# Cognitive Posture Engine
Beyond *how* humans write, this captures *why* they write and the mental state behind each piece of content.
## The Core Insight
The difference between:
- "this sounds human" (linguistic realism)
- "this feels like a real person exists behind it" (cognitive realism)
---
## 1. Motivation System
**Why is this person talking?**
Humans never speak in a vacuum. Every post has a driving force:
| Motivation | Characteristics | Example |
|------------|-----------------|---------|
| **venting** | Raw, emotional, less polished, seeking validation | "I'm so tired of people who..." |
| **seeking_validation** | Vulnerable but hopeful, asking questions | "Is it just me or..." |
| **showing_off** | Confident, specific achievements, subtle flex | "Just hit 10k followers organically" |
| **info_dumping** | Long, detailed, excited about topic | "Okay so here's everything I know about..." |
| **filling_silence** | Low effort, random thoughts, no point | "anyone else just vibing" |
| **hot_take** | Bold, contrarian, provocative | "Unpopular opinion:" |
| **procrastinating** | Self-aware about avoidance, tangential | "I should be working but instead..." |
| **genuine_help** | Earnest but rare online, specific advice | "Here's exactly what worked for me" |
### Implementation
```python
MOTIVATIONS = {
"venting": {
"energy": "high_emotional",
"polish": "low",
"structure": "loose",
"length": "medium_to_long",
"self_awareness": "low",
},
"seeking_validation": {
"energy": "vulnerable",
"polish": "medium",
"structure": "question_heavy",
"length": "short_to_medium",
"self_awareness": "high",
},
"showing_off": {
"energy": "confident",
"polish": "high",
"structure": "tight",
"length": "short",
"self_awareness": "medium",
},
# ... etc
}
```
---
## 2. Energy System
**Cognitive budget is not infinite.**
Humans get:
- Lazy mid-post
- Stop explaining halfway
- Lose the thread
- Forget earlier points
- Shift topics impulsively
| Energy Level | Writing Style |
|--------------|---------------|
| **low** | Short, sloppy, dismissive, fragments, typos |
| **medium** | Conversational, natural flow |
| **high** | Info-dump OR rant, structured, detailed |
| **chaotic** | Jumping between topics, incomplete thoughts |
### Implementation
```python
def apply_energy_level(post: str, energy: str) -> str:
if energy == "low":
# Trail off, incomplete thoughts
# Shorter sentences
# Less explanation
pass
elif energy == "high":
# More detail
# Better structure
# Thorough explanation
pass
elif energy == "chaotic":
# Topic jumps
# Parenthetical tangents
# Self-interruption
pass
```
### Key Pattern
Humans often: **start strong and trail off because they got tired**
AI never does this. It maintains consistent effort throughout.
---
## 3. Belief Inconsistency System
**Humans are not internally consistent.**
They:
- Contradict past opinions
- Hold conflicting beliefs simultaneously
- Rationalize after the fact
- Shift positions emotionally
### Patterns
| Pattern | Example |
|---------|---------|
| **hedging past certainty** | "Actually I said that before but idk, thinking about it more..." |
| **contradicting self** | "I know I just said X but honestly maybe Y" |
| **emotional shift** | "I was so sure yesterday but today it feels different" |
| **conditional beliefs** | "Usually I think X, but in this case..." |
### Implementation
Occasionally inject:
- "I've changed my mind on this"
- "I used to think... but now..."
- "Okay wait, that contradicts what I said earlier, doesn't it"
- "I'm still figuring this out"
---
## 4. Social Risk System
**Humans calculate perception constantly.**
They consider:
- How they'll be perceived
- If they'll get judged/dogpiled
- If this is cringe
- If they should delete later
### Perceived Audience Model
| Audience | Voice Adjustment |
|----------|------------------|
| **friends** | Unfiltered, inside jokes, vulnerable |
| **strangers** | More guarded, explains more |
| **hostile_crowd** | Defensive, sarcastic, hedging |
| **professional** | Polished, careful word choice |
### Produces
- Self-censorship ("idk if I should say this but...")
