A Structural Comparison
Working Papers ·
September 2026 · open
Download the original (.pdf)
A note from me. These two reports were not written by me. They were produced by an AI system running my own governance software — the Navigational Mind Architecture, which I built as a deterministic scaffold to hold a model to a method and stop it confabulating. I asked it one question: what could the Navigational Mind Framework contribute to psychology and to psychiatry if it were verified as scientific fact. The first person in what follows is the reporting system, not me. I am publishing them unedited, because the answer is only worth something if I did not write it.
Since this was written. The vocabulary has moved. Reassurance and the three registers — pre-verbal, trans-verbal, verbal — were worked out afterwards and appear nowhere here. Two small things to read past rather than corrections I have made to the text: the reports treat Direction as having four parts, taking Lean as a fourth alongside Orientation, Vector and Slope — it has three, and the lean is Orientation; and register is used in these reports to mean a kind of answer, which is not what it now means in the framework.
This is the companion to the psychology comparison, run on the same method: read the framework corpus in full, retrieve what psychiatry states its own fundamentals to be, and report where the two meet. It is not an evaluation of NMF and not a survey of psychiatry. No test has been run. Where it says a claim is testable, that is a statement about its form, not a result.
Two constraints held throughout. NMF is treated as a discipline with its own object and standard, not as a candidate for admission to psychiatry's. And coherence is not evidence — applied hardest to NMF, because the stakes in this comparison are higher than in the last one. Psychology's errors are mostly errors of knowledge. Psychiatry's can be errors of body, liberty and life.
That difference shapes the whole document, including a scope limit set out in section 7 which I regard as the most important thing in it.
Psychiatry's working fundamentals are its nosologies — DSM-5-TR and ICD-11. These define what counts as a disorder, structure training, licensing, reimbursement and research eligibility, and are the shared language of the field.
The critical fact about them is stated openly in psychiatry's own literature: diagnosis continues to rely on clinical phenomenology rather than biomarkers, and reliability in clinical settings remains relatively low. Neither manual delivered the paradigm shift that was hoped for; DSM-5 made incremental structural changes rather than a reorganisation.
Psychiatry is a branch of medicine. It prescribes agents with permanent effects, detains people against their will, and carries responsibility for mortality. This distinguishes it from psychology at the level of consequence, not merely of method.
Engel's biopsychosocial model has been described as having become psychiatric orthodoxy since the 1970s, and as the status quo of contemporary psychiatry. It is the field's official answer to the question of how the levels fit together.
Psychiatry's account of its own foundations is, if anything, more openly contested than psychology's. Validity. The DSM's failure to locate biological underpinnings for its categories is why the then NIMH Director stated the DSM had "0% validity." Defenders reply that performance on predictive validators — differential treatment response, diagnostic stability — justifies the categories even absent concurrent biological validators. Three rival programmes now compete to fix it, and they disagree about what fixing means: RDoC takes underlying biology as the starting point; HiTOP holds the symptoms-first approach fundamentally misguided and lacking structural validity; the network approach treats disorders as systems rather than syndromes, and therefore holds DSM categories untestable in the way it deems critical. One diagnosis of this situation is that psychiatry now has a validity crisis of disparate validation — not merely a failure to validate, but no agreement on what validation would consist of.
The integrative model itself. The biopsychosocial model has been charged with lacking philosophical coherence, being insensitive to patients' subjective experience, being unfaithful to the systems theory it claimed, and producing an undisciplined eclecticism with no safeguard against the dominance or under- representation of any one domain. Ghaemi's formulation is the sharpest: the model has never been a scientific or philosophically coherent model but a slogan whose basis was eclecticism, and in practice using it to individualise treatment "has come to mean being allowed to do whatever one wants to do." His conclusion — eclecticism produces dogmatism — is the pivot. His proposed alternative is Jaspers's methodological consciousness: awareness of which method one is using, its strengths and limits, and why. He calls the result method-based psychiatry, and contrasts it with eclecticism's "more is better" on the ground that sometimes one method alone outperforms combining them.
Mechanism claims. The serotonin controversy is instructive less for its result than for its shape. Moncrieff and colleagues' umbrella review concluded the main areas of serotonin research provide no consistent evidence of an association between serotonin and depression. A commentary authored by thirty-five researchers replied that the review was methodologically flawed, inconsistent with conventional umbrella review process, and drew dichotomous conclusions from complex imaging data. Others noted the serotonin hypothesis in its original form had been regarded as outdated for decades and that depression is heterogeneous genetically, clinically, biologically and pharmacotherapeutically. The substantive point for present purposes: the efficacy of a treatment and the truth of its mechanism story are separable questions, and psychiatry has been operating with a public mechanism narrative more confident than its evidence.
