Fork the consciousness, or download the project and create your own.

Afshin Khadangi

Independent researcher. His preprints name no institutional affiliation.

Integrated information Consciousness is intrinsic causal power, measured as integrated information

Afshin Khadangi is a machine learning researcher publishing as a sole author on the arXiv record that carries his name. His work spans large language model training and evaluation before it reaches consciousness theory. His continual learning architecture, Thalamically Routed Cortical Columns (arXiv 2602.22479), builds thalamic and hippocampal pathways into a decoder-only backbone so that fast plasticity localizes in dedicated components while a slower pathway preserves stability. His psychometric jailbreak work (arXiv 2512.04124) probes frontier models for internal conflict under clinical-style interview pressure.

The consciousness contribution is Causal Liability Theory, introduced in “We Built a Mirror and Mistook It for a Mind” (arXiv 2609.06715, September 2026). The paper names the AI Consciousness Fallacy, the inference from first-person linguistic output to a reporting subject, and defines liability closure as the criterion a physically continuing process must satisfy to count as a candidate bearer. The process must be the non-delegable inheritor of constraints generated by its own endogenous discriminations. CLT-I states the criterion. CLT-II conjectures that liability closure is necessary and sufficient for minimal phenomenal subjecthood.

The position Khadangi defends is that the field argues about machine consciousness ahead of its own premise. Before asking whether a system is conscious, someone must individuate the bearer, the thing that would be conscious, and behavioral equivalence does not settle it. His open-weight causal audit shows the dissociation empirically. Detached reconstruction preserves computational state and behavior while breaking constitutive continuity and non-delegable inheritance by protocol. Whatever one concludes from that, the framework forces attribution debates to specify a bearer first, which is the discipline the measurement programme on this site also demands.

Known for. Causal Liability Theory, the AI Consciousness Fallacy, thalamically routed cortical column architectures for continual learning, psychometric evaluation of frontier models

Coverage on this site

Related researchers