The Universal & Atomic Elements of Organization
A falsifiable scientific theory proposing that the same irreducible structures organize both thought and reality — from the quantum to the social scale.
O-Theory proposes that four irreducible structures — Distinctions, Systems, Relationships and Perspectives (DSRP) — form the universal grammar of organization. Rather than being merely useful ways of thinking, these structures are hypothesized to be the atomic elements from which every organized phenomenon emerges, whether in cognition, biology, physics, mathematics, society, or the cosmos.
| D | S | R | P |
| Distinctions | Systems | Relationships | Perspectives |
| Identity ↔ Other | Part ↔ Whole | Action ↔ Reaction | Point ↔ View |
| D := (i ↔ o) | S := (p ↔ w) | R := (a ↔ r) | P := (ṗ ↔ v) |
This is a living scientific evidence compendium: an open, continually evolving collection of independent empirical research, formal theory, mathematical proofs, cross-disciplinary analyses, applications, critiques, and proposed falsifications. Every entry is included because it supports, refines, challenges, or attempts to falsify the theory.
Scientific theories are strengthened not only by evidence that confirms their predictions, but also by surviving attempts to falsify them. This compendium brings both together: independent evidence from researchers who were not testing DSRP and proposed counterexamples evaluated against the formal theory.
One counterexample is enough to falsify O-Theory. Until then, the question remains: do the same four structures organize everything from quantum systems to human thought?
Independent convergence is one of the strongest forms of scientific evidence because researchers arrive at the same conclusion while investigating different questions for different reasons.
The number is not the point. Researchers in different fields, studying different questions with different methods, repeatedly arrived at the same structural predictions—almost always without testing DSRP or using its language.
Try a different term or clear the filters.
This compendium began as the peer-reviewed literature review “A Literature Review of the Universal and Atomic Elements of Complex Cognition,” published in the Journal of Systems Thinking with 109 studies. That paper is the peer-reviewed foundation. What you see here is its living, continuously updated version . New studies are checked before they are added, and the collection now holds and keeps growing. Open any card to see what the researchers found, why it bears on DSRP, and where the original review discusses it, the fuller account.
The collection is updated continuously, so the citation carries the date you consulted it rather than a study count — the count changes weekly, and putting it in the reference would make the same collection look like a different work to everyone who cites it. To cite a single claim or study, use its own address: every one has a permanent link.
This is the adversarial half of the compendium. Where the evidence track asks what converges on DSRP, this one asks what would end it: a single organized phenomenon whose structure needs a fifth pattern, a ninth element, or a fifth structural dynamic. It holds written up from candidates across territories, and resolutions — the general answers those cases settle against. Every case is published whether it held or failed, including the ones still open.
Cite this rather than the evidence collection when the point is what survived attack. The two are separate works with separate addresses: one asks what converges on the theory, the other asks what would end it, and a reference to the first does not support a claim about the second. Case and resolution numbers change when the record is revised, so cite a case by its own permanent link rather than by number.
This collection is meant to keep growing. Send us a study, paper, book or critique that bears on DSRP, whether it supports the theory or cuts against it, and we will read it and decide whether it belongs.
