Wiki

Wiki: How to read source types, status labels, and evidence classes

Auxiliary page to avoid confusing badge, source, status, and what a paper directly strengthens

Mind Uploading Research Project

Public Page Updated: 2026-03-26 Reading guide

How to use this page

Read this first to avoid getting lost

This page is an auxiliary page that organizes labels such as ``Scopus,'' ``arXiv,'' ``Review,'' ``Media,'' ``source_logged,'' ``curated,'' and evidence classes that appear in collections and bibliographic maps for beginners. The important thing here is to read these labels separately and understand what information each one represents.

  • We explain publication source, document type, site status, and evidence class as separate axes.
  • We show how to separate source labels from the question of what a paper directly strengthens.
  • You can use the table to see where to stop and return to the original text.
  • Human measurement papers are kept separate as observability-class advances rather than being folded into demo or hidden-state bins.
  • Language-facing demos are read through route split and neural-contribution audit, not as one solved brain-to-text category.
Helpful next pages
Collection of papers Literature map Casework
Best for
People who don't want to misread labels on collections of papers or bibliographic maps, and people who want to know where to go back to the original text.
Reading time
8-12 minutes
Accuracy note
Labels are a reading aid and do not automatically guarantee the correctness of the article content. When using them as evidence, always refer back to the DOI, main text, and primary research.

Relatively clear at this stage

What we know now

  • Even for the same document, publication source, document type, site status, and evidence class must be viewed separately.
  • Scopus is an index, arXiv is a preprint storage, and Review is a document type, and they do not have the same meaning.
  • source_logged and curated are labels that indicate how organized this site is.
  • A human PET/MRS/MRI paper can strengthen observability without closing hidden state or proving causal sufficiency.
  • A language demo can strengthen semantic reconstruction, fixed-segment retrieval, known-onset decoding, prompt-conditioned generation, or communication performance without becoming unrestricted thought reading.

Still unresolved beyond this point

What we still do not know

  • The extent to which individual papers ultimately remain central evidence will depend on subsequent scrutiny.
  • Top stories traced from reviews and news may weaken when returned to primary research.

Learn the basics

Check the basics in the wiki

What the wiki is for

The wiki is a learning aid. For the project's official current synthesis, success criteria, and operating rules, always return to the public pages.

First divide into four parts

What can be confusing on literature pages is that where a paper is listed, what type of literature it is, its current status on this site, and what it directly strengthens can all appear to be the same thing. By separating these four things first, you can significantly reduce misreadings.

Labels have four axes

axis What does it represent Example That's not all I can say
Publisher/Badge It's about where you can trace the document and which entrance you picked it up from. Scopus / arXiv / Academic journal page / DOI It is not determined that the content is strong, correct, or central.
Document type Whether the document is primary research, a review, or news. Primary / Review / Media It is necessary to separately confirm whether there is new data or whether it is strong enough to be used directly for verification.
Site status Indicates whether input has been accepted, organized, or excluded on this site. source_logged / curated / noise_excluded It does not necessarily mean that the content of the paper is true or of low value.
Evidence class Indicates what the paper directly strengthens in the technical reading flow. Direct validator / system demo / observability-class advance / benchmark / hidden-state boundary It is still not a claim that the whole problem is solved. It only fixes which evidential axis moved.
Multiple attributes overlap in the same document

For example, a document may be a review article on Scopus, a preprint of primary research on arXiv, or a curated human-MRI paper that belongs to the observability-class-advance bin. Do not rely on only one label.

How to read publication source and type

Display In everyday language Useful points Notes
Scopus These are documents found via the academic literature index database. It is easy to organize bibliographic information and becomes an entry point for picking up a wide range of academic literature. Scopus is an index and does not automatically guarantee the strength of peer review or the degree of direct connection to the issue.
arXiv This is a public storage area for preprints. You can find new research quickly. Many papers have not yet been peer-reviewed at the time of submission, so you need to confirm the methods and limitations yourself.
Review This is a review article that summarizes multiple studies. It is suitable for grasping the map of the entire field and the main points of discussion. Since our own new experiments are not always the main character, we return to primary research when using it as strong evidence.
Media News articles, explanatory articles, and general introductions. It can be used as an entry point to a topic or as a clue for searching for primary literature. It is weak as an academic basis as it is, and it is necessary to trace it back to the original paper or presentation.
Primary Primary research that directly produces new data, experiments, and analyses. The evidence can be traced most directly by reviewing the method, evaluation, and limitations. Even primary research can have weak sample conditions, statistics, and reproducibility, so it cannot be used unconditionally.

How to read evidence class

Evidence class What it directly strengthens Typical examples What it still does not justify
direct validator / causal calibration Error sources, calibration limits, and local ground-truth checks. Mikulan (2020), Unnwongse (2023), Hao (2025) Whole-brain uniqueness recovery or full hidden-state closure.
task-limited system demonstration Task-conditioned language decode, prompt-conditioned generation, phoneme-sequence decoding, or closed-loop communication performance on a specific scaffold and interface. Tang (2023), Défossez (2023), d'Ascoli (2025), Ye (2025), Willett (2023), Littlejohn (2025), Wairagkar (2025), Singh (2025) Whole-brain emulation, unrestricted thought reading, identity continuity, or hidden-state completeness.
dataset / benchmark / standard / toolchain Comparability, synchronization, governance, and reproducibility. EEG-BIDS, Motion-BIDS, LSL, MOABB Biological sufficiency or mechanistic truth.
observability-class advance / human proxy ladder What humans can directly observe or approximate at a specific measurement class. Shapson-Coe (2024), Johansen (2024), Lucchetti (2025), Baadsvik (2024), Rzechorzek (2022), Hirschler (2025) State-complete measurement. Each proxy or atlas still has a claim ceiling.
mechanistic boundary / hidden-state evidence What still remains latent or omitted even after connectome, cell type, or a strong demo. Gouwens (2021), Hengen (2016), Xu (2024), Looser (2024), Cahill (2024) Direct validation or a finished implementation.
review / synthesis Field map, terminology, and issue clustering. Review articles and benchmark syntheses Strong conclusions without returning to primary research.
context / philosophy / law / culture Context around the topic and its surrounding debates. Ethics, legal analysis, metaphysics, cultural studies The technical or natural-science frontier by itself.
Language-facing demos need one more split

On this site, `task-limited system demonstration` is still too broad if readers treat every language-facing result as one `brain-to-text` category. Tang et al. (2023) constrain within-subject semantic reconstruction, Défossez et al. (2023) constrain fixed-segment speech retrieval, d'Ascoli et al. (2025) constrain known-onset word decoding, Ye et al. (2025) constrain prompt-conditioned generation, and Singh et al. (2025), Willett et al. (2023), Littlejohn et al. (2025), and Wairagkar et al. (2025) constrain different invasive speech-decoding or communication-subsystem routes. Therefore, when a paper emits fluent text or speech, this site asks for the Neural Contribution Card before it is promoted beyond task-conditioned evidence.

How to read site status labels

Label In everyday language What you need next Easy to misread
source_logged This is the stage where we accept URLs and DOIs as entry points and include them in our tracking targets. Relevance confirmation, primary literature tracking, U number assignment, and abstract scrutiny are required. It does not mean "recruitment confirmed" or "high quality confirmed".
curated This site has been organized and arranged according to the points of discussion. It will continue to be necessary to eliminate duplicates, replace with strong evidence, and track updates. It is not a label that guarantees that the content is true.
noise_excluded We are at the stage where we have determined that the relationship is weak, the contamination is large, and we will not use it at this time. It is important to leave the reason for exclusion and prevent contamination of the same species. This does not mean that the paper itself has no value, but it may be out of sync with the main points of this site.

Where to stop and return to the original text

Current purpose You can stop here Return to original text
I want to pick up a wide range of topics Once you know whether you are interested by Badge, summary, or 5-point arrangement. When you want to support a strong argument by citing the paper.
Looking at source_logged items When you understand that you are still at the entrance stage. When you want to judge whether you can pursue primary research or whether it will be included in the U number.
Looking at Reviews and Media When you have a map of the issues and related keywords. When you want to use evaluations, methods, and numerical values as evidence.
Comparing documents from Scopus and arXiv When you can figure out which entrance it was picked up from. When you want to check the peer review status, experimental conditions, limitations, and possibility of follow-up testing.

Common misreadings

Misread

  • “Scopus is strong”: Scopus is an index. Check the text and method to see how it applies to the issue.
  • “It's all weak because it's arXiv”: There are important entry points even before peer review. Check the content to see if it's weak or not.
  • “Review is enough”: Review articles are useful as maps, but for direct evidence you need to go back to primary research.
  • “Source_logged, so accepted”: This is still an acceptance log, and the decision to integrate or exclude is a follow-up work.
  • “curated so it is true”: Being organized and ultimately correct are two different things.
  • “A human proxy paper solved hidden state”: A human PET/MRS/MRI advance often raises an observability class, not a full state-complete readout.

Where to return next

If you want to have a broad view of the papers, go back to Collection of Papers, if you want to see how they correspond to unresolved issues, go to Literature Map, and if you want to organize them based on the differences in their roles, go back to How to read the literature and evidence page.