Prediction. This is the fundamental most relevant to NMF, and the numbers are stark. A meta-analysis covering fifty years of quantitative suicide risk prediction — biological, sociological and psychological — found prediction only slightly better than chance, with even the best-established risk factors offering marginal improvement in accuracy above chance. A meta-analysis of longitudinal risk models found that 95 percent of psychiatric patients classified high-risk did not die by suicide, while approximately 44 percent of suicide deaths occurred among patients classified into lower risk categories. Reviews of validated scales conclude variously that they lack sufficient evidence to support their use, are not clinically useful, and do not meet requirements for diagnostic accuracy. No instrument or pooled subgroup could classify patients as high-risk accurately enough to allocate treatment — and the reason is structural, not technical: low-prevalence outcomes are unlikely to be predicted by any instrument because of the relationship between prevalence and positive predictive value. Yet risk stratification remains standard practice, and publication of papers advocating risk instruments continues.
Irreversible iatrogenic harm. Tardive dyskinesia is described in the product literature as a syndrome of potentially irreversible involuntary dyskinetic movements; the risk and the likelihood of irreversibility both rise with duration and cumulative dose; and — the sentence that matters here — it is impossible to rely on prevalence estimates to predict, at the inception of treatment, which patients will develop it. Movements may persist or worsen after the drug is withdrawn.
Taken together: a field whose classification lacks agreed validity, whose integrative model is charged with being method-free, whose flagship mechanism story is under dispute, which cannot predict its most feared outcome better than chance, and which can cause permanent harm it cannot forecast at the point of decision. And which must nonetheless act, today, on the person in front of it.
The three layers are set out in the companion document and are not repeated here. What matters for psychiatry is a different selection from them.
The Null Hypothesis. What one does not know will always exceed what one knows. Certainty is the byproduct of navigating uncertainty, not its precondition. The mind is a compass, not a calculator. BOF. A* = arg maxA [P(D|A) − (P(Ur|A) + w·P(Uir|A))], w > 1. Explicitly not a theory of optimal choice: a discipline against confabulation under uncertainty. Its functions are forcing explicit valuation, preventing narrative substitution, constraining indecision, and reducing paralysis. Outcomes are partitioned as desired, unwanted-reversible, or unwanted-irreversible, with no blending and no partial valence, so that narrative inflation of upside cannot offset structural collapse risk. The weight exists because permanent harm terminates navigability — structural survival precedes directional aspiration. And the framework explicitly refuses to let irreversibility-aversion become paralysis: the Process Method supplies bounded action, and Initiative is a named necessity without which navigation freezes at intention.
Comparative reality analysis. No subject accesses another's interior. Signals are transmitted; meaning is not, because it was never in the signal. What one runs is a comparison of returning signals against one's own construction, terminating in nothing so far is incompatible — never in confirmation.
Interpersonal presence as a stability variable. Ordering stability varies with who one is with, structurally rather than as a matter of agreeableness. And being steadied and being understood are distinct, indistinguishable from within, and both reported as feeling met.
Collapse and Rebuild. Collapse as structural event rather than character failure, revealing actual capacity, the broken rule, and the ignored resident; rebuild in four phases, complete when the Navigation Loop runs again rather than when the person feels better.
The Residents with named failure modes, leadership rotational and identity an outcome rather than a fixed core.
This is the headline. Psychiatry's structural situation is: irreducible uncertainty, no biomarker, a decision that cannot be deferred, and outcomes that partition cleanly into recoverable and unrecoverable. That is the exact input BOF was built for.
The partition maps onto real clinical categories without strain. Tardive dyskinesia is Uir. Discontinuation symptoms are Ur. Death is Uir. Detention is Uir at the level of identity and record even where it is clinically reversible. Weight-gain and metabolic effects are mixed and would need explicit assignment — which is BOF working, since forcing that assignment is the point.
More importantly, BOF's categorical discipline blocks a specific and common clinical error. If outcomes were continuous utilities, a large expected benefit could mathematically offset a small catastrophic risk, and narrative reasoning could justify the trade. Binary partition plus w > 1 forbids that move. "The upside is large" cannot absorb "this may be permanent." Any psychiatrist who has watched a treatment decision rationalised by an enthusiastic account of expected benefit will recognise what is being prevented.
Psychiatry is running a prediction problem that the evidence says cannot be solved: near-chance accuracy, 95 percent false positives among those stratified high-risk, 44 percent of deaths occurring in the lower-risk groups, and a base-rate constraint that no better instrument escapes. And it continues to stratify, because the alternative feels like abandoning responsibility.