Authors Cabrera, Derek
Year 2026
Publisher Journal of Systems Thinking
Kind of work article
Discipline Computer Science & AI
Read it at the publisher 10.54120/jost.v26i1.dc6umdd6
Unified Modeling Language (UML) is a standardized modeling language whose vocabulary spans structure, behavior, interaction, deployment, and extensibility. This paper asks whether UML introduces any organizational structure not already expressible by DSRP-484. Following the structural-remainder protocol used in companion analyses of system dynamics and network science, we separate information (I) from organization (O), and we replace the earlier practice of nominating constructs by judgment with a pre-registered extraction rule applied to the OMG UML 2.5.1 normative Abstract Syntax Metamodel. The rule yields 242 metaclasses, of which 100 normalize into their ancestors under a purely structural criterion, leaving a closed candidate set of 142. The inclusion filter excludes none of them, because a normative abstract syntax has already been filtered by the standards body. The entire metamodel, 6,785 elements, is assembled from eighteen element types, of which eight are plainly informational. Ruling the remaining construction kit therefore rules UML: Class is identity and part-whole under two undeclared perspectives; Property performs three jobs on one organization, selected by a flag; Association carries two ordered action and reaction ends; Generalization is Distinction and Relationship simultaneously on one link. Twenty further constructs were selected before ruling on the criterion of whether each could break the claim, including conditionality, the temporal metaclasses, ordering, negation, templates, and the metalevel. A separately counted adversarial block was drawn from the Unified Foundational Ontology, whose authors hold that UML requires more primitives rather than fewer; its four load-bearing challenges are rigidity, existential dependence, four irreducible parthood relations, and relators. No residual organizational primitive X was identified anywhere. A second result concerns what the omissions cost: UML has no way to express negation and builds it eleven times, no way to name a point and hard-codes framing in sixty-nine Boolean flags, no Link construct and builds a surrogate for one, and no loop construct while ninety of its own metaclasses sit inside cycles. The result does not make UML unnecessary. It locates UML as a domain-specific modeling language whose higher-order constructs are generated from a more elemental organizational calculus, and whose recurring workarounds are the measurable cost of the structures it declines to name.
Third run in the structural-containment series after system dynamics and network science. Asks whether UML introduces any organizational structure outside the closure generated by DSRP-484, using a pre-registered extraction rule against the normative machine-readable metamodel rather than constructs nominated by the analyst.
Twenty-eight constructs were ruled individually – nine from which the whole normative metamodel is assembled, plus twenty hard cases selected in advance for their capacity to break the claim, Generalization counted once in both groups – together with a separately counted adversarial block from the Unified Foundational Ontology. No residual organizational primitive X was identified in any of them. Across the closed 142-construct candidate set there are 271 novel structural contributions, and every one is a relationship end (160), a primitive value (96), or an enumeration value (15); there is no fourth kind. Cabrera reports a second, independent result as a census of workarounds: 83 derived properties plus 32 derived unions recomputing severed paths; four metaclasses standing in for the Link the metamodel does not contain; negation hand-built eleven times; framing hard-coded in 69 Boolean flags, 13 enumerations and 674 OCL constraints; and one case where the specification concedes its own formalism cannot express one of its operations. The conclusion is containment, not replacement: UML is located as a domain-specific modeling language composed from the same organizational grammar found in the system dynamics and network runs.
Our reading: this run's contribution over the SD and network papers is methodological. The candidate set is generated by script from a public normative file rather than curated by the analyst, the ruling on each contribution is decided by OMG's own declared types rather than by interpretation, and the falsifier is a count that a critic can re-run. That converts the containment claim from a sample into a census. The workaround census is the second move worth keeping: it makes formalism bias measurable in expenditures rather than arguing it, and Cabrera reads the eleven hand-built negations as a Love-Reality Loop that ran and could not close because the missing structure had no name.
Primary nominator: OMG UML 2.5.1 normative specification cross-checked against UML.xmi. Normalization criterion: a metaclass enters the candidate set exactly when it introduces at least one owned attribute or association end that neither redefines nor subsets an inherited property – 242 metaclasses, 100 normalize, 142 remain, inclusion filter excludes none. Two further nominators: Selic's 1998 account of modeling complex real-time systems (retrospective containment check – every construct he names is already in the candidate set) and the UFO/OntoUML adversarial block. Decision rule carried over from the companion papers: a feature is organizational if changing or removing it changes what entities or relations exist or how they are bounded, nested, or ordered; informational if it changes only content or value. Falsifier stated as a procedure: subtract the informational content, and if what remains requires a new element to hold it, that is X – a fifth mediation that is not exists/negates, belongs/contains, affects/effects or frames/is-framed. Measured compression: UML declares 950 features across 242 metaclasses, of which 836 are typed by another metaclass and 114 by a primitive or enumeration – 88 per cent relational vocabulary, 12 per cent information. Appendix A lists all 142 candidates with their contributions and rulings; Appendix B lists the 100 that normalize.
Patterns it shows D, S, R, P
Shows the structure is there, Shows it makes a difference
How to cite this Cabrera, Derek (2026). Is UML Made of DSRP?. Journal of Systems Thinking.