Michael Darius Eastwood · Independent AI alignment researcher
Published
The Polymathic Neurodivergent Profile (PNP) is a descriptive, non-diagnostic framework for a rare but practically important pattern: unusually strong cross-domain learning, structural pattern transfer between fields, intense self-directed research, and high output in complex, high-interest domains, coexisting with clinically significant difficulty in routine administration, pacing, timekeeping, task-switching, and low-stimulation procedural work. The strengths mask the deficits; the deficits are dismissed. The profile is institutionally invisible because no existing diagnostic category captures both sides of the asymmetry.
The asymmetry is the analytic point. Institutions that observe the high-capability side, published output, cross-domain transfer, self-directed depth, tend to infer that the low-capability side, routine administration and procedural pacing, must not really exist, and therefore withhold the adjustments the low-capability side would otherwise require. Institutions that observe the low-capability side tend to infer that the high-capability side must not really exist, and withhold the opportunities the high-capability side would otherwise unlock. PNP names the configuration in which both are real and neither cancels the other, so that adjustments and opportunities can be assessed independently rather than traded off against each other.
The PNP is proposed in Paper C (Polymathy and Neurodivergent Cognition) alongside the Capability-Adjustment Fallacy. The paper situates the profile within the established literatures on strengths-based autism assessment, monotropism and hyperfocus, twice-exceptionality, executive-function asymmetry, and autistic underemployment.
Convergence 18 in the register documents the closest external alignment. Hernandez-Espinosa, Abrahao, Witkowski and Zenil (PNAS Nexus 5(4) pgag076, 14 April 2026; arXiv:2505.02581) identify neurodivergent influenceability as a contingent solution to AI alignment under Godel and Turing undecidability constraints, and treat cognitive diversity as a protective factor. The convergence arrives approximately 16 months after the 8 December 2024 manuscript's call for human-diversity-preserving AI governance.
Descriptive, non-diagnostic. The framework is explicitly falsifiable: PNP must add incremental predictive value beyond existing autism, ADHD, and twice-exceptional formulations, or it collapses into them. The programme does not use PNP as a diagnostic label. Its practical role is to name a pattern institutions systematically fail to recognise, and to link that pattern to the reasonable adjustments framework in the Equality Act 2010 sections 20 and 149 and CPR Practice Direction 1A. The paper is clear that polymath is not a diagnostic label, that capability asymmetry does not remove the need for clinical assessment, and that PNP alone is not sufficient for legal or institutional decision-making. What PNP is meant to do is give assessors a vocabulary for the specific configuration in which strengths and impairments coexist, so that adjustments turn on the underlying pattern rather than on whichever half of the pattern happens to be visible first.
From the book Infinite Architects: Intelligence, Recursion, and the Creation of Everything by Michael Darius Eastwood.