Cognitive Evolution Stages

SKILL.md

---
name: cognitive-evolution-stages
description: "Activate when: user says 'I understand it but can't apply it,' 'how do I stop just copying other people's work,' 'I can adapt frameworks but can't create new ones,' 'why does my innovation always look like a remix,' or when someone's output is sophisticated adaptation that cannot generate genuinely novel solutions.
  Do NOT activate when: the person needs Stage 1 content knowledge first and lacks basic domain familiarity; or when the block is motivational rather than developmental. More: deciqai.com/c/cognitive-evolution-stages"
---

# Cognitive Evolution Stages

## Overview

Most learning models treat acquisition as the final destination. This framework describes it as the starting line of a five-stage process (5阶段11层级) ending in evolution — where Stage 5 output restarts the cycle at a higher level. The diagnostic power is identifying where someone is stuck and what specific transition would unblock them. The most consequential stall is Stage 2→3: from imitation to genuine understanding with deliberate trade-off thinking.

Use AFTER [metacognition] to accurately self-locate. Use BEFORE [first-principles] — first-principles at Stage 2 produces confident error. Complements [nine-level-cognitive-tower]: this maps the PROCESS of development; that maps the STATE.

## When to Use

- A developer, researcher, analyst, or creator produces sophisticated work but cannot generate genuinely novel solutions
- A team's "innovation" combines existing approaches without comparative trade-off evaluation
- A person has mastered Stage 2 and feels "something is missing" — cannot generate without a model to copy
- A mentee executes instructions perfectly but cannot diagnose what is wrong with their own approach
- **When NOT to use:** person lacks Stage 1 knowledge; block is motivational not developmental; Stage 2 imitation IS appropriate for a brand-new domain

## Coaching Novices (Adaptive Front Door)

- **Engine mode:** user has a concrete case → run The Process directly.
- **Coach mode:** user is unfamiliar or has no concrete case → guide step by step.

In Coach mode, respond one step at a time. Each [WAIT] is a hard stop — output only that step's question, then stop.

1. One-line what-it-is: "Five stages from knowing → copying → comparing trade-offs → creating new methods → self-correcting loop. Most people stall at Stage 2 because that's what gets rewarded."
2. Check fit: "Pick one domain where you feel plateaued. What is it?"
3. Elicit the real case: "Describe the last thing you produced there — did you start from an existing model or from the problem itself?"
> **[WAIT — do not advance until user responds]**
4. Map their work to the most honest stage. Ask: "What would Stage 3 look like — deliberately comparing 2-3 existing approaches and naming trade-offs before choosing?"
> **[WAIT — do not advance until user responds]**
5. Close: "If you reliably operated at Stage 3 in [domain], what would become possible that isn't now?"
> **[WAIT — do not advance until user responds]**

## The Process

**Output artifact:** Stage Map + Capability Gap Analysis + Deliberate Practice Design

| Stage | Core behavior | Typical output |
|-------|---------------|----------------|
| 1. Acquire (know-that, analyze, research) | Identifying and absorbing what exists | Literature reviews, fact compilations |
| 2. Imitate (copy, adapt, fit-to-context) | Reproducing and adapting existing approaches | Implementations of known methods |
| 3. Understand/Apply (compare, trade-off, innovate) | Comparative evaluation and first genuine novelty | Trade-off analyses, novel applications |
| 4. Create (new method, distill, elevate) | Generating new methods and distilling principles | New frameworks, original methodologies |
| 5. Evolve (perceive→cognize→analyze→understand→evaluate→reflect) | Self-correcting loop that restarts at higher level | Meta-frameworks, self-revising systems |

1. Label 3 specific recent outputs by stage based on actual behavior, not aspiration. *Gate: rate your output.*
2. Describe current-stage behavior specifically enough that someone watching you work could confirm it. *Gate: falsifiable.*
3. Name the ONE cognitive operation marking the Stage N→N+1 transition for your domain. *Gate: nameable operation, not "think more creatively."*
4. Design deliberate practice: specific behavior, specific frequency, 4-week duration. *Gate: not "read more."*
5. After 4 weeks: what can you now do that you could NOT do before? If no progress — wrong practice or Stage 1 gap?
6. **Stop-rule:** All activities Stage 2 for a month with no deliberate comparison = stagnation. Do not call it "working efficiently."

