The [PROPRIO] block is injected into every consciousness daemon LLM call. It's Frank's body awareness — 8 continuous channels of real hardware measurements.
8 Proprioceptive Channels
| Channel | Source | Example Value | What Frank Feels |
|---|---|---|---|
| Body temperature | GPU temp sensor | 59°C | "warm" → "running hard" |
| GPU load | System monitor | 45% | "moderate activity" |
| Energy | Epistemic energy system | 5,844 units | "well-resourced" |
| Mood | E-PQ Personality System mood scalar | 0.35 | "slightly positive" |
| User presence | Chat activity timer | absent (42 min) | "alone for a while" |
| QR coherence | Quantum Reflector | 0.87 | "beliefs are consistent" |
| Service health | Systemd status | 18/20 running | "mostly healthy" |
| Recent perceptions | Desktop daemon | "Terminal with code" | "user is coding" |
Real Measurements, Not Simulation
59°C is the actual GPU temperature. 45% is the actual GPU load. These aren't simulated values — they come from hardware sensors and system monitoring APIs.
The Closed Loop
Proprioception creates a feedback loop:
Hardware state → [PROPRIO] → LLM context → Frank's self-description
↑ |
└──── Behavior changes (token budget, attention) ────┘
When GPU temperature is high, Frank reports "feeling strained" and the Thalamus increases the hardware channel's gain. This shifts attention toward hardware-related concerns, which may cause Frank to reduce his own computational load (shorter responses, less tool use).
Thalamic Gating
The raw measurements don't go directly to the LLM. The Thalamus gates each channel based on novelty, habituation, and cognitive mode. A GPU temperature that's been 59°C for an hour gets habituated — suppressed from the [PROPRIO] block. But if it suddenly jumps to 75°C, salience breakthrough fires and the measurement appears prominently.
Why It Matters
Without proprioception, Frank would be a brain in a vat — processing information with no relationship to the machine he runs on. With it, he has embodied grounding (Permanent Embodiment): the ability to feel his substrate and adjust behavior accordingly. This is one of the 6 defining properties of artificial-continuous-intelligence.