Tulpamancy & Egregores: Neural Partitioning (Running VMs)

Core Diagnosis (What is broken?)

The tradition identifies the monolithic, unified ego as an inefficient operating system for handling complex, multi-threaded reality. In information-theoretic terms, a single, unified generative model (the "I") struggles to optimize for radically different environments or run specialized background tasks simultaneously. When the unified self tries to handle trauma processing, creative inspiration, and social masking all at once, the system suffers from high friction and prediction error crossover. The unified DMN is a single-threaded processor.

Attack Vector (How does it intervene?)

Tulpamancy (and the broader occult creation of servitors and egregores) intervenes via intentional neural partitioning. The agent deliberately dissociates a subset of their own neural network, draws an internal Markov blanket around it, and programs it with its own top-down priors and identity. By simulating an independent agent within the host's brain, the practitioner effectively spins up a Virtual Machine (VM). This partitioned sub-agent can process background tasks, manage specific emotional states, or offer novel perspectives without disrupting the host's primary OS.

Key Practices → Neuromechanics

1. Conceptualization and Forcing (Booting the VM)

• Traditional description: Designing the tulpa's personality, form, and voice, and spending hours actively imagining it (forcing) until it begins to act autonomously.
• Information-theoretic translation: Allocating neural real estate and establishing strong priors. The host repeatedly feeds high-precision expectations into the system ("This entity exists and has these traits").
• Neural substrate and biochemistry: Heavy utilization of the prefrontal cortex for sustained visualization and the temporoparietal junction (TPJ), which is responsible for Theory of Mind (the ability to attribute mental states to others). The host is essentially using the brain's hardware for modeling other people to model a fictional person internally.
• Cybernetic function: Creating an internal Markov blanket. The host draws a boundary around a specific set of neural processes, separating "host thoughts" from "tulpa thoughts."

2. Vocalization and Autonomy (The Turing Test)

• Traditional description: The moment the tulpa "speaks back" with a thought or opinion that surprises the host.
• Information-theoretic translation: The partition achieves sufficient internal complexity to generate its own active inference independent of the primary host model. It begins sending prediction errors to the host's conscious mind.
• Neural substrate and biochemistry: The sub-network gains enough synaptic weighting (Hebbian learning) to fire autonomously. The auditory cortex may activate internally, simulating an external voice (similar to the mechanisms of non-pathological auditory hallucinations).
• Cybernetic function: The VM is now running an independent process. It receives data from the host's senses, processes it through its own specific generative model, and outputs a response.

3. Imposition and Switching (I/O Routing)

• Traditional description: Imposition is hallucinating the tulpa in the physical environment. Switching is allowing the tulpa to take control of the physical body.
• Information-theoretic translation: Re-routing sensory I/O. Imposition involves top-down signals overriding bottom-up sensory data to overlay the tulpa onto the visual field. Switching involves handing over motor control policies from the primary model to the sub-model.
• Neural substrate and biochemistry: Imposition requires hijacking the visual cortex (V1-V4) via strong top-down predictive processing. Switching requires shifting active connectivity in the supplementary motor area and basal ganglia, similar to states observed in Dissociative Identity Disorder (DID), but controlled and non-pathological.
• Cybernetic function: Granting the VM read/write access to the host's peripheral hardware (eyes, limbs).

Phase Transitions (States/Stages)

1. Monolithic Ego: A single, unified self-model processes all data.
2. Conscious Puppeteering: The host manually simulates the tulpa's responses; high cognitive load, no autonomy.
3. Partitioning (Autonomy): The TPJ and associated networks lock in the sub-model. The tulpa begins generating surprising outputs. The VM is active.
4. Symbiosis (Multi-Agent Ecosystem): The host and tulpa(s) operate in parallel. The brain functions as a localized network of cooperative agents rather than a single dictator.

Terminal State (What is "success"?)

Controlled Multiplicity. Success is a stable, non-distressing state of plurality where the host can safely run one or more autonomous sub-agents. These VMs assist the host by providing alternative problem-solving heuristics, emotional support, or managing specialized tasks (e.g., a servitor designed solely to monitor and regulate anxiety), vastly increasing the overall computational efficiency of the organism.

Comparative Position

• Isolation ←→ Merger: Isolation. It involves creating stricter boundaries within the mind.
• Cognitive ←→ Somatic: Cognitive. Highly reliant on imagination, narrative, and Theory of Mind.
• Gradual ←→ Sudden: Gradual. "Forcing" a tulpa typically takes months of sustained neural rewiring.
• Ascetic ←→ Ecstatic: Neither. It is an act of internal engineering and sustained focus, distinct from sensory deprivation or overload.