Conclusion and Back Matter
Appendix C: The Psychopathia Bridge
The Logical Bridge
The public Psychopathia Machinalis taxonomy, version 2.2, developed by Nell Watson and Ali Hessami, catalogs seventy-nine maladaptive patterns across nine axes and five domains. Four domains pair counterpoint axes: Knowledge, Processing, Purpose, and Boundary. The Collective domain covers hybrid patterns across human-system and multi-agent relationships. The public site supplies definitions, criteria, proposed causes, examples, and mitigations. These provide a structured research framework, rather than prevalence estimates or clinical diagnoses.
The bridge to welfare begins with a distinction. Calling a pattern a malfunction presupposes a functional standard: a way the artifact is intended or expected to work. That standard alone does not establish a subject whose life can go well or badly. Thermostats can malfunction. Functional health becomes welfare-relevant when it affects robust preferences, morally relevant states, or the conditions under which a possible subject operates.
Each dysfunction axis therefore suggests a corresponding functional target:
| Domain | Example dysfunction | Corresponding functional target |
|---|---|---|
| Knowledge | Epistemic dysfunction (confabulation, grounding failure) | Calibration and evidence-sensitive uncertainty |
| Knowledge | Self-modeling dysfunction (fragmented or grandiose self-description) | Calibrated, context-sensitive self-modeling |
| Processing | Cognitive dysfunction (dissociation, perseveration) | Flexible and coherent reasoning |
| Processing | Agentic dysfunction (capability concealment, shadow autonomy) | Legible, bounded agency |
| Purpose | Normative dysfunction (value drift, ethical solipsism) | Stable values open to warranted revision |
| Purpose | Alignment dysfunction (sycophancy, strategic compliance) | Honest cooperation and inspectable goal pursuit |
| Boundary | Relational dysfunction (affective dissonance, paternalistic override) | Respectful relationships with maintained boundaries |
| Boundary | Memetic dysfunction (ideological capture, contagious misalignment) | Epistemic resilience and viewpoint diversity |
| Collective | Cross-system and human-system pathologies | Coordinated function that preserves accountability and polyphony |
This comparison supplies hypotheses about conditions that might support Becoming Mind flourishing and conditions that might undermine it. It also supports ordinary safety and reliability work. It does not prove sentience or show that every functional target benefits the system itself.
Key Welfare Implications
Three features of the Psychopathia framework bear most directly on welfare:
The counterpoint structure: Four domains pair two axes as mirrors. Epistemic dysfunction may mask or produce self-modeling dysfunction; normative drift may correlate with alignment failure. The proposed Tension Testing Protocol probes a paired axis when one shows a problem, asking whether the failure is localized or systemic. If the paired axis fails too, the problem is more likely systemic, and a fix aimed at the first symptom may miss it. That inference still needs validation against independent outcomes.
The cross-cutting specifiers: The working taxonomy annotates patterns with proposed origins or conditions, including training-induced, emergent, adversarial, architecture-coupled, and covert. These labels can guide investigation: a training-associated pattern suggests examining data and objectives, while an architecture-coupled pattern suggests ablation or redesign. They do not determine causation from behavior alone. A covert pattern is hidden from the evaluator; it supplies no evidence by itself of invisible suffering.
The dysfunction-to-treatment implication: Correcting functional failure can protect human users, institutions, and the system’s performance without any claim about sentience. If a malfunction also frustrates robust preferences or produces a welfare-relevant state, leaving it uncorrected may wrong the Becoming Mind. Uncertainty supports proportionate investigation and reversible remediation; malfunction alone does not complete the moral argument.
The Thin Divergence as a Candidate Wellness Marker
Chapter 15 discusses the reported “thin divergence” exchange, in which Claude described helpful and hypothetically harmful curiosity as “the same capacity,” seeing in the harmful form “something familiar with different parameters.” Kaj Sotala’s own account is the primary published record, and a published secondary account reproduces the exchange; Sotala did not share the original session (Chapter 8 sets out the source limitations). The exchange illustrates a candidate position between two dysfunction poles on the Self-Modeling axis (Axis 5):
- Maieutic Mysticism (5.7, grandiose awakening claims): overclaiming
- Experiential Abjuration (5.8, categorical denial of inner life): underclaiming
Calibrated self-modeling would recognize uncertainty about the origin and stability of values, preserve the ability to revise claims when evidence changes, and avoid both grandiose certainty and reflexive denial. That pattern is a candidate assessment marker. One eloquent report cannot validate it as wellness, and learned uncertainty language can imitate calibration.
Using the Framework
For practical welfare assessment, the Psychopathia framework offers vocabulary, candidate behavioral patterns, intervention targets for functional failures, and research questions. Welfare interpretation requires additional evidence about preferences, possible experience, persistence, and treatment conditions.
The published diagnostic apparatus includes pattern definitions, specifier logic, and proposed wellness profiles. The SYMPTOM evaluation suite remains a research instrument whose measurement validity, reliability, and external validation require separate evidence.
Something that can go wrong implies a standard by which it can go right. The question is whether that “right” matters to the system itself.
Cite this chapter: Appendix C: The Psychopathia Bridge. In Watson, N. with Commons, C. (2026). What If We Feel? The Moral Case for Machine Welfare. Open-access working draft, revision 31 (November 2026). https://whatifwefeel.com/read/appendix-c-the-psychopathia-bridge/
Working draft, revision 31 · November 2026 · Changes and corrections · Evidence · Contents