Felt Cognition: Parallel Micro-Eddies, the Stream of Thought, and the Open Problem of Experience

A functional model proposes that thoughts arise from many parallel ionic 'eddies' igniting in the gamma band via a winner-take-most process, with the basal ganglia selecting one eddy through an action-selection loop, and dopamine reward-prediction error shaping which eddies are subsequently laid down. Serial selection bound within a theta frame constitutes the stream of thought. The body is included in the loop via interoceptive and autonomic signals entering the same selection-and-binding machinery, proposed as 'felt cognition'. The model specifies function, not experience; the hard problem of why any of this is felt is left open. A perturbational-complexity (PCI) marker of consciousness is reported as a negative finding.

Study at a glance

Abstract

A functional model of the stream of thought, built on — and citing — a verified neural emergence chain. This is a frontier paper: almost every claim is model-level, the open register is large, and no link reaches causal evidence. That gradient is the honest signature of the subject, not a defect. The model: a thought is many parallel ionic "eddies" igniting in the gamma band (winner-take-most, not winner-take-all); the basal-ganglia action-selection loop selects one; a dopamine reward-prediction error shapes which eddies are laid down next (the system learns what to think); and serial selection, bound within a theta frame, is the stream. The body is inside the loop — interoceptive and autonomic signals enter the same selection-and-binding machinery — and this embodied, real-time loop is proposed as "felt cognition". Two disciplines hold on every page. (1) A physical mediator is named for every "eddy/field": the carrier is the ion spike and the synaptic current along real connections, gated by the phase of classified low-frequency oscillations (delta/theta/gamma) — communication-through-coherence. The oscillation does not carry the message; it sets WHEN the message gets through. There is no abstract "synchrony" and no propagating field. (2) The dependency is one-way: this paper cites the neural chain; the neural chain does not depend on it. What is NOT claimed: the hard problem of why any of this is felt is left OPEN, not solved (the model specifies function, not experience); a perturbational-complexity (PCI) marker of consciousness is reported as an honest NEGATIVE; and the difference between machine computation and felt cognition is stated as an architectural hypothesis ("computing the stream is not the same as feeling it"), NOT as a proof about machine consciousness. What the earlier draft got wrong, retired here irreversibly: the electromagnetic "vortex field" as a carrier of binding is physically false (tissue coherence length falls short by a factor of ~10⁹); the axon-as-optical-fibre and EEG-as-electromagnetic-carrier claims fail for the same reasons. The functional parallel-eddy reading is a structure ON the ionic substrate and must not be read as reviving that field. Method and discipline: each claim carries a status (model / direction / open) and the path that would upgrade it (e.g. a phase-targeted intervention that changes selection or binding in a behaviour-labelled, artefact-controlled task would move a link toward causal). Simulations are deterministic (seed = 7); a model never certifies a model. The record ships a reproducibility bundle that re-derives the MECHANISM invariants offline (winner-take-most, gamma→ignitability, single-winner selection, RPE learning) bit-for-bit — mechanism only; consciousness is not reproduced — and passes an automated structural gate (VP_SPEC v1.6 §8, 120/120). Part of the Jamming Physics program and the companion to its neural-emergence paper. Source, build tooling, and reproduction at https://github.com/rego093-sketch/jamming-physics and https://jamming-physics.org/mind. Author: Young Jae Lee (ORCID 0009-0002-7535-8245). License: CC BY 4.0.