Atlas Specification

Governing principle

The atlas is claim-and-relationship centric, not merely publication centric. A paper may support several claims at different evidentiary levels, and a later paper may extend one claim while constraining another.

Core node types

study: empirical paper or formal empirical project.
theory: explicit theoretical model or synthesis.
claim: narrow proposition attributable to one or more sources.
finding: directly reported empirical result.
constraint: null, mixed, contradictory, boundary-setting, or non-superiority result.
construct: measured or inferred psychological/biological concept.
method: measurement, statistical, or experimental method.
intervention: deliberate attempt to alter a mechanism or outcome.
population: the group actually studied.
person: researcher/collaborator relevant to intellectual or methodological development.
project: grant, consortium, lab program, or funded initiative.
source: publication, institutional page, conference abstract, dataset, or evidentiary record.
fork: one unresolved problem generating two or more sustained research lines.
cascade: finding -> unresolved question -> targeted test -> later consequence.
convergence: previously separate streams become jointly relevant to a later claim/program.
inference-gate: transition requiring a qualitatively stronger evidentiary basis.

Relationship types

documents
supports
constrains
contradicts
extends
replicates
fails-to-replicate
predicts
operationalizes
tests-mechanism-of
manipulates
intervenes-on
motivates
converges-with
contextualizes
measures
samples
funds
introduces-method

Relationship record

Every consequential edge should eventually have its own record containing:

source node
target node
relationship type
confidence
basis for connection
supporting citation/source
author-explicit vs analyst-reconstructed
date added/reviewed

Connection confidence

documented: source explicitly identifies the connection.
strongly-inferred: chronology, constructs, authorship, and research question strongly support the relationship, but the explicit statement has not yet been located.
tentative: analytically plausible; must not be rendered as established genealogy.

Verification levels

bibliographic: title/authors/date/DOI verified.
abstract-verified: key claims checked against abstract/official record.
fulltext-verified: methods, results, and limitations checked in full text.
replication-checked: related confirming/contradictory literature explicitly reviewed.

Inference gates

Association
Temporal prediction
Candidate mechanism
Experimental causal manipulation
Intervention efficacy
Distal clinical/behavioral outcome
Durability
Real-world effectiveness / translation

No arrow may silently cross a gate.

Mandatory empirical-node fields

Every study note should record:

publication dates (online-first, issue, conference where relevant)
research question
sample and population
developmental stage
constructs and construct-specific roles
design/method
direct findings
constraints/null findings
what the study establishes
what it does not establish
prior unresolved question: author-stated vs analyst-reconstructed
downstream question: author-stated vs analyst-reconstructed
endpoint type
evidence class
verification level
source(s)
relationship confidence

Reader-depth rule

There are no person-specific versions. Each node exposes the same factual core with progressive depth:

Question
Finding
Why it matters
Method
Construct status
Inference boundary
Alternative explanations
Downstream genealogy

Simplification may remove detail but may not strengthen a claim.