmed. (B5, B24)
  7. Treat dormancy as information. A long-dormant paper may be premature or wrong; check whether its prerequisites have since arrived. (A14, A16)

The aggregator: Ground News, rebuilt for research

Ground News runs four layers — cluster, rate, own, reveal. Each has a science translation, and the building blocks for all four already exist. What doesn't exist is the integration.

Layer 1

Cluster

Ground News: same story, many outlets

Science needs claim-level clustering: every paper, preprint, review, and replication addressing one claim, linked into a maintained claim graph with stable claim IDs. OpenAlex, Semantic Scholar, and scite's citation contexts are the parts; the graph itself is the missing piece.

Layer 2

Rate

Ground News: three outside raters, averaged

Rate with methodological raters, not political ones: GRADE certainty per outcome, RoB 2 / ROBINS-I per study, funding and COI flags, scite support/contradict tallies. The aggregator must never rate in-house — it aggregates named, auditable raters and shows where they disagree.

Layer 3

Own

Ground News: who owns the outlet

Funder + institution + author-network mapping: who funded the study, who employs the authors, who owns the journal, and which author clusters dominate the claim's citations. Most of it already sits in Crossref, ORCID, and OpenAlex metadata — never surfaced to readers in one place.

Layer 4

Reveal

Ground News: the Blindspot

Surface asymmetric attention: claims heavily cited but never replicated; replicated only by the originating lab; null results registered but never published; retracted papers still being cited. A Blindspot for science shows the file drawer and the unreplicated.

Requirements the evidence imposes

The honest gaps

What the research could not establish — stated as plainly as the findings.

The experiment that has never been run

Is there any empirical work on individuals predicting specific findings of papers they have not read? None was found. Extensive searching turned up group-level convergence (multiples), aggregate forecasting (markets), and literature-based discovery — but not the mission's precise proposition. A preregistered experiment is straightforward to design — elicit hypotheses from domain-adjacent non-readers, then score them against the literature — and it has apparently never been run.

The first run flagged one caveat: such work might exist under terminology not searched (for example “precognition” in parapsychology) — parapsychology meta-analyses were deliberately excluded as outside the peer-reviewed mainstream framing. That follow-up has now been run (~50 entries across precognition, panpsychism, and the bridge literatures, Nov 18, 2026), and it does not close the gap: the precognition literature tests prediction of random future stimuli, never of specific empirical findings, and panpsychism predicts nothing about who will guess which result. The gap research ↓

Synchronicity: no falsifiable formulation, no confirming test — Jung's own included. Without a testable formulation the concept remains outside science, however meaningful the coincidences feel.

Open questions the run leaves behind

  1. What is the base rate of “intuitive rediscovery”? Sleeping beauties are common (A16) — how often does an apparently original idea already exist in dormant literature? A bibliometric study linking idea-generation experiments to the dormant literature would quantify it.
  2. Can ML forecasting be fixed? SCORE's three ML systems failed while humans hit 76–78%. Whether modern models with citation-context features can beat that benchmark is open — and the benchmark itself is anchored to a ~50% base rate.
  3. What are the true flag thresholds? The contradiction-ratio and citation-concentration rules in the framework are calibrated to single studies; their behaviour at scale is unknown.
  4. Does a bias aggregator change behaviour? Ground News has no published evidence that its bias display improves reader accuracy. A science aggregator would need outcome studies: do users cite better, retract faster, fund more wisely?
  5. Can claim-level clustering be maintained? The claim graph is the hardest infrastructure piece; no public system yet maintains stable claim IDs with replication and contradiction links.
  6. Can credit be reformed? Given the Matthew effect and three centuries of priority disputes, can CRediT, open notebooks, and timestamped preregistration reduce priority conflict — or merely move it?

The gap research: precognition, panpsychism, and the bridge

The territory the first run deliberately excluded, searched Nov 18, 2026 — ~50 entries across three domains, each carried with its strongest critique. (The request said “panpsychology”; no established discipline exists under that name distinct from panpsychism, so it is treated as panpsychism and its psychological angles.)

