# Neural Decoding **Entity class:** Technology, method, or technical system **Domain:** Computational Neuroscience / Machine Learning **Doc Type:** Concept Node **Maturity:** Active Research and Deployment **Related:** [[Neural Signal Acquisition]], [[Generative BCI Decoding]], [[Brain-Computer Interfaces]], [[Neural Data Provenance]] --- ## Definition **Neural decoding** estimates an intended action, stimulus, linguistic unit, perceptual category, or internal state from measured neural activity. A decoder is a model conditioned by its sensor modality, task, training data, temporal window, preprocessing, and target representation. ## Evidentiary boundary Decoded output is an inference, not a transparent copy of thought. Accuracy depends on the experimental distribution and may not generalize across people, devices, sessions, or contexts without calibration. ## Neurotech cluster route **Collection:** [[collections/Neurotech|Neurotech]] **Source articles:** [[articles/The Architecture of Continuity and Emerging Neuroinformatics Standards|The Architecture of Continuity and Emerging Neuroinformatics Standards]] ## Relationships [[wiki/Neural Decoding|Neural Decoding]] is related to [[collections/Neurotech|Neurotech]] through [[articles/The Architecture of Continuity and Emerging Neuroinformatics Standards|The Architecture of Continuity and Emerging Neuroinformatics Standards]]. <!-- BEGIN HUMANIZED RELATIONSHIPS 2026-09-11 --> This entry's documented connections are expressed in its definition and related-work routes, with provenance retained in the source-linked material. <!-- END HUMANIZED RELATIONSHIPS 2026-09-11 --> ## Related Work in the Corpus <!-- BEGIN HUMANIZED CORPUS ROUTES 2026-09-11 --> - In [[articles/The Architecture of Continuity and Emerging Neuroinformatics Standards|The Architecture of Continuity and Emerging Neuroinformatics Standards]], ****3\. Standardization of Brain-Computer Interfaces (BCI)**** provides the narrative context for **Neural Decoding**: Decoding and control: neural decoding · generative BCI decoding · cross-modal neural decoding · closed-loop BCI · human-in-the-loop BCI. <!-- END HUMANIZED CORPUS ROUTES 2026-09-11 --> ## Research Inference Attractors <!-- BEGIN DEEP INFERENCE ATTRACTORS 2026-09-11 --> These are secondary semantic placements for the inference attractor network. Each statement keeps its original ID and tier; its canonical cluster page links back to every destination. Source register: [[research/Research Inferences|Research Inferences]]. Interpretive context: [[articles/Technologies for Consciousness Mapping and Transfer|Technologies for Consciousness Mapping and Transfer]] and [[articles/Mind Uploading and AI — The Host is Reusable and the Person is the Delta|Mind Uploading and AI — The Host is Reusable and the Person is the Delta]]. Collection route: [[collections/Neurotech|Neurotech]]. - **INF-0007 — Strongly indicated.** Academic speech BCIs reached 62–78 words per minute with implanted electrodes and model-side decoding, which means the recent gains are attributable to the decoder more than the sensor. Decoder improvement is free of surgical risk and compounds across every existing implant already in a head. - **Canonical cluster:** [[wiki/Neural Interfaces and Continuity Architecture|Neural Interfaces and Continuity Architecture]] - **INF-0016 — Plausible.** Prosthetic control without sensory return plateaus quickly because the loop is open. Writing proprioceptive feedback into somatosensory cortex closes it, and the closed loop is where the boundary between the device and the body's own model of itself starts to dissolve. - **Canonical cluster:** [[wiki/Neural Interfaces and Continuity Architecture|Neural Interfaces and Continuity Architecture]] - **INF-0023 — Plausible.** Below roughly a hundred milliseconds, a decoded action stops feeling like operating a tool and starts feeling like moving. Sub-perceptual latency is the threshold at which the interface is incorporated into the body schema, and body-schema incorporation is the psychological precondition for substrate transfer. - **Canonical cluster:** [[wiki/Neural Interfaces and Continuity Architecture|Neural Interfaces and Continuity Architecture]] - **INF-0082 — Established.** Continuous language has been reconstructed from non-invasive recordings by aligning brain activity to a language model's semantic space. The alignment, not the scanner, does the work, which is why decoding quality now improves with each generation of language model. - **Canonical cluster:** [[wiki/Cognitive Interface Layer|Cognitive Interface Layer]] - **INF-0083 — Strongly indicated.** Work through 2026 has moved from decoding externally supplied captions toward recovering participants' own inner speech via subject-specific neural-semantic alignment. Inner speech is the first genuinely private content to become instrumentable, and the technique requires no change to the underlying language model. - **Canonical cluster:** [[wiki/Cognitive Interface Layer|Cognitive Interface Layer]] - **INF-0085 — Analytic.** Decoders currently require per-subject alignment, which is the main obstacle to deployment. The moment cross-subject transfer works from a short calibration, decoding becomes a service rather than a study, and the cost per decoded mind falls by orders of magnitude. - **Canonical cluster:** [[wiki/Cognitive Interface Layer|Cognitive Interface Layer]] - **INF-0094 — Plausible.** Foundation models trained on aggregated neural recordings will exhibit the same scaling behavior as other modalities, meaning decoding quality becomes a dataset-size problem. The entity holding the largest standardized neural corpus will hold decoding capability that cannot be replicated by better hardware alone. - **Canonical cluster:** [[wiki/Cognitive Interface Layer|Cognitive Interface Layer]] - **INF-0095 — Plausible.** A pretrained neural foundation model with few-shot subject adaptation would collapse setup from weeks to minutes. That single change moves neural interfacing from a procedure with a technician to a product with an onboarding flow. - **Canonical cluster:** [[wiki/Cognitive Interface Layer|Cognitive Interface Layer]] <!-- END DEEP INFERENCE ATTRACTORS 2026-09-11 --> ## Simple Reminders, Quotations, and Thoughts > "Brain-machine interfaces continue to be improved, initially for physically impaired people, but eventually to provide a seamless boundary between people and the monitoring network." > **— Marti Hearst**, *2015, Edge annual question “What Do You Think About Machines That Think?”* [[reminders/Neural Interfaces/Brain Interfaces May Dissolve Into the Monitoring Network by Marti Hearst|Brain Interfaces May Dissolve Into the Monitoring Network by Marti Hearst]] > "I envisage the human-computer interface as like having a helpful partner, and the more intelligent machines become the more helpful they can be partners." > **— Lawrence M. Krauss**, *2015, Edge annual question “What Do You Think About Machines That Think?”* [[reminders/Neural Interfaces/Human-Computer Interfaces Could Become Helpful Partners by Lawrence M. Krauss|Human-Computer Interfaces Could Become Helpful Partners by Lawrence M. Krauss]] > "Few doubt that machines will surpass more and more of our distinctively human capabilities—or enhance them via cyborg technology." > **— Martin Rees**, *2015, Edge annual question “What Do You Think About Machines That Think?”* [[reminders/Neural Interfaces/Machines Will Surpass More Human Capabilities by Martin Rees|Machines Will Surpass More Human Capabilities by Martin Rees]]