Journal of Systems Thinking · Volume 6, Issue 1 · 2026-07-03
Hidden in plain science
Testing a universal ontology in the results of 13 scientific disciplines
Derek Cabrera
DOI 10.54120/jost.v26i1.dc6uosd6 (registration pending)
Abstract
ABSTRACT: DSRP theory makes a strong claim: Distinction (identity–other), System (part–whole), Relationship (action–reaction), and Perspective (point–view) are not merely useful ways of thinking about the world but universal structures of organization itself. If that is true, these structures should appear wherever humans encounter reality and report what they find—including in scientific writing produced by people who have never heard of the theory. We tested that claim at this interface. Twenty-seven paragraphs were drawn from published research papers across 13 disciplines by a mechanical rule fixed in advance—the third prose paragraph of each paper's Results or Findings section—and coded by three independently built large language models working blind: one Pattern per pass, one pass per chat, no titles, no discipline labels, no mention of the theory, and a verbatim quotation required for every recorded structure. The predicted structures were recovered in every passage: all 27 paragraphs contained at least one structure found independently by two or more coders (97 consensus structures, 38 unanimous). The predicted epistemological bias appeared with them: Distinction ranked first in 16 of 27 paragraphs (p=1.6×10−4) and Perspective last in 23 of 27 (p=8.3×10−11), reproducing the Cognitive Pareto Bias in scientific Results sections. The predicted DSRP-484 dynamics also appeared: observed pole–complement pairings exceeded every one of 20,000 random re-pairings across all four Patterns, the same entity occupied multiple structural roles in 16 distinct cases, and every paragraph containing a Relationship also contained a Distinction. Disciplinary thought styles received preliminary support: hard sciences marked relatively more Relationship and soft sciences relatively more Perspective, producing the predicted double dissociation (p=0.017); the direction persisted across sensitivity analyses, although its magnitude remains imprecise at this corpus size. Across these distinct tests, every directional prediction moved in the direction specified by DSRP-484, producing convergent evidence from a single mechanically sampled corpus for structures, biases, and dynamics predicted by the same theoretical architecture. No single corpus can establish a universal ontology. This study adds a methodologically distinct line of evidence to a larger convergent research program by finding both sides of the DSRP claim where new empirical knowledge is reported: the predicted structure in what scientists describe and the predicted epistemological bias in what they make explicit and leave implicit. The corpus, codebook, raw coder outputs, and analysis scripts are available upon request.
© 2026 Journal of Systems Thinking (JoST) · CC BY-NC-ND 4.0