The headline: the gap is not closed

Nothing in this territory — not the precognition meta-analyses, not panpsychism, not the bridge literatures — is empirical work on individuals predicting specific findings of papers they have not read. That study has still not been done, in any framing, under any terminology found. Precognition research tests prediction of random future stimuli (pictures, words, RNG outputs), never of specific empirical findings. Panpsychism is a metaphysical thesis about the ubiquity of experience; by its leading defender's own admission it is untestable, and it predicts nothing about who will guess which result. The bridge literatures explain how convergence can appear to happen without anomalous access — forgotten prior exposure, tacit absorption of a field's regularities, fast cue-based inference — but none of them tests prediction of unseen findings either.

The closest mundane analogue is the forecasting literature (DellaVigna & Pope 2018): experts predicting outcomes of experiments whose designs they were shown — a different and much easier task.

Domain A · Precognition and anomalous cognition — claim, then critique

Bem (2011) — “Feeling the Future”
JPSP · 9 experiments, 1,000+ participants
Claim: time-reversed psychological effects — responses recorded before the stimulus was randomly selected — significant in 8 of 9 experiments; combined z = 6.66, mean d = 0.22 [published]; erotic images guessed at 53.1% vs 50% chance [reported]. Critique: failed in direct preregistered replications (below); Alcock documented mid-study protocol changes, uncorrected multiple tests, and effect sizes shrinking as samples grew (r = −0.91 [reported]). Never retracted — and widely credited with accelerating psychology's preregistration reform.
Wagenmakers et al. (2011) — the Bayesian reply
JPSP, same issue
Claim: with default Bayesian tests, Bem's evidence is weak to nonexistent — of 10 critical tests, 1 substantially favoured psi, 3 favoured the null, 6 were anecdotal [reported]; a meta-analytic Bayes factor of ~40 is negligible against precognition's prior implausibility. Reply: Bem, Utts & Johnson argued the prior was inappropriately diffuse. Assessment: the dispute is genuinely about priors — it relocated the argument rather than settling it.
Bem et al. (2015) — meta-analysis of 90 experiments
F1000Research · 33 labs, 14 countries
Claim: overall z = 6.40, Hedges' g = 0.09 [published]; excluding Bem's own experiments, g = 0.06 [published]; fail-safe N = 544 [published]; p-curve estimates the true effect at 0.20 [published]. Critique (Lakens 2015): only 18 of 90 studies were significant, from just 7 labs; his publication-bias-corrected estimate is 0.008, with confidence intervals including zero [reported] — “no good reason to assume that anything more than publication bias is going on.” Venue note: F1000Research uses post-publication open review; the paper appeared there after rejection at conventional journals.
Honorton & Ferrari (1989) — forced-choice precognition
Journal of Parapsychology · 1935–1987
Claim: 309 studies, ~50,000 participants, ~2 million trials [published]; effect size 0.02, homogeneous subset ES = 0.012, fail-safe N = 14,268 [published]. Critique: mean methodological quality 3.3 of 8 criteria [reported]; the database predates modern preregistration and reporting standards — and effect sizes did not grow over 50 years, which the authors themselves called “bad news.”
Storm et al. — free-response & forced-choice meta-analyses
Psychological Bulletin 2010; updates to 2018
Claim: ganzfeld free-response ES = 0.142, hit rate 32.2% vs 25% chance [published]/[reported]; nonganzfeld noise-reduction ES = 0.110 [published]; standard free response null (−0.029 [published]); forced-choice 1987–2010 ES = 0.01 [published]. Critique: Rouder et al.'s Bayesian reanalysis found the evidence collapsed from ~6 billion:1 to at most 330:1 once manually randomized studies were excluded and omitted studies added [reported] — the manual-vs-computerized randomization moderator is the sharpest single methodological point against this literature.
Presentiment — Mossbridge et al. (2012)
Frontiers in Psychology · 26 reports, 1978–2010
Claim: pre-stimulus physiology (skin conductance, heart rate, EEG…) anticipates arousing vs calm stimuli: ES = 0.21, z = 6.9 [published] — and higher-quality experiments produced larger effects, the opposite of the usual artifact signature. The authors preferred “unexplained anticipatory activity” and placed the cause in natural physical processes. Critique (Schwarzkopf 2014): circular inference in fMRI data, ~2:1 target/control imbalances enabling learned prediction, baseline-correction artifacts, and one mechanism spanning measures with vastly different timescales is biologically implausible.
