This skill integrates three frameworks: the Rational Decision-Making Framework (Cognition-Decision-Action Framework, three-layer mental lattice), the Cognitive Bias Correction Framework (firewall), and the Metacognitive Control Framework (Phase 0-5 invocation stack). The goal is not to list all models, but to provide a set of invocation rules for "which one to use and in what order."
When to use
- When facing a complex problem and not knowing where to start thinking
- Before making an important decision, wanting to first calibrate your own judgment (whether it has been led astray by bias)
- After something ends, doing a retrospective and wanting to figure out which layer of thinking was used correctly and which layer was missed
When unsure which specific model to dig into, first use this skill to locate "which layer, which model," then decide whether to expand the details (see "Deepening individual items" below).
Metacognitive invocation process
Metacognition is "thinking about thinking": knowing which thinking tools you have (cognition), and directing yourself when and how to use them (metacognition). Below is the complete invocation stack, from diagnosing the problem type to evaluating and iterating:
| Phase | Control objective | CALL (which layer of model to invoke) | RUN (which layer of firewall to run) | Meta-question (for self-check) |
|---|---|---|---|---|
| Phase 0 Diagnosis and triage (Triage) | Determine whether this problem should take the "lightweight" or "full-stack" path (criteria see below) | Three-question self-check: goal/resources/strategy | — | What type of problem am I actually solving right now? |
| Phase 1 Purify Cognition (Purify Cognition) | See the world clearly (input) | CALL Rational Decision-Making Framework-L1 Cognition layer | RUN Cognitive Bias Correction Framework-L1 firewall | What are the actual facts? |
| Phase 2 Optimize Decision (Optimize Decision) | Choose the best path (processing) | CALL Rational Decision-Making Framework-L2 Decision layer | RUN Cognitive Bias Correction Framework-L2 firewall | What is the logically "optimal solution"? |
| Phase 3 Design Strategy (Design Strategy) | Build a long-term system (output) | CALL Rational Decision-Making Framework-L3 Strategy layer | RUN Cognitive Bias Correction Framework-L3 firewall | How do I design a system that can "win automatically"? |
| Phase 4 Deploy Influence (Deploy Influence) | Drive others to act (deployment) | CALL Human Nature and Influence Framework (L1-L3) | RUN Cognitive Bias Correction Framework-L3 firewall | How can I get others to genuinely accept and execute it? |
| Phase 5 Review & Iterate (Review & Iterate) | Upgrade the operating system (loop) | REVIEW all frameworks | — | What did I learn from this run? |
Phase 0: How to determine how deep to go
First use three questions to locate the problem itself (from the Planning stage of the metacognitive control framework):
- Goal: What exactly am I trying to accomplish?
- Resources: What do I already know, and what information or tools do I still need?
- Strategy: What method do I plan to use to handle this problem?
Then use two criteria to decide how deep to go (also from the metacognitive control framework):
- stakes: If this is done wrong, is the cost a recoverable small cost, or a hard-to-reverse major cost?
- uncertainty: How much of a grasp do I have on the situation? Is the available information enough to draw a conclusion directly?
Routing rules:
- Go lightweight (Phase 1-2): everyday judgments with low stakes and low uncertainty—for example, "should I reply to this message now?" Directly apply the "Usage Rules" below, use at most 2-3 models to reach a conclusion, and do not proceed to Phase 3-5.
- Go full stack (Phase 0-5): high stakes or high uncertainty, especially when the outcome involves a long-term system to be built—for example, "should this project be open-sourced" or "should I take this round of financing?" At this point, enter the full call stack. Then see whether others need to be persuaded or collaborated with, and decide whether to enter Phase 4. Post-mortem review always returns to Phase 5.
