For researchers

Method and limitations

This page sets out how the corpus is built, on what criteria a document gets in, how retrieval works and what known defects it carries. The last part matters as much as the first: a documentary tool that does not state its limits asks you to trust it, and the point here is the opposite.

90,000+documents
550,000+indexed passages
300,000+concepts
400,000+relations

Figures read from the graph itself when the page loads. The rest of the data is as of 7 September 2026.

Where it comes from, and what gets in

The corpus is fed by three automated routes and one manual one, all of them daily. None of them searches for “whatever exists about drugs”: each walks forty-four axes defined by discipline, so the whole does not end up being pharmacology with trimmings.

Route Documents What it contributes, and why
Crossref39,374Everything with a DOI: humanities, law, botany
PubMed / Europe PMC5,385Peer-reviewed biomedical literature
OpenAlex2,538Social sciences and fields with no biomedical coverage
Own library and archive3,286Grey literature: reports, proceedings, theses, work without a DOI
Cannabis Magazine700Own back catalogue since 1997

Inclusion criteria. From the automated routes, only work with a deposited abstract gets in — without text there is nothing to index — and only what passes a relevance filter of field terms applied to title and abstract. Anything already held is dropped by comparing DOI, address and normalised title. There is no impact or journal-prestige criterion: a report from a risk-reduction organisation weighs the same as a paper in Nature, and the reader sees which one each claim comes from.

A document gets in if it deals with a psychoactive substance or the field around it: pharmacology, drug policy, harm reduction, addiction, forensic toxicology, the anthropology and history of use. In September 2026 everything collected up to then was reviewed against that criterion and 62,929 documents were set aside: they had got in on a single generic word in their abstracts — “ritual”, “conservation”, “taxonomy”, “drug” — and were not about psychoactives at all: philosophy of law, agronomy, materials chemistry. Nothing was deleted, but they are neither searched nor counted. Material added by hand (library, archive, books and the magazine back catalogue) does not go through that filter: there a person made the selection.

How it is indexed

  1. Segmentation

    Each document is split into overlapping passages of a manageable size. The overlap keeps a claim from being cut in half and lost to retrieval.

  2. Dual index

    Each passage is indexed by meaning, using a multilingual embedding model running on our own server, and by literal text. The first finds what is said in other words; the second finds the acronyms and proper nouns an embedding dilutes.

  3. Graph extraction

    Concepts are extracted from the passages — substances, receptors, brain regions, effects, clinical conditions — along with the relations between them and the textual evidence that supports each one. That is what lets a question about a receptor also surface what has been written about the plant that activates it.

  4. Authors and full text

    Recorded coverage varies between documents. The corpus may contain abstracts or other passages; a full-text flag alone does not establish that the publication is complete. To answer, the model receives only the retrieved passages and the references shown. Check the original document to assess coverage.

How retrieval works, and why that bounds invention

A question triggers not one search but several. The query is rewritten and translated into the corpus languages; retrieval runs in parallel by meaning and by literal match; the two rankings are fused; and a local reranking model judges the candidates again on content rather than language, because an embedding scores text in the question's own language higher and that was skewing answers toward the Spanish corpus.

The model is instructed to write from the retrieved passages and cite the corresponding sources. It may misinterpret a source, add unsupported claims or omit information. A linked citation lets you check the text; it does not by itself demonstrate that the text supports the claim.

There are two more modes besides the written answer. “Search documents” returns the list of documents without writing anything, with filters by year, evidence level, source, author and journal, and can be exported to BibTeX, RIS or CSV: it is the same retrieval as above, minus the final step. And the “extended answer” (with an account) retrieves fourteen sources instead of six and writes section by section, for someone doing a review rather than settling a quick doubt.

Study types: classification and limitations

The filters place documents in five groups: evidence syntheses; experimental studies; analytical observational studies; descriptive studies and series; and preclinical research and opinion. These are the tool’s document groups, not quality scores. Answer sources represent retrieved passages, and one document may contribute several.

Noosphere shows the classification recorded for the document in the corpus. Check that grouping against the original publication: a label alone does not verify the actual design, conduct or relevance to the question. Generating an answer does not perform a new individual methodological assessment.

Documents without a recognised group are shown as unclassified. Clinical groupings may be unsuitable for analytical chemistry, botany, ethnography, law or history. A missing label does not establish that the document lacks value.

These five groups are not a GRADE assessment. Certainty of evidence is assessed for each outcome and considers, among other factors, risk of bias, inconsistency, indirectness, imprecision and publication bias. A review or trial does not guarantee high certainty. See the Cochrane Handbook on certainty of evidence.