Ritchie, Wiseman & French (2012) — preregistered replication
PLoS ONE · 3 labs
Result: three preregistered exact replications of Bem's Experiment 9, using Bem's own software, all failed — combined n = 150, p = .83 [published], with 99.92% power to detect Bem's effect [reported]. JPSP desk-rejected it (“this journal does not publish replication studies”) [reported] — a canonical illustration of publication bias against replications.
Galak et al. (2012) — seven replications
JPSP · N = 3,289
Result: seven experiments with predetermined sample sizes failed to replicate Bem's Experiments 8–9; across all 19 known replication attempts (N = 4,091), average d = 0.04, CI including zero [published]. A forward control condition produced d = 1.99, showing the apparatus detects the ordinary memory effect [reported]. The authors' summary: the only thing that predicted positive results was whether Bem himself ran the experiment [reported].
Transparent Psi Project & AMP-TPP (2020–2025)
Royal Society Open Science; PLOS ONE 2025
Result — the strongest evidence in the domain, and it is negative. The TPP, an adversarial collaboration with preregistration, consensus protocols, and independent auditing, retested Bem's Experiment 1 across 10 labs: 2,220 subjects, 37,836 trials, erotic images predicted 49.89% of the time — almost exactly chance [verified live]. Walleczek et al. (2025): three studies, 26,483 participants, 420,472 trials — no replication; the largest study found 50.07%, p = 0.496 [reported]. Proponent response: the conditions suppress the effect — unfalsifiable in its strong form, and a concession that the effect is not robust to standard safeguards.
Dream precognition — Maimonides (1964–1972)
Dream laboratory, Brooklyn
Claim: sleeping receivers' dream reports matched randomly selected art prints above chance; Child's 1985 meta-analysis put the combined series at roughly p = 1/1,000 [reported], though some results rested on as few as 8 data points. Status: never independently replicated; a comparable Wyoming protocol found no effect [reported]; isolation of the target from the subject was disputed (Hansel 1985).
Remote viewing / Stargate — Utts vs Hyman (1995)
AIR evaluation of the U.S. government program
Claim (Utts): a statistically significant effect — best-controlled experiments ~34% hits vs 25% chance [reported]; “anomalous cognition is possible and has been demonstrated.” Critique (Hyman): the same judge scored all ten SAIC experiments; never replicated by an independent lab; output “vague, general, and way off target.” Agreed verdict: the AIR report concluded remote viewing produced no usable intelligence — no report ever provided actionable information — and the program was terminated in 1995 [reported].
RNG / PEAR / Global Consciousness Project
Princeton PEAR lab; GCP from 1998
Claim: intention biases random number generators at “just under 51 percent” where 50% is chance [reported]; GCP's formal series (513 registered events, 1998–2015) reached cumulative Z = 7.31 [reported]. Critique: Bösch et al.'s meta-analysis of 380 studies found a small-study effect and extreme heterogeneity — publication bias the most parsimonious explanation [published]; the 9/11 result depends on its chosen time window [reported]; Carroll: deviations before an event count as precognition and after as response, so nothing could count against the hypothesis [reported].
The file-drawer critiques — and the 2026 counter-literature
Francis 2012 · Alcock 2011 · Tressoldi & Storm 2026 · Wiseman & Watt 2026
Francis's Test for Excess Success: Bem's reported success rate (9 of 10) is too high to be credible given his own effect sizes [reported] — a problem he later found in >80% of multi-experiment articles in two mainstream journals, so it is general to psychology, not specific to psi. The field's own 2026 metascience complicates the simple story: published null rates in ESP research run higher than in mainstream psychology [published] — yet in the Koestler registry, preregistered studies produced significant results only 16% of the time vs 46.6% for non-preregistered ones [published]. Preregistration deflates psi results even where the file drawer is smaller than critics assumed.

Claimed effect sizes, by paradigm

What the meta-analyses claim, at face value — against the bias-corrected estimate and the preregistered record. Effect size (ES / Hedges' g) as tagged in the research.

And the preregistered record: Transparent Psi Project hit rate 49.89% on 37,836 trials where chance is 50% [verified live]; Walleczek et al.'s largest study 50.07% [reported]. The debate's current shape: meta-analyses keep returning small positive effects; preregistered replications keep returning nulls.

Forced-choice vs free-response

The design pattern across the whole literature — and why it matters for the core question.

Forced-choice designs (guess one of N fixed alternatives) yield ES ≈ 0.01–0.02 [published]. Free-response designs (describe a rich target) yield ES ≈ 0.11–0.14 [published] — an order of magnitude larger. Standard free response without noise-reduction procedures (ganzfeld, dreaming, meditation) is null.

Proponents read this as: meaningful settings and selected participants are genuinely more psi-conducive. Critics read it as: free-response designs carry more analytic degrees of freedom — judging, target selection, randomization method — and the randomization moderator supports the artifact reading. Note the trap for the core question: free-response is the closest analogue to “predicting a specific finding,” and it is exactly the design most vulnerable to judging flexibility.

Domain B · Panpsychism — a metaphor for convergence, not a mechanism

Panpsychism answers a different question — why experience exists at all — and no position in this literature predicts that one mind will converge on the specific content of another's unread findings. Cosmopsychism comes closest to a convergence-friendly picture (all minds as aspects of one cosmic mind), but offers no channel, no test, and no prediction about specific content. One could accept panpsychism wholesale and still hold that predicting unseen findings is impossible; the two questions are orthogonal, and no panpsychist author surveyed claims otherwise.

