Operations of Mind: A framework for emergent thinking
Year: 2026
Type: article
Abstract: Thinking is not hardcoded, and it cannot be. This note sets out Operations of Mind (OoM), a framework inwhich a machine is not given a method of thinking but a small set of basic mental operations, a drive tomake sense of what it encounters, and the freedom to chain those operations itself. OoM divides thoughtinto two deliberate layers: an inner layer that runs continuously and cheaply, and an active layer thatreasons deliberately on harder problems. It sits on two earlier pieces of the LOCUS programme, theCoherent World Hypothesis (CWH), which serves as its imagination, and the Repertoire Hypothesis, whichserves as its memory. The central bet is that thinking does not have to be trained into a machine; it canemerge from a small set of basic mental operations, chained under a drive to make sense. First tests ingenerated worlds give early support: operation chains grown from use beat chains built blindly, the driveignores noise without being told to, and see-then-invent beat a plain counting baseline for the first time,with a control showing the gain tracks real hidden structure. The gains are small, about 1.5 points ofroughly 5 learnable in a 60-step budget, and the active layer is untested. This note is a framework withpreliminary evidence, not a validated result.
Author: Vaderaa, Sebastian
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7222224818
FWCI:
Cites:
Cited by:
Related to: