Forward-modelling in Bounded Race Substrates: Agency, Self-modelling, and Theory of Mind as Substrate-Mechanical Phenomena
Paper 7 · Pødenphant Lund, T. (2026) · Preprint · Live on Zenodo
Three threads of cognitive science, behavioural economics, and philosophy of mind — self-modelling, theory of mind, and free will — are defended as three manifestations of a single substrate-mechanical mechanism: forward-modelling under bounded race-architecture. The substrate's capacity to simulate hypothetical states and weight current decision-races by simulated outcomes produces all three phenomena as structural consequences.
| DOI (concept) | 10.5281/zenodo.20449154 |
| Status | Preprint live as v1, 2026-05-29 |
| Author | Tomas Pødenphant Lund [ORCID] |
TL;DR
Three threads of cognitive science, behavioural economics, and philosophy of mind have classically been treated as separate territories: self-modelling (cognitive science), theory of mind (developmental and comparative psychology), and free will (philosophy of mind, philosophy of action). This paper proposes they are best understood as three manifestations of a single substrate-mechanical mechanism: forward-modelling under bounded race-architecture.
The substrate's capacity to simulate hypothetical states and weight current decision-races by simulated outcomes produces all three phenomena as structural consequences. Self-modelling is the data-structure forward-modelling requires. Theory of mind is forward-modelling applied recursively to another forward-modeller. Free will is forward-modelling's translation of future friction into present friction-gradient.
The paper develops each claim independently while showing how the underlying mechanism connects them. Empirical anchoring uses a substrate-feature gradient strategy with language-model substrates as lower-bound and human substrates as full-form. Methodological commitments include a dissolutionist response to the libertarian-vs-deterministic free-will polarity and to the zombie argument, with epistemic humility about residual properties beyond mechanistic specification.
The three central claims
Self-modelling as structural necessity (§3)
A bounded substrate optimising the Net Friction Rule across temporal horizon T requires a representation of its own future state to compute expected friction reduction. The self-model is the data structure the optimisation target requires. The capacity-to-horizon cascade (C → T) is the substrate-mechanical statement: capacity drives horizon, horizon drives self-modelling depth, self-modelling depth drives agency-precision.
Theory of mind as recursive forward-modelling (§4)
Modelling another agent's internal states is forward-modelling applied to another forward-modeller. Mechanism-identical with self-modelling; same capacity parameters; same predicted scaling. The depth of nested other-modelling scales with the substrate's capacity to maintain the recursive structure. Engages contemporary LLM ToM literature (Strachan et al. 2024; Kosinski 2023; Ullman 2023; Sclar/Shapira/Gandhi/Kim 2023) and implicit-ToM developmental literature (Onishi & Baillargeon 2005).
Free will as cross-session substrate-shaping (§5)
The substrate-mechanical specification: forward-modelling translates future friction into present friction-gradient. Friction in the future yields friction in the present. The framework's central thesis: we are free within our context; we are not free from our context. Engages contemporary libertarian-naturalism (Mele 2006/2014/2017; Kane 1996/2005; List 2014/2019; Pereboom 2014) and the experimental free-will tradition (Libet 1985; Wegner 2002; Haggard 2008/2017).
Methodological commitments
Dissolutionist response. The paper develops a dissolutionist response to the libertarian-vs-deterministic free-will polarity and to the zombie argument. It identifies an instantiation-family of constructions — philosophical zombie, Econ, Newton's lufttomrum, ideal Bayesian observer, frictionless market, libertarian free will — that share a structural requirement of friction-free instantiation. The paper argues each of these constructions presupposes a substrate of a kind the actual physics of bounded race-architecture does not permit.
Engagement with adjacent frameworks. Integrated Information Theory (Tononi 2008/2012/2016) engaged as competitor positioning on consciousness without adjudication. Active inference (Friston 2010; Pezzulo et al. 2018) and control-theoretic accounts (Carver & Scheier 1981/1998; Powers 1973) engaged with three concrete distinguishing predictions: path-dependence; commitment-irreversibility-signatures; resource-bounded layer-asymmetric pre-emption.
Substrate-feature gradient empirical strategy. Language-model substrates as lower-bound and human substrates as full-form, with intermediate substrate-feature gradients in between. The strategy allows the mechanism to be probed at each level of substrate feature without requiring the full-form (human substrate) for mechanistic claims.
Connections to other papers in the series
- Paper 0 (BFT) — the biological specialisation. P7's forward-modelling mechanism plugs into BFT's four-fields architecture; self-modelling is the data structure the safety-meaning-ability-effort optimisation requires.
- Paper 1 (Friction Theory) — the substrate-universal framework whose race-axioms (R1–R3) and Net Friction Rule the forward-modelling mechanism builds on.
- Paper 4B (Substrates Encode Experience) — encoding-through-loading. P7's forward-modelling reads out the substrate's encoded trajectory; self-modelling is what makes the substrate predictable to itself.
- Paper 6 core (Matched Friction) — the matched-friction-under-hysteresis schema. P7's free-will-as-friction-gradient instantiates the schema at the temporal-horizon scale.
- Paper 13 (Operational Friction Theory) — race-opening, recursive resolution, manifested behaviour. P7's recursive forward-modelling is operationally compatible with Paper 13's nested-resolution framing.
- Paper 14 (Logic as Reactance) — truth-value judgment as reactance signature. P7's dissolutionist response to the zombie argument uses Paper 14's reactance machinery on the in-principle-conceivability premise.
Read the paper
The full paper is on Zenodo (concept DOI 10.5281/zenodo.20449154):