Chalmers — the hard problem
The Conscious Mind (1996) · Testable? No
Easy problems (functions) vs the hard problem (why physical processes are accompanied by experience at all). His Russellian monism: physics describes only structure; phenomenal properties are matter's intrinsic nature. For: the zombie conceivability argument. Against: illusionists (Dennett, Frankish) call the hard problem a “hunch.” Makes no empirical prediction distinguishing it from physicalism. PhilPapers 2020: 62.4% of philosophers say the hard problem is genuine [reported].
Strawson — realistic monism (2006)
Journal of Consciousness Studies · Testable? No
A real physicalist must be a panpsychist: experience is real, everything real is physical, and experience cannot intelligibly emerge from wholly non-experiential matter — so fundamental matter is experience-involving. Against: the combination problem; his own “ignorance response” concedes the gap rather than closing it. The argument is a priori and predicts nothing observable.
Goff — Galileo's Error, constitutive panpsychism
OUP 2017 / 2019 · Testable? No — by his own admission
Galileo exiled qualities from science, making consciousness “essentially and inevitably mysterious”; human consciousness is constituted of micro-level consciousness. His answer to subject-summing is “phenomenal bonding,” conceded to be non-transparent. In his own words: “Can panpsychism be tested? No. You can't do experiments to test whether an electron or a field has feelings.” Panpsychism overall: 7.55% of philosophers accept or lean toward it (PhilPapers 2020) [reported].
Nagel — Mind and Cosmos (2012)
OUP · Testable? No
The materialist neo-Darwinian picture cannot explain life, consciousness, reason, or value; his positive proposal is “natural teleology” — immanent principles predisposing nature toward mind. Against: a promissory note — it names what needs explaining, with no mechanism, no predictions, and no experiment proposed.
Cosmopsychism — Mathews, Shani, Nagasawa & Wager
Priority cosmopsychism · Testable? No published test found
Inverts micropsychism: the cosmos as a whole is conscious and prior to its parts, dissolving the combination problem upward. Against: the decombination problem (Chalmers) — how does one cosmic subject yield many mutually opaque finite subjects? The problem is relocated, not solved. This is the report's one point of poetic contact with convergence — and it licenses no inference about information transfer.
The combination problem
James → Chalmers · Unsolved
William James: tell each of twelve men one word of a sentence — the men together do not know the sentence. How do micro-subjects, micro-qualities, and micro-structure yield one unified human experience? Responses — phenomenal bonding, cosmopsychism, emergentist panpsychism, the ignorance response, denying the intuition, IIT's exclusion axiom — none commands consensus, and panpsychism's leading defenders acknowledge it as the central challenge.
Integrated Information Theory (Tononi)
IIT 4.0 (2023) · Testable? Partially — uniquely here
Consciousness is identical to integrated information (Φ); any system with Φ > 0 is conscious to some degree — a structured panpsychism Tononi accepts. Its clinical descendant (PCI) stratifies unresponsive patients [reported]. Against: Φ is computationally intractable for real brains; Aaronson's reductio (an inactive XOR-gate grid could out-Φ a brain); a 2023 open letter from 124 scholars called the theory untestable [reported]; the 2025 Nature adversarial collaboration challenged key tenets of both IIT and its rival GNWT [reported]. The panpsychist implication (low-Φ systems slightly conscious) remains untestable even where the theory's other predictions are not.
Orch OR (Penrose & Hameroff)
Physics of Life Reviews 2014 · Testable? Yes, in principle
Consciousness from orchestrated quantum coherence in neuronal microtubules, ending in gravity-mediated objective reduction. The most empirically engaged theory in this domain: recent work finds anesthetics reduce energy diffusion through microtubules by 12–15% [reported] and a microtubule-stabilizing drug delayed loss of consciousness in rats by ~69 seconds (n = 8) [reported]. Against: Tegmark's decoherence calculation (~10−13 s in the warm brain) [reported]; the Gran Sasso experiment ruled out the simplest Diósi–Penrose collapse model the theory uses [reported]; nothing yet shows microtubule quantum effects produce consciousness.
The critics — Dennett, Churchland, Papineau, Searle, Seth
The mainstream reply
Dennett: the hard problem is a “hunch”; explain the illusion of phenomenality instead. Churchland: folk psychology is a false theory a mature neuroscience will replace; the hard problem is premature. Papineau: the gap is epistemological, not ontological — though he concedes his view is “not unlike panpsychism” in refusing a bright line in nature [reported]. Searle: panpsychism “does not get up to the level of being false” [reported]. Seth: “it explains nothing and does not generate testable predictions” [reported].
Empirical angles claimed for panpsychism
Bottom line
Direct testability: none. Panpsychism is compatible with any physical observation because it concerns a hidden aspect of matter; Schwitzgebel argues the standard correlational method cannot even establish whether a snail is conscious [reported]. The adjacent empirical programs (IIT, Orch OR) test theories of consciousness, not panpsychism as such. No empirical angle in this literature tests — or could test — whether one person will converge on the specific content of another's findings.