What coverage there is

By decadeThe distribution by year is uneven and changes as documents are added. The public corpus breakdown shows the years declared in the metadata; an early or future date should be checked against the original document.
By discipline14,851 documents carry a discipline from the OpenAlex classification, spread over 26 fields, led by medicine, psychology, neuroscience, pharmacology and toxicology, and the social sciences. The rest come in through the thematic axes of collection: mycology, risk reduction, ethnobotany, analytical chemistry, phytochemistry, religious studies, sociology and toxicology, among others.
By journalClose to 12,000 distinct journals after merging the variant spellings each source uses for the same title. The largest contributor, Addiction, is under 4% of the whole.
DepthCoverage of a publication may be partial. Public bibliographic records and query-corpus records are separate inventories: inclusion does not establish access to the full article, methodological appraisal or exhaustive coverage of a topic.

Known limitations

What follows are not courtesy warnings: they are measured defects, with their magnitude. Anyone using this tool for research should know them before citing anything.

Severe language bias70,591 documents in English against 1,652 in Spanish and token figures in Portuguese, German and French. 96% of the corpus comes from the anglophone tradition, with everything that implies about which questions get treated as researchable. Retrieval is multilingual, but it cannot retrieve what is not there.
Reliance on abstractsCoverage varies between documents: some provide a title, abstract and metadata; others have a full-text flag in the corpus. The model receives retrieved passages. An abstract may include methods and results while omitting details needed to appraise the study. The document flag alone does not verify that the full publication is held. Check the original before relying on it for a conclusion.
Metadata that arrives wrong at sourceA 1960 paper on psilocybin is listed as published in Corrosion Science. We checked it against the Crossref API: the error is in the publisher's deposit, not in our matching. There are a handful of cases like this and they cannot be fixed from here.
Our own mismatchesWhen assigning DOIs to archive documents that lacked one, twelve came out wrong: two papers on Banisteriopsis caapi point to a course on vegetative propagation. They are identified and flagged for manual review.
Inclusion is by keywordA term filter does not understand what a text is about. Measured against the body of the documents, around 15% mention no substance or discipline from the field; many are neuroscience method papers that serve as background, but others simply do not belong. The bias runs both ways: relevant material written in another vocabulary is also discarded.
Publication and recency biasThe corpus inherits what the literature publishes: positive results over negative ones, and almost half the documents dated in this decade. A field that reawakened fifteen years ago produces a corpus that looks more like the present than like its own history.
The answer is not reproducibleRepeating the same question may retrieve different sources or generate different wording. A query link does not archive the answer or freeze the corpus or model. Bibliographic references can also be corrected: retain the details you consulted and check claims against the original publication.
No peer review and no expert curationNobody reviews what comes in document by document. The controls are automated — relevance, duplicates, integrity — and audited periodically, but they do not amount to the curation of an institutional repository. This project is not run by a university.

How to cite it

To support a claim, consult and cite the original publication: authors, title, year and DOI or URL. A citation in an answer does not show that the source supports the claim. When describing your use of the tool, identify Noosphere, Psiconáutica, its URL and your access date; disclose that use according to your institution’s or journal’s rules.

“Link to this query” includes the question and the response-mode and study-type settings selected when copying it. Check those controls: they are not bound to the original answer’s settings. Opening it may show a prepared answer; to generate a new one, submit the question. The link does not archive the answer or its sources, or identify a corpus or model version. A new query may produce a different result. Share it only if you want the question to be visible.

To document a search, record its date and time with time zone, exact question, language, mode, applied filters and retrieved references. Retain the passages you checked and state whether you consulted an abstract or the full text. You can export references from Search documents; a bibliographic export does not archive the session. If a corpus update date or count is shown, record it as descriptive information: it is not a version identifier and cannot reconstruct the selection. If a detail is not shown, record it as unavailable. The Cochrane Handbook, section 4.5, explains how to document a search process; using Noosphere does not amount to conducting a systematic review. Each answer offers a local research-record download containing the question, received text and available sources. For new queries it retains the submitted options; for prepared or history answers, unavailable original options remain unknown. It also distinguishes incomplete reception from scientific validation and may contain sensitive information: review it before sharing.

Psiconáutica (2026). Corpus documental multidisciplinar sobre sustancias psicoactivas (Noosphere). https://brain.psiconautica.org/corpus.html

If you spot an error, say so

A wrong metadata field, an impossible attribution, a document that should not be there: write to noosphere@psiconautica.org. We are equally interested in literature indexed nowhere that is being lost: grey material, proceedings, work that was not published in English.

Corpus fact sheet Try Noosphere

Noosphere is a document research tool by Psiconáutica. References let you check an answer; they do not guarantee correctness or exhaustive coverage.

How to verify and cite · Corpus fact sheet · Method and limitations