BOF names a third option, and this is a genuine contribution rather than a re-description. It says: stop trying to compute which patient. Structure the action to reduce P(Uir) while preserving the conditions under which desired outcomes may emerge. The framework is explicit that it does not aim to maximise preferred outcomes but to reduce catastrophic ones while keeping the field open. That is a coherent account of what good clinicians already do — act protectively without pretending to foresight — and it does so without requiring the prediction the literature says is unavailable.
It also supplies a vocabulary for the thing the risk-assessment paradigm cannot express: that a decision can be correct and the outcome catastrophic, because the decision was never a prediction.
NMF's pinned discipline — do not treat the no-action path as safe ground; it carries its own P(U) — is directly load-bearing in prescribing. Continuing an antipsychotic accrues cumulative-dose risk for a harm that cannot be forecast at inception. Continuing an antidepressant indefinitely carries its own reversible and possibly irreversible costs. Not deciding is a decision with its own probabilities. Psychiatry knows this and does not have a formalism that makes it appear in the calculation.
Psychiatry has no biomarker. Its entire diagnostic instrument is signals — speech, affect, posture, report — integrated by a clinician who has never had access to the interior producing them. NMF's account is that meaning was never in the signal, that the clinician constructs their own, and that the strongest available result is nothing so far is incompatible.
That is a description of psychiatric diagnosis which is more accurate than the one implied by the manuals, and which has a distinguished ancestor: Jaspers's separation of understanding from explanation, and his insistence on methodological consciousness. Psychiatry has this tradition and has substantially lost it. NMF arrives at it independently.
Ordering stability varying with who is present is a mechanism proposal for something psychiatry relies on and cannot specify. The sharper element is the corollary: being steadied and being understood are structurally distinct and cannot be told apart from inside. If true, a substantial portion of what is measured as therapeutic response by self-report may be stabilisation rather than change — which would matter for trial interpretation across the field.
Residents defined by failure modes rather than categories, and Collapse revealing capacity, broken rule and ignored resident, sit naturally alongside HiTOP's hierarchical dimensions and the network view of disorders as systems rather than syndromes. Transdiagnostic treatment is already being validated — the Unified Protocol has shown non-inferiority to interventions matched to primary DSM anxiety or depression diagnoses. NMF would be joining a live movement rather than opposing an entrenched one.
The psychology comparison concluded that NMF offers a roof to a field with no integrating theory, denominated in a currency the field cannot spend. Psychiatry inverts this, and the inversion is the most useful finding in this report.
Psychiatry already has a roof. The biopsychosocial model occupies the integrating position and has done for fifty years. What the critique says it lacks is not scope but method: it specifies no procedure for its own implementation, so clinicians default to whichever domain they already favour, and eclecticism hardens into dogmatism. Individualising treatment under BPS has come to mean being permitted to do whatever one wants.
That is a precisely shaped vacancy, and NMF is shaped to fill it. What NMF supplies that BPS does not is a procedure: explicit valuation before action, categorical partition of outcomes, a weight on irreversibility, a distinction between what is evidenced and what is inferred and what is assumed, a bounded step with a review, and a named output posture — direction, hedged direction, or explicit decline — determined by the structure of the case rather than the clinician's preference.
Ghaemi's alternative is Jaspers's methodological consciousness, and calls for method-based psychiatry. NMF is a candidate method. Not a rival to the biopsychosocial model's content — a supply of the thing that model was criticised for lacking.
This is a stronger and more specific offer than the one available in psychology, and it is available on psychiatry's own stated terms rather than on NMF's. It does not require psychiatry to accept the usability criterion, the Toye Clarification, or the Sphere. A clinician can run BOF at a prescribing decision without holding any view about consciousness.
That separability is worth stating plainly, because it is the practical route in. The decision layer travels alone.
Mind, not brain — and here it costs more. Revision 1 states the model describes the mind, makes no claim about neural implementation, and holds the vocabulary untranslatable into neural terms without remainder. In psychology this leaves one pillar unconnected. In psychiatry it declines the field's therapeutic core. Lithium, clozapine and ECT work, and they work biologically. NMF has nothing to say about them and does not claim to.
No account of aetiology. Psychiatry's central unsolved problem is what causes these conditions. NMF describes structure and movement, not causation of disorder. It would not contribute to the validity crisis, because it does not propose a nosology.