Domain C · The bridge — mundane explanations and near-misses

Cryptomnesia — the leading mundane explanation
Brown & Murphy (1989) · Source monitoring (Johnson et al. 1993)
Recalling information from an external source while experiencing it as your own original idea — a source-monitoring failure. In the lab paradigm, participants inadvertently reproduced others' contributions in 3–9% of cases [published], most often from the person who spoke immediately before them. Content outlasts context: delay, incidental exposure, and cognitive load all raise the rate. This is the single most important mundane account of “intuiting” a published result — an abstract skimmed, a talk half-heard, a corridor conversation, source tag lost, content kept. The phenomenology is exactly “it just came to me.” Any decisive test must audit prior exposure; without that control, cryptomnesia is an unexcluded confound in every convergence anecdote.
Implicit learning & tacit knowledge
Reber (1967; 1989) · Polanyi
People absorb the hidden structure of an environment — Reber's artificial grammars — and classify novel cases accurately without being able to state the rules [published]. “We know more than we can tell.” A researcher immersed in a field absorbs which hypotheses are plausible and which effects “should” exist, and can generate a correct-feeling prediction from that tacit base. Genuine knowledge — but knowledge of the field's structure, acquired through exposure, not of a specific unseen finding.
Thin-slicing (Ambady & Rosenthal)
Psychological Bulletin (1992) · 38 studies
Brief slices of expressive behaviour (<300 s) predicted interpersonal outcomes at mean r = .39–.41 [published]; 30-second silent clips of teachers predicted end-of-term evaluations at r = .76 — but with n = 13 teachers and no large independent replication located [reported]. Fast inference from minimal cues is real but moderate, uneven by trait, and partly shared cultural bias. It concerns people, where valid cues exist — no precedent for inferring an unseen finding, where no cues exist at all.
Intuitive expertise (Kahneman & Klein 2009)
American Psychologist · Adversarial collaboration
Skilled intuition is recognition, and it develops only when both hold: a high-validity environment with learnable cues, and prolonged practice with rapid, unambiguous feedback. Predicting specific findings of unread papers fails the first condition structurally — random assignment, sampling noise, and investigator choices are not inferable from a topic — and usually the second as well. The framework therefore predicts confident-but-unreliable intuition for exactly this task, and warns there is no subjective marker separating correct intuitions from confident wrong ones.
Morphic resonance (Sheldrake) — with appraisal
A New Science of Life (1981)
Claim: non-local “morphic fields” carry the habits of past systems, so forms become more probable the more often they occur — the closest thing in this whole report to a proposed mechanism for cross-person convergence. Appraisal: widely classified as pseudoscience; no mechanism specified; independent staring experiments found nothing beyond chance and identified design flaws [reported]; no supporting experiment has been replicated by independent skeptical researchers with adequate controls [reported]. Even if granted, it predicts convergence on repeated forms, not prediction of novel specific findings.
Synchronicity — summary only
Covered in the first run; not re-litigated
Jung's own statistical test — 483 married couples against ~32,220 recombined “false couples” [reported] — failed: each batch produced a different “hit” aspect and the pooled result was not significant; Jung himself concluded statistics could not establish it. Zeng (2025): “synchronicity cannot be verified by natal astrology” [published]. It remains a clinical/philosophical concept, not an explanation of convergence.
“Anomalous cognition” as a term
May, Utts & Spottiswoode (1995) · May (2010)
Coined as a mechanism-neutral replacement for ESP / remote viewing [published] — and defined negatively: May concedes “we do not have a positive definition of AC” [published]. May & Marwaha's own summary concludes the phenomenon “resists training for excellence (and replication studies will not yield results)” [reported] — by its chief proponent's account, not a reliable channel, which undercuts its usefulness as an explanation of systematic intuition-convergence.
Forecasting experimental results (DellaVigna & Pope 2018)
Journal of Political Economy · The closest mundane analogue
208 academic experts [published] forecast results of 15 treatments in a real-effort experiment (~10,000 subjects), given 3 benchmark treatments to calibrate: the average expert forecast closely predicted the results and outperformed 96% of individual forecasts [published]; citations, rank, and field did not predict accuracy. Limitation: forecasters were shown the designs and benchmarks. No study in this literature asks anyone to predict findings of papers they have not read — informed forecasting from design knowledge is a different task from anomalous access to unseen results.