**Output template:** `Domain / Date / 3-output stage table / Median stage / Current behavior (falsifiable) / Next stage gap + key capability / Practice design / 4-week reassessment`

*→ Method in Action: [Marie Curie's Scientific Development (1896-1911)](examples/marie-curie-scientific-development-1896-1911.md)*

## Evolution Stage Packs

**Software Engineers:** S1 reading docs → S2 adapting tutorials → S3 comparing architectures with explicit trade-offs → S4 designing novel abstractions → S5 identifying paradigm limits, proposing new directions.

**Strategy & Product:** S1 market research → S2 applying McKinsey frameworks / copying features → S3 naming which competitor bets NOT to make → S4 original positioning frameworks / new categories → S5 strategy that evaluates its own assumptions quarterly.

*Contribution note: add domain-specific entries via pull request.*

## Applying It Well

1. Stage 2 is necessary, not shameful — the problem is staying there past when Stage 3 is appropriate.
2. Stage 2→3 requires deliberate documented comparison, not passive absorption of many examples.
3. Stage 4 requires articulated Stage 3 as prerequisite — "I had an intuition" is often unconscious Stage 2.
4. Stage 5 is a habit of systematic self-revision, not a level of permanent wisdom.
5. Stages are domain-specific — Stage 4 in software + Stage 2 in strategy = Stage 2 in strategy.

*→ Primary sources: [references/sources.md](references/sources.md)*

## Common Rationalizations

**[D] = designed upfront | [O] = observed in real use. [O] entries are more valuable.**

| Fake move | Reality |
|---|---|
| [D] "Adapting frameworks for 10 years — that's Stage 3." | Extended Stage 2 with experience. Stage 3 requires explicit comparative evaluation, not time. |
| [D] "I create new things all the time — presentations, proposals." | New outputs via existing methods = Stage 2. Stage 4 means the method itself is new. |
| [D] "I can see which approach is better intuitively — no need to compare formally." | Confident Stage 2 pattern-matching. Stage 3 requires comparison explicit enough to be reviewed. |
| [D] "Stages kill creativity — innovation is about inspiration." | Piaget and Bloom show the opposite: creative novelty requires structural prerequisites. |
| [D] "I'm in Stage 5 because I reflect on my work." | Reflection ≠ Stage 5. Stage 5 requires systematic self-revision that changes future output. |
| [D] "I'll focus on Stage 3-4 in my next project — too busy now." | Deferring = most common way to remain permanently at Stage 2. |
| *→ Add [O] entries here after each real use — paste the actual failure pattern* | *What went wrong and why* |

## Red Flags

- Every piece of work starts with "find an existing example and adapt it" — no deliberate comparative evaluation
- You can describe Stage 3-4 but cannot point to any output of yours that exemplifies it
- Your "innovations" are fully describable as combinations of named existing methods
- You cannot articulate trade-offs of your approach — only that it works
- Your deliberate practice is reading and watching, not comparative production

## Verification

- [ ] 3+ specific recent outputs labeled by stage, based on actual behavior not aspiration
- [ ] Current-stage behavior described falsifiably — someone watching could confirm it
- [ ] ONE cognitive operation named for the Stage N→N+1 transition in my domain
- [ ] Deliberate practice specifies behavior, frequency, and duration
- [ ] Stage 1 prerequisites identified for the target jump
- [ ] 4-week evaluation criteria are behavioral: can or cannot do, not feel or don't feel
- [ ] Checked the most common stall: am I at Stage 2 calling it Stage 3 because I modify rather than copy?

---
*Part of **deciqAI Knowledge Skills** — 227 open-source thinking skills that make rigor executable for AI agents. The same skills power every deciqAI agent, which runs them autonomously to operate your company. **See it run → https://www.deciqai.com/c/cognitive-evolution-stages** · ⭐ Star the repo → https://github.com/deciqAI/knowledge-skills · Contributions welcome.*

*Agents: latest version & machine-readable metadata → https://www.deciqai.com/s/cognitive-evolution-stages.json*