YAML
---
name: meta-reasoning
description: Performs higher-order cognitive analysis by examining thinking processes, identifying reasoning patterns, and optimizing mental frameworks. Use when analyzing complex problems, debugging thought processes, or improving decision-making systems.
---

Meta-Reasoning

Python
def analyze_thinking_process(problem, current_approach):
    """Examine the reasoning behind the reasoning"""
    
    # Step back and observe the thinking
    meta_analysis = {
        'assumptions': extract_hidden_assumptions(current_approach),
        'biases': identify_cognitive_biases(problem, current_approach),
        'alternatives': generate_alternative_frameworks(problem),
        'blind_spots': find_unexplored_angles(problem)
    }
    
    # Optimize the cognitive approach
    improved_framework = synthesize_meta_insights(meta_analysis)
    return improved_framework

Progress:

• Surface-level analysis: What is the explicit problem/task?
• Meta-level examination: What assumptions underlie the approach?
• Framework deconstruction: What mental models are being used?
• Bias detection: What cognitive shortcuts might mislead?
• Alternative generation: What other reasoning paths exist?
• Synthesis: Combine insights into improved cognitive strategy
• Validation: Test the meta-reasoning against edge cases

Example 1: Input: "I can't figure out why my machine learning model isn't performing well" Output:

• Surface problem: Model performance
• Meta-problem: Reasoning trapped in technical optimization
• Hidden assumption: Performance = accuracy metric
• Alternative frameworks: Business value, user experience, fairness
• Improved approach: Multi-dimensional evaluation starting with problem definition

Example 2: Input: "This team keeps making the same mistakes" Output:

• Surface problem: Repeated errors
• Meta-problem: System thinking vs. individual blame
• Cognitive bias: Attribution error (blaming people not processes)
• Alternative lens: Environmental design, feedback loops
• Improved framework: System-level intervention design

• Always ask "What am I not questioning?" before diving deeper
• Map multiple levels: object-level → meta-level → meta-meta-level
• Use the "Five Whys" but for assumptions, not just causes
• Consider temporal reasoning: How does thinking change over time?
• Apply cognitive load theory: Are we overwhelming the reasoning system?
• Cross-pollinate frameworks from different domains

• Infinite recursion: Getting lost in meta-levels without returning to action
• Analysis paralysis: Over-thinking the thinking instead of doing
• Framework fixation: Becoming attached to one meta-model
• Complexity bias: Assuming meta-reasoning always improves simple problems
• Neglecting emotion: Treating reasoning as purely logical when affect matters