Synthesis: face value vs the critics

Taken at face value, the meta-analyses imply a genuine, small, domain-general anomalous anticipation effect — forced-choice ES ≈ 0.01–0.02, free-response ≈ 0.11–0.14, presentiment ≈ 0.21, Bem-paradigm ≈ 0.09 [published] — which would make anomalous cognition a live candidate mechanism for intuition-convergence: “intuiting” a paper's finding could, in principle, be access to a future or distant state of affairs through the same channel. The critics' reply, in descending order of force:

Net: taking the meta-analyses at face value requires discounting the replication record; taking the replication record at face value requires treating the meta-analyses as artifacts. Both readings are internally coherent — the replication record is the more recent and methodologically stronger body of evidence, and it is negative. As of 2024–2026 the debate has shifted from “is there a meta-analytic signal?” (there is, in the published literature) to “does the signal survive preregistration and adversarial replication?” (so far, no) — and the field's own metascience now argues about how much publication bias exists rather than whether the effect is real.

What a decisive test of the core question would look like

Designed to the standards this literature's history shows are necessary — the standards whose absence produced 90 years of inconclusive meta-analyses:

  1. Preregister, adversarially. Predictors, skeptics, and proponents agree the protocol in advance (the TPP model); analysis plan frozen; data open; independent audit.
  2. Choose targets predictors cannot have seen. Specific, quantitative findings from papers under embargo, from non-English journals outside the predictors' field, or from a registered-report pipeline before results are known — plus fabricated control findings (plausible but false) to measure base-rate guessing and cryptomnesia.
  3. Audit exposure — the cryptomnesia control. Document each predictor's prior contact with the topic: publications, conferences, social-media exposure, and a recognition test for the source material. Lab source amnesia runs 3–9% even under controlled conditions; in the wild it is presumably higher.
  4. Run a mundane baseline. A comparison arm of field experts predicting from design knowledge alone (the DellaVigna & Pope model). If predictors without design knowledge match experts with it, that is the anomaly; if they match chance, there is none.
  5. Pre-specify the scoring. Exact quantitative predictions — direction plus magnitude band — scored against published values, with the chance baseline computed from the empirical distribution of findings in the pool, not from 50%.
  6. Power it, replicate it, publish regardless. Sized to detect the effect the precognition literature claims (d ≈ 0.1–0.2): hundreds of predictors, thousands of predictions, a built-in replication sample, and a commitment to publish either way — the registry lesson.
  7. Test the boundary conditions. Vary expertise and topic familiarity. If accuracy appears only where valid cues and feedback exist (Kahneman & Klein), the mundane explanation wins by the framework's own logic.

Decisive for: above-chance quantitative prediction of specific unseen findings, surviving the exposure audit, beating the mundane baseline, replicating independently. Decisive against: chance performance in a well-powered preregistered test — the outcome the TPP and Walleczek studies already delivered for the adjacent, and easier, precognition claim.

Open questions the gap research leaves behind

  1. Why do the two literatures diverge? Psi meta-analyses persistently return small positive effects while preregistered replications return nulls. Is the residual entirely publication bias and QRPs (Lakens, Francis), or a small real effect that adversarial conditions suppress (Tressoldi, Radin)? The 2026 metascience sharpens but does not settle this.
  2. The Walleczek Study 2 anomaly. A below-chance result (49.65%, p = 0.013 [reported]) that replicated once, then didn't: fluke, artifact, or something else? A case study in why single replications are insufficient.
  3. Can “anomalous cognition” get a positive definition? May (2010) concedes it does not have one. A field that defines its phenomenon negatively — “information transfer we can't explain” — cannot accumulate theory.
  4. The combination problem. Panpsychism's central unsolved problem; cosmopsychism relocates it as decombination. No response commands consensus.
  5. IIT's and Orch OR's empirical fates. Whether IIT 4.0 survives its next adversarial round after the 2025 Nature results, and whether the 2023–2025 microtubule findings replicate and connect to consciousness at all.
  6. The core question itself. No study has asked people to predict specific findings of papers they have not read. Until one does — with the controls above — every claim about intuition-convergence, anomalous or mundane, rests on anecdote and extrapolation.

Could not verify — carried from the research