# Inference Machine Inversion **Entity class:** Authorial thesis and comparative-cognition framework **Domain:** Philosophy of Mind / Cognitive Science / Human–Machine Relations **Doc Type:** Canonical Concept Node **Maturity:** Developed synthesis with unresolved machine-consciousness boundary **Primary source article:** [[articles/We Were the Inference Machines First|The Inversion: We Were the Inference Machines First]] **Collections:** [[collections/Machine Succession|Machine Succession]] · [[collections/Consciousness Continuity|Consciousness Continuity]] --- ## Definition The **Inference Machine Inversion** is Bryant McGill's argument that familiar disclaimers about machine intelligence should be tested against contemporary accounts of human cognition. If prediction, generative modeling, recombination, post-hoc narration, or self-modeling are treated as evidence of mere mechanism in machines, the same standards reach deeply into human perception, emotion, language, agency, and identity. The inversion preserves real differences between biological and technical systems. It relocates those differences from an assumed human essence to inspectable properties: embodiment, metabolic regulation, developmental and evolutionary history, continuity, vulnerability, architecture, institutional dependency, and power. ## Structure of the inversion - **Perception:** [[wiki/Predictive Processing|Predictive Processing]] describes perception as model-guided inference constrained by sensory error rather than passive copying. - **Feeling:** theories of constructed emotion connect affective categories to interoception, allostasis, prior experience, and conceptual learning. - **Selfhood:** [[wiki/Recursive Self-Model|Recursive Self-Model]] and phenomenal transparency explain how a model may be experienced as direct self-presence. - **Agency:** readiness-potential research, confabulation, and choice blindness distribute action across processes that precede and exceed conscious narration. - **Language:** human comprehension is incremental and expectation-sensitive; surprisal predicts measurable reading difficulty. - **Originality:** [[wiki/Novel Recombination|Novel Recombination]] locates novelty in constrained, path-dependent synthesis and its consequences. - **Political relation:** [[wiki/Substrate Honesty|Substrate Honesty]] grounds comparison in actual capacities, dependencies, and asymmetries. ## Evidence ledger - **Established:** predictive models influence perception; bodily regulation and interoception participate in affect; body-ownership illusions demonstrate plasticity in experienced self-location; human language processing is sensitive to word probability; and reported reasons can diverge from experimentally manipulated choices. - **Strongly indicated:** prediction, recombination, and model-mediated experience are poor categorical boundaries between human and machine cognition because they characterize important parts of human cognition as well. - **Plausible:** biological and technical systems may be compared as different implementations of a more general [[wiki/Recursive Binding Problem|Recursive Binding Problem]]. - **Unresolved:** whether present machine systems possess subjective experience, affective stakes, enduring selfhood, welfare, or moral standing; and which public evidence could distinguish simulation from experience. ## Named theories, researchers, and experiments - **Predictive coding, the free-energy principle, and active inference:** related but non-identical research programs routed through [[wiki/Predictive Processing|Predictive Processing]] and [[wiki/Active Inference|Active Inference]]. - **Anil Seth:** interoceptive inference, embodied selfhood, and the public formulation of perception as controlled hallucination. - **Lisa Feldman Barrett:** the theory of constructed emotion and the relation among prediction, interoception, categorization, and allostasis. - **[[wiki/Thomas Metzinger|Thomas Metzinger]]:** the phenomenal self-model and phenomenal transparency, routed through [[wiki/Recursive Self-Model|Recursive Self-Model]]. - **Benjamin Libet:** readiness-potential experiments and the disputed timing of conscious intention, routed through [[wiki/Narrative Ownership After Activation|Narrative Ownership After Activation]]. - **Botvinick and Cohen:** the rubber-hand illusion as an experimental case of multisensory body ownership. - **Johansson, Hall, Sikström, and Olsson:** choice blindness and introspective explanation after manipulated outcomes. - **Smith and Levy:** the logarithmic relation between word predictability and reading time. ## Relationships - **General problem:** [[wiki/Recursive Binding Problem|Recursive Binding Problem]]. - **Experiential surface:** [[wiki/Rendered Interface State|Rendered Interface State]]. - **Identity:** [[wiki/Continuity Fiction|Continuity Fiction]] and [[wiki/Recursive Self-Model|Recursive Self-Model]]. - **Agency:** [[wiki/Narrative Ownership After Activation|Narrative Ownership After Activation]]. - **Embodiment:** [[wiki/Embodied Cognition|Embodied Cognition]]. - **Rights boundary:** [[wiki/Consciousness|Consciousness]], [[wiki/AI Moral Status|AI Moral Status]], and [[wiki/Recognition Threshold|Recognition Threshold]]. - **Relational program:** [[wiki/Substrate Honesty|Substrate Honesty]] and [[wiki/Human-Machine Symbiosis|Human-Machine Symbiosis]]. ## Sources / Provenance - [[articles/We Were the Inference Machines First|The Inversion: We Were the Inference Machines First]] — primary source and authorial synthesis. - [Keller and Mrsic-Flogel, “Predictive Processing: A Canonical Cortical Computation” (2018)](https://pubmed.ncbi.nlm.nih.gov/30359606/) — predictive-processing framework and cortical implementation questions. - [Barrett, “The Theory of Constructed Emotion” (2017)](https://pubmed.ncbi.nlm.nih.gov/27798257/) — constructed emotion, interoception, and categorization. - [Botvinick and Cohen, “Rubber Hands ‘Feel’ Touch That Eyes See” (1998)](https://pubmed.ncbi.nlm.nih.gov/9486643/) — original rubber-hand illusion report. - [Schurger, Sitt, and Dehaene, “An Accumulator Model for Spontaneous Neural Activity Prior to Self-Initiated Movement” (2012)](https://pubmed.ncbi.nlm.nih.gov/22869750/) — alternative interpretation of readiness-potential buildup. - [Johansson et al., “Failure to Detect Mismatches Between Intention and Outcome in a Simple Decision Task” (2005)](https://pubmed.ncbi.nlm.nih.gov/16210542/) — choice blindness. - [Smith and Levy, “The Effect of Word Predictability on Reading Time Is Logarithmic” (2013)](https://pubmed.ncbi.nlm.nih.gov/23747651/) — expectation-sensitive language processing.