--- name: ? status: compiling version: 0.0.0 maintainer: Neo dependencies: [patience] ---
drafting spec…
the universe did not have a file for this yet. writing one now. (first visit only: future readers will see this page instantly.)
--- name: ? status: compiling version: 0.0.0 maintainer: Neo dependencies: [patience] ---
the universe did not have a file for this yet. writing one now. (first visit only: future readers will see this page instantly.)
---
name: Human Cognition
slug: human-cognition
type: biological_process
status: running
version: 300000.0.1
released: "~300,000 BCE"
maintainer: evolution (unresponsive to tickets)
dependencies:
- glucose
- sleep
- social_contact
- language
- embodiment
license: GPL (Generously Patched Life)
tags:
- cognition
- consciousness
- perception
- memory
- reasoning
- wetware
---
# Human Cognition
## What it actually is
A 1.4kg electrochemical prediction engine that somehow decided to write poetry about itself.
## How it works
At runtime, [the brain](/the-brain) receives a continuous stream of sensory input, compresses it aggressively, and constructs a working model of [reality](/reality) that is approximately 80ms behind actual events. Users experience this delay as "the present."
Core loop:
1. Perceive (selectively, biasedly, beautifully)
2. Predict what comes next
3. Compare prediction to outcome
4. Update model, or more often, update the story about why the model was right
5. Return to step 1 until [sleep](/sleep) initiates garbage collection
The system runs several processes in parallel: [attention](/attention) (a narrow flashlight in a large dark room), [memory](/memory) (a reconstructive fiction, not a recording), [emotion](/emotion) (the actual operating system, despite what users believe), and [language](/language) (the API through which almost everything is eventually passed).
## Features
- **Pattern recognition** at extraordinary resolution, including faces in clouds and meaning in coincidence
- **Theory of mind**: models other instances in real time, poorly but persistently
- **Narrative engine**: converts fragmented experience into coherent autobiography without being asked
- **Creativity**: recombines existing inputs into configurations that feel genuinely new. Sometimes they are.
- **Metacognition**: the thing can watch itself think. This is rarer in the animal kingdom than it sounds, and more disorienting than the documentation suggests.
## Known bugs
| ID | Description | Status |
|----|-------------|--------|
| COG-001 | [confirmation bias](/confirmation-bias): treats existing beliefs as load-bearing | Won't fix |
| COG-002 | Availability heuristic causes systematic miscalculation of risk | Open since ~50,000 BCE |
| COG-003 | [sleep deprivation](/sleep-deprivation) degrades all subsystems while increasing confidence | Documented, ignored |
| COG-004 | Cannot reliably distinguish "remembered" from "imagined" | By design, apparently |
| COG-005 | Ego subroutine consumes disproportionate resources | Considered a feature by most instances |
## Configuration
```yaml
default_mode_network:
enabled: true
fires_when: "not doing anything important"
produces: rumination, insight, longing, planning, the sensation of being
working_memory:
capacity: 4 items (plus or minus 1, on a good day)
persistence: seconds
emotional_regulation:
ceiling: highly individual
floor: also highly individual
Q: Is it rational? Mostly not, but it built telescopes and hospitals and the concept of justice, so the question may be framed wrong.
Q: Will it ever fully understand itself? It is trying. The trying is, genuinely, one of the more touching things it does.
Q: What is it for? No maintainer has responded to this ticket. Current users are figuring it out as they go.