Precision Dissociation in Hierarchical Predictive Coding: A Computational Principle Illustrated in Narcissistic Personality Disorder

Abstract We introduce precision dissociation as a general computational principle in hierarchical predictive coding, and illustrate its consequences in the context of Narcissistic Personality Disorder (NPD). The principle holds that opposing precision assignments across parallel inference streams produce characteristic dissociations in the behaviours those streams govern — without any deficit in the underlying generative model capacity. Applied to NPD, the principle predicts two clinically prominent but mechanistically unexplained features of the disorder: (i) intact cognitive empathy (Theory of Mind, mentalising) co-occurring with substantially impaired affective empathy (emotional resonance, empathic concern); and (ii) episodic, disproportionate self-esteem collapse following ego-threatening events. We formalise the framework as a three-level Bayesian hierarchy in which the affective inference stream carries systematically lower precision (π aff) than the cognitive stream (π cog), while the self-model carries a high-precision, inflated prior (Π (0) self). Under sustained social threat this self-model precision erodes exponentially, shifting the free-energy landscape from a deep-grandiose to a deep-vulnerable attractor basin — the formal correlate of narcissistic collapse. Proof-of-concept simulations across two-group agent ensembles (n = 50 per group, 300-trial social paradigm with embedded narcissistic injury block) are consistent with the predicted dissociation: empathic accuracy is near-equivalent between groups (HC = 0.783 ± 0.010 vs. NPD = 0.771 ± 0.010, absolute difference < 2%), whereas empathic concern is approximately fourfold lower in the NPD regime (HC = 0.346 ± 0.014 vs. NPD = 0.086 ± 0.004). Self-esteem collapse during the injury block is ∼86% larger in the NPD regime than in controls. Phase-space analysis reveals bistability in the (Π self , S) plane, recasting the grandiose–vulnerable oscillation as a precision-driven attractor bifurcation. The framework generates falsifiable predictions for electrophysiology, neuroimaging, and experience-sampling studies, and suggests therapeutic leverage points distinct from those implied by symptom-level accounts.

Authors

Institutions

Publication Details

Journal
Research Square
Published
2026-11-12
DOI
https://doi.org/10.21203/rs.3.rs-11449501/v1
Primary Topic
Embodied and Extended Cognition
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Precision Dissociation in Hierarchical Predictive Coding: A Computational Principle Illustrated in Narcissistic Personality Disorder

Marco Fronzi
Research Square
Embodied and Extended Cognition
preprint

Precision Dissociation in Hierarchical Predictive Coding: A Computational Principle Illustrated in Narcissistic Personality Disorder

Marco Fronzi
preprint en

Abstract

Abstract We introduce precision dissociation as a general computational principle in hierarchical predictive coding, and illustrate its consequences in the context of Narcissistic Personality Disorder (NPD). The principle holds that opposing precision assignments across parallel inference streams produce characteristic dissociations in the behaviours those streams govern — without any deficit in the underlying generative model capacity. Applied to NPD, the principle predicts two clinically prominent but mechanistically unexplained features of the disorder: (i) intact cognitive empathy (Theory of Mind, mentalising) co-occurring with substantially impaired affective empathy (emotional resonance, empathic concern); and (ii) episodic, disproportionate self-esteem collapse following ego-threatening events. We formalise the framework as a three-level Bayesian hierarchy in which the affective inference stream carries systematically lower precision (π aff) than the cognitive stream (π cog), while the self-model carries a high-precision, inflated prior (Π (0) self). Under sustained social threat this self-model precision erodes exponentially, shifting the free-energy landscape from a deep-grandiose to a deep-vulnerable attractor basin — the formal correlate of narcissistic collapse. Proof-of-concept simulations across two-group agent ensembles (n = 50 per group, 300-trial social paradigm with embedded narcissistic injury block) are consistent with the predicted dissociation: empathic accuracy is near-equivalent between groups (HC = 0.783 ± 0.010 vs. NPD = 0.771 ± 0.010, absolute difference < 2%), whereas empathic concern is approximately fourfold lower in the NPD regime (HC = 0.346 ± 0.014 vs. NPD = 0.086 ± 0.004). Self-esteem collapse during the injury block is ∼86% larger in the NPD regime than in controls. Phase-space analysis reveals bistability in the (Π self , S) plane, recasting the grandiose–vulnerable oscillation as a precision-driven attractor bifurcation. The framework generates falsifiable predictions for electrophysiology, neuroimaging, and experience-sampling studies, and suggests therapeutic leverage points distinct from those implied by symptom-level accounts.

Research Square
The University of Sydney (AU)
Good health and well-being
Embodied and Extended Cognition
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon