Journal of Systems Thinking · Volume 6, Issue 1 · 2026-08-21
Contextual Identity and Cognitive Dark Matter in DSRP-484
The Advanced Is/Is Not Operation and the Structure That Minds and Formalisms Leave Unwritten
Derek Cabrera
DOI 10.54120/jost.v26i1.dc6cicm6 (registration pending)
Abstract
ABSTRACT: What is a thing? We propose that its identity in a model is the organized account of what it is and is not, what it contains and belongs to, what it relates to, and how it is framed: its DSRP structures — Distinctions, Systems, Relationships, and Perspectives — constitute its contextual identity. Cognitive dark matter is the Organization a system uses that is absent from its own account of itself: CDM = O − I(O). We tested these constructs in two large language model families using an invented identity, vorg, holding its definition constant while varying the surrounding Organization and asking questions the definition left unresolved. The supplied Organization, not the definition, determined the answers: GPT matched it at .84–.86 and Claude at 1.00; identities with the same structure counts but different structures produced disjoint answers; removing a Pattern removed the answers it supported; changing one identity altered inferences about neighboring identities; and neither system collapsed unknown into false across 1,440 unidentified cases. Taken together, the findings show that identity behaves as an organized field of what is, is not, and remains unknown, and that this field can extend beyond what is explicitly supplied. With only a definition, GPT built a road-vehicle identity for vorg from familiar terms; when the same world was rewritten with nonce terms, that imported structure disappeared. Cognitive dark matter was therefore measurable as the gap between operative and reported Organization: GPT used .86 of the identity while reporting .25, whereas Claude showed little gap until the instruction “use only the supplied material” was removed, after which it too reported mainly the definition while continuing to use the wider identity. Perspective thus changed both what Organization entered the task and whether the Organization doing the work remained visible. The same structural machinery supported disciplined inference when visible and unsupported assumption when it was not. In real-world judgment, this means that what people or machines conclude can depend less on the information in front of them than on the unreported Organization through which that information is being interpreted.
© 2026 Journal of Systems Thinking (JoST) · CC BY-NC-ND 4.0
