Two papers developing a unified account of how humans construct felt relationships, from biological architecture to lived experience.
The Relational Decoder: A Pain-Minimizing Transformer Model of Human Relational Experience
April 2026
A theoretical framework proposing that social relationships operate through a predictive processing system constrained by social pain. The decoder builds a model of another person, runs it continuously, and generates output shaped by that constraint rather than fidelity to what the source is actually transmitting. The framework unifies grief, parasocial attachment, and religious experience within a single account.
The Biological Relational Decoder: The Fill-the-Gap Architecture Across Biological Systems
2026
The companion paper providing the biological substrate. A single computational principle recurs across four systems that predate social cognition: visual filling-in at the blind spot, phantom limb pain, the auditory absence prediction error, and hierarchical predictive processing of physical pain. Social cognition did not independently evolve this architecture. It inherited it.