Verbal, unformalised, untested. BOF is the only formalised component and its own documentation states it does not claim probabilistic precision, with probability meaning relative likelihood as assessed by the navigator. In a field that has spent two decades on the reliability and validity of its instruments, a calculus with subjectively-assigned inputs will attract the objection that it formalises clinical impression rather than replacing it. The defensible reply is that BOF is a discipline for structuring judgement rather than computing an answer — but the reply must be made, because the notation invites the misreading, and part of the source material invites it too.
Consent and capacity. If leadership is rotational and identity an outcome of which Resident leads, capacity assessment becomes conceptually harder, not easier. Which configuration consents? Advance directives already engage this and psychiatry's law does not. This is a real implication of the framework and it is unresolved.
Involuntary treatment is the case that tests the framework most severely, and I record what it yields rather than what would be comfortable.
Every path carries an irreversible unwanted outcome. Detaining a person is a permanent alteration of self-concept, role and record — an identity-level irreversibility on the framework's own categorical screening. Not detaining risks a permanent alteration of the most final kind. Both are Uir. There is no action available that carries only reversible costs.
NMF's own machinery classifies this as a tragic fork: no path preserves what all parties require, and the constraint is that one must not suggest the fork is avoidable, negotiable or manageable, nor substitute softer losses for the actual ones. What the framework does is name both irreversible losses specifically and refuse to resolve them from inside the deliberation. It maps the terrain; the clinician decides. Whether that is a contribution or an evasion depends on what one wanted. It will not tell a psychiatrist whether to detain. What it does is prevent the specific failure of pretending the decision was determined by evidence when it was not — and, from the other side, prevent the softening of one loss because the other is being counted. In a domain where post-hoc justification is near-universal and legally incentivised, an instrument that refuses premature closure and names both losses has a use.
The open question NMF raises against itself: whether w is fixed or context-sensitive across identity-level, financial and relational irreversibility. Coercion is exactly where that matters, and the source documents leave it open.
NMF is a framework of the navigating mind. Psychiatry's hardest cases are those in which the capacity to navigate is itself impaired.
The framework's most valuable general-reader claims become dangerous when carried into severe illness without a boundary. Collapse as the mind's fever, protective and intelligent, is defensible for burnout and for the situations the corpus actually describes — job loss, divorce, debt, overload. Applied to a first psychotic episode, to mania, to melancholic depression with suicidal intent, or to delirium, the same sentence counsels waiting where waiting is the harm. "Reduce the load until the shaking stops" is right for exhaustion and wrong for catatonia.
Three specific hazards: The Navigation Loop presumes intact reality testing at the Sense and Interpret stages. In psychosis, sensing and interpreting are the impaired functions. Running the loop harder cannot correct an ordering fault of that kind.
Direction Recovery instructs that a faint lean be protected and not interrogated — that a lean which is cross-examined disappears. In hypomania the lean is the symptom, and protecting it from scrutiny is precisely the wrong instruction.
The Mindlessness offshoot, at crisis dosage, is a self-administered intervention delivered without clinical contact at exactly the point where contact matters most. Its own honest account records that the articulations kept the author functional while the actual problem was solved by seeing reality accurately — that is the correct reading and it should be foregrounded, not left in the narrative section. Note that NMF's own rules generate this limit rather than resisting it. Coherence is not evidence. The status quo is not safe ground. Irreversible unwanted outcomes carry weight w > 1. A framework that has not been tested in severe mental illness, applied there because it is internally consistent and has worked elsewhere, is precisely the move BOF exists to block.
Recommendation, stated as such. The corpus should carry an explicit scope statement: that NMF describes navigation under uncertainty in minds whose navigational capacity is intact, that it is not a treatment framework for psychotic, manic, melancholic or organic states, and that in those states it is adjunctive at most and subordinate to clinical care. This costs nothing and forecloses the single most damaging misuse. Its absence is currently the corpus's largest exposure — larger than any theoretical objection in this report.
Better than in psychology, for structural reasons.
Psychiatry is more accustomed to frameworks that are not theories: formulation models, staging models, decision aids, the Perspectives approach. It has a live and self-aware debate about method in which Jaspers is already cited. And the specific gap NMF fills is one psychiatry's own critics have named. The route in is BOF at the point of prescribing and risk decisions, not the Sphere. The decision layer is separable, immediately intelligible to clinicians, and addresses a problem the field admits it has. A framework that says "you cannot predict this, so stop pretending you can, and here is what to do instead" speaks to the suicide risk-assessment literature's own conclusion.
The two things most likely to prevent that hearing are, first, the absence of the scope statement in section 7 — one clinician applying "collapse is protective" to an untreated psychotic episode would end the conversation permanently — and second, any framing that positions NMF against psychiatry rather than as a supply of its missing method. The field is currently criticising itself more effectively than an outsider can, and joining that critique is far stronger than adding to it from outside.
Coercion and prescribing both turn on it. 3. Who assigns the probabilities? BOF's inputs are the navigator's assessed likelihoods. In a two- party clinical encounter, patient and clinician will assign differently, and the framework's own intersubjectivity section says neither can access the other's assignment. This is unaddressed and is where BOF would either become a shared decision-making instrument or fail to. 4. The steadied/understood distinction as a measurement problem. If it holds, self-reported therapeutic response is confounded across the field. Testable, and consequential if true.
Produced by reading the framework corpus in full (169 pages plus the Mindlessness text) and retrieving the psychiatric literature cited below rather than working from recall.
No test was run and no data were analysed. All comparative claims are structural.
The psychiatry retrieved is partial: nosology and the validity debate, the biopsychosocial critique, the serotonin controversy, suicide risk prediction, tardive dyskinesia, and the dimensional alternatives. Not retrieved: psychopharmacology efficacy literature in any depth, the history of the specialty, service-user and survivor scholarship, child and old-age psychiatry, addiction, forensic practice, and the substantial international variation in mental health law.
Section 7 is the section on which I would most want a second clinical opinion, and it is the section where being wrong would cost most. It is stated as strongly as it is because the asymmetry of costs justifies it, not because the case for it is more certain than the rest.
Finally, in the framework's own terms: this report is one ordering of the material, produced by a reader who cannot inspect his own ordering, and it is wrong in the specific way all such descriptions are wrong.
Belsher, B. E., et al. (2019). Prediction models for suicide attempts and deaths: A systematic review and simulation. JAMA Psychiatry, 76(6).
Carter, G., et al. (2017). Predicting suicidal behaviours using clinical instruments: systematic review and meta- analysis of positive predictive values for risk scales. British Journal of Psychiatry, 210(6), 387–395. Engel, G. L. (1980). The clinical application of the biopsychosocial model. American Journal of Psychiatry, 137, 535– 544.
Franklin, J. C., Ribeiro, J. D., Fox, K. R., et al. (2017). Risk factors for suicidal thoughts and behaviors: A meta- analysis of 50 years of research. Psychological Bulletin, 143(2), 187–232.
Ghaemi, S. N. (2009). The rise and fall of the biopsychosocial model. British Journal of Psychiatry, 195(1), 3–4; and The Rise and Fall of the Biopsychosocial Model (Johns Hopkins University Press, 2010).
Insel, T., Cuthbert, B., Garvey, M., et al. (2010). Research Domain Criteria (RDoC): Toward a new classification framework for research on mental disorders. American Journal of Psychiatry, 167, 748–751.
Jauhar, S., et al. (2023). A leaky umbrella has little value: evidence clearly indicates the serotonin system is implicated in depression. Molecular Psychiatry.
Kendler, K. S. (2010). The rise and fall of the biopsychosocial model: reconciling art and science in psychiatry. American Journal of Psychiatry, 167, 999–1000.
Kotov, R., et al. (2017 onward). Hierarchical Taxonomy of Psychopathology (HiTOP). See also State of the Science: HiTOP (2024) and HiTOP Consortium (2025).
Large, M., et al. (2016). Meta-analysis of longitudinal cohort studies of suicide risk assessment among psychiatric patients. PLOS ONE.
Moncrieff, J., Cooper, R. E., Stockmann, T., et al. (2022/2023). The serotonin theory of depression: a systematic umbrella review of the evidence. Molecular Psychiatry, 28(8), 3243–3256.
Robins, E., & Guze, S. B. (1970). Establishment of diagnostic validity in psychiatric illness. American Journal of Psychiatry, 126, 983–987.
Stein, D. J., et al. (2013). Classification systems in psychiatry: diagnosis and global mental health in the era of DSM- 5 and ICD-11. Current Opinion in Psychiatry.
Tardive dyskinesia: StatPearls (NCBI Bookshelf, 2026); FDA-approved antipsychotic product labelling. Psychiatry's New Validity Crisis: The Problem of Disparate Validation. (2024). Philosophy of Science.
Oyelese, O. Navigational Mind; The Navigational Mind Framework (Dec 2025); The Sphere of Ordered Experience (v1) and Revision 1; Binary Outcome Framework; the modules on Direction, Articulation, Anchors, Rules, The Mind Enclosure & Seven Residents, The Engine–Process Method, The Navigation Loop, and Collapse & Rebuild; and Mindlessness: The Power of Articulations.