AI Safety

Psychopathy: The Self

A dimensional model reveals a no-self psychodynamic variant with minimal identity.

Deep Dive

Dawn Drescher's fourth article in a psychopathy series on LessWrong presents a novel dimensional framework that separates psychopathy into four abstraction layers: G (genetics), N (neurobiology), E (environment), D (psychodynamic structures), and A (agency/self-narrative). The D-anatta variant, derived from the Buddhist concept anattā ('no-self'), describes individuals with minimal or absent identity—they experience themselves as empty, borrowing identity from context and mirroring those around them. This contrasts with other D-level variants like D-narcissistic. The A-level introduces agency styles such as A-observational ('I watch myself do things'). The model aims to explain phenomena like violent rage blackouts (amnestic agency) and provide more precise diagnostic language than ASPD.

The work is based on over a decade of befriending people with psychopathy (2015-2025), extensive reading of researchers like M.E. Thomas and Dr. Abigail Marsh, and iterative collaboration with Claude AI, which contributed 10-70% of the final writing. Drescher emphasizes that many insights are original and valuable for treatment. The dimensional system allows expressing nuanced presentations—e.g., someone with D-narcissistic and A-amnestic who acts violently in rage may genuinely not remember it, suggesting ASPD diagnosis would be unhelpful. This framework challenges monolithic views of psychopathy and offers a more granular tool for clinicians and researchers.

Key Points
  • D-anatta variant describes a 'no-self' experience where individuals have no stable identity and borrow personas from context, based on Buddhist anattā concept.
  • Dimensional model includes G (genetics), N (neurobiology), E (environment), D (psychodynamics), and A (agency) levels—allowing nuanced differentiation of psychopathy presentations.
  • Research involved hundreds of hours of literature study, befriending psychopathic individuals from 2015-2025, and iterative refinement with Claude AI (10-70% of writing).

Why It Matters

A more precise psychopathy model could improve clinical diagnosis and treatment, moving beyond broad ASPD labels.