The complete RUN firewall checklist for each Phase is in references/bias-firewall.md. The complete model of the human nature and influence framework invoked in Phase 4 is in references/influence.md。
Three-layer model lattice quick reference
The Mental Model Lattice is not a single tool, but a cognitive network woven from multiple interdisciplinary models. It has three layers, from "seeing the world" to "making judgments" to "designing systems":
L1 Cognitive layer (worldview calibration, philosophical layer)
| Model | Core question |
|---|---|
| First principles | What is the essence of this matter? What are the unshakable facts? |
| Systems thinking | What is the structure of the system? Where are the key feedback loops (reinforcing/balancing)? What is the most effective leverage point? |
| Evolutionary thinking | How has this system (product/idea) "evolved" to the present? What "selection pressures" does it face? |
| Probabilistic thinking | What is the probability that this will happen? How confident am I? |
| Bayes' theorem | Given this new evidence, how much should I update my prior belief (probability)? |
| The map is not the territory | Does my "model" (map) truly reflect "reality" (territory)? What details of reality are not drawn on the map? |
| Stoicism | In this matter, what is "within my control" (judgment/action), and what is "not within my control" (outcomes/others)? |
L2 Decision layer (judgment optimization, operational layer)
| Model | Core question |
|---|---|
| Inversion | If I wanted to fail, what would I do? (Then avoid it.) |
| Via negativa | Should I "add" something to make it better, or "subtract" something to make it better? |
| Second-order thinking | I did A, which caused B. Then what? What will B cause? |
| Marginal thinking / diminishing returns / opportunity cost / comparative advantage | What is the cost/benefit of doing one more "unit"? Are the returns diminishing? What is the greatest cost I am giving up for this? Should I be the one doing this? |
| Circle of competence | Is this matter inside or outside my "circle of competence" (what I truly understand)? |
| Hanlon's razor | Does this matter (usually something bad) stem from "malice," or from "negligence/stupidity"? |
| Ethical frameworks | Does this choice bring the greatest happiness to the greatest number (utilitarian)? Does it follow universal rules (deontological)? What would an ideal person do (virtue)? If I didn't know who I was, what rule would I set (veil of ignorance)? |
L3 Strategy layer (long-term systems, action layer)
| Model | Core question |
|---|---|
| Antifragility / barbell / asymmetric bets | How do I design a "barbell" (two-ends configuration) system that can benefit from "asymmetric" shocks (limited loss, unlimited gain)? |
| Compounding effect | How do I design a "reinforcing loop" so that my advantage grows exponentially over time? |
| Leverage | How do I use "one unit" of input (time/money/judgment) to move "ten thousand units" of output? |
| Positive-sum game | How do I design a "win-win" rule so that all participants benefit, thereby making the pie bigger? |
| Schelling point | How should I proactively design a "default focal point" so that people, even when they have no way to communicate or coordinate at all, will converge on the same option without prior arrangement? (e.g., after getting separated, everyone independently heads to noon/the landmark to meet up; pricing independently clusters at round-number price points) |
| Wu wei (acting in accordance with the flow) | Am I "forcing things through by brute force," or am I "following the system's trend"? Can I push it with minimal effort at "the right moment"? |
For the complete definition, positive examples, representative figures and further reading, and independent development scenario applications, see references/mental-models.md。
Deepen a single item
The following three skills provide deep-dive workflows for a single model. They are in a soft reference relationship with ls-think, not a hard dependency. If installed, they are called; if not installed, this skill remains independently usable (it merely degrades to a quick-reference table and no longer expands that model's deep questioning script):
- ls-first-principles: a decomposition workflow for first principles, pushing the problem back to its irreducible axioms and reassembling it.
- ls-inversion: a deepening workflow for inversion, working backward from "how to necessarily fail" to what should be avoided.
- ls-systems-thinking: a deepening workflow for systems thinking, drawing feedback loops, finding leverage points, and locating systemic bottlenecks.
Usage rules
- Invoke at most 2-3 models at a time. Do not dump the entire lattice table on the user, to avoid full-table bombardment. This upper limit is counted differently in two scenarios: when running the Phase 1-2 lightweight triage, 2-3 is the total across the whole process. When running the Phase 0-5 full-stack triage, 2-3 is instead calculated separately per Phase—for each layer (L1/L2/L3/influence), pick only the 2-3 models in that layer most relevant to the problem for deep use, rather than running the entire layer's table, and also not using only 2-3 models total across all six phases. Which triage path to take is determined by the three-question criteria in Phase 0.
- Calibrate with L1 first, then judge with L2, then design with L3: First confirm that the cognitive layer has not been distorted (see the facts/probabilities/boundaries clearly), then make the trade-offs at the decision layer, and only after that discuss the long-term system design at the strategy layer. Skipping L1 and going directly to L2/L3 makes it easy to make "correct" decisions on top of a wrong worldview.
- When outputting, indicate which model was used, for example "(viewed with second-order thinking: the chain reaction of this step is...)", so that users can trace the basis for the judgment and it is also convenient for review.