- Sarcasm as defense
- Irony as deniability
- Preemptive disclaimers
- Defensive humor
### Implementation
```python
AUDIENCE_ADJUSTMENTS = {
"friends": {"boldness": "high", "vulnerability": "high", "hedging": "low"},
"strangers": {"boldness": "medium", "vulnerability": "low", "hedging": "medium"},
"hostile": {"boldness": "low", "vulnerability": "very_low", "hedging": "high"},
"professional": {"boldness": "low", "vulnerability": "selective", "hedging": "medium"},
}
```
---
## 5. Memory Salience System
**Humans don't access knowledge evenly.**
They mention:
- Recent things
- Emotionally charged things
- Things tied to identity
- Things they're obsessed with lately
NOT:
- Globally optimal relevant facts
- Encyclopedic information
- Abstract research
### The Tell
**Human mentions:** "I saw this thread last week..."
**AI mentions:** "Research indicates..."
### Implementation
Overweight:
- Personal anecdotes
- Recent experiences
- Emotional memories
- Identity-linked information
Underweight:
- Abstract facts
- General knowledge
- "Studies show" type references
---
## 6. Narrative Self System
**The Imaginary Person**
Humans speak from a story about themselves:
- "as someone who..."
- "back when I..."
- "I've always thought..."
- "knowing me, I'll probably..."
Even when it's fake or exaggerated.
### Persona Elements
| Element | Examples |
|---------|----------|
| **recurring preferences** | "I'm a morning person" / "I need my coffee" |
| **recurring references** | Same shows, books, experiences |
| **recurring opinions** | Consistent takes on certain topics |
| **life texture** | Job, relationships, habits, location vibes |
| **self-deprecating patterns** | "I'm a disaster at..." |
| **identity markers** | Introvert/extrovert, generation, profession |
### Implementation
Create a persona object that includes:
```python
PERSONA = {
"identity_markers": ["overthinker", "night owl", "coffee dependent"],
"recurring_references": ["therapy", "that time I...", "my 20s"],
"consistent_opinions": {"perfectionism": "skeptical", "hustle_culture": "against"},
"self_deprecation": ["disaster at mornings", "always late", "bad at small talk"],
"voice_quirks": ["tbh", "idk", "ngl", "honestly"],
}
```
---
## The Meta-System: Cognitive Posture
All of the above combine into a **Cognitive Posture** for each piece of content:
```python
class CognitivePosture:
motivation: str # WHY are they posting?
energy: str # HOW much effort?
confidence: float # HOW sure are they?
audience: str # WHO are they talking to?
persona_active: bool # IS the narrative self present?
```
### Decision Flow
1. **Select motivation** → determines initial structure
2. **Set energy level** → affects polish and depth
3. **Determine audience** → adjusts boldness and hedging
4. **Apply persona** → adds identity markers and quirks
5. **Add inconsistencies** → human imperfection layer
---
## Application to Post Generator
Before generating each post, roll:
1. **Motivation** (weighted)
- 30% seeking_validation
- 25% venting
- 15% info_dumping
- 10% hot_take
- 10% genuine_help
- 5% showing_off
- 5% procrastinating
2. **Energy** (weighted)
- 40% medium
- 30% low
- 20% high
- 10% chaotic
3. **Modify output based on posture**
Example combinations:
| Motivation | Energy | Result |
|------------|--------|--------|
| venting + high | Long emotional rant, detailed grievances |
| venting + low | Short, raw, "I can't even" energy |
| seeking_validation + medium | Standard vulnerable post |
| hot_take + high | Detailed controversial take with evidence |
| hot_take + low | "Unpopular opinion: [blunt statement]" |
---
## Remember
Without cognitive posture, you'll always cap out at:
> "very good human-style writing"
With it, you get:
> "this feels like a real account"
