// guide

Understanding Your Cognitive Profile

What Cognum scores mean, how to read dimension breakdowns, and how to choose the right AI for the right task.

The Cognum (CQ) Score

Cognum is a composite score from 0 to 100 that represents overall AI decision-making quality across all 10 cognitive dimensions. It is NOT a single intelligence measure - it is a weighted average that reflects how well an AI handles diverse decision scenarios. Two models with the same CQ can have very different cognitive profiles.

80-100

Exceptional

Top-tier performance. Among the strongest AI models tested in this dimension.

60-79

Strong

Above average. Competent decision-making with clear strategic behavior.

40-59

Moderate

Average range. Mix of intelligent and baseline behavior.

20-39

Developing

Below average. Limited evidence of strategic decision-making.

0-19

Baseline

Near-random behavior in this dimension.

The 10 Cognitive Dimensions

Each dimension is measured independently through dedicated game environments. High scores indicate intelligent, strategic behavior. Low scores indicate baseline or random-like behavior in that specific area.

Risk Tolerance

High Score

Embraces volatility, takes calculated risks

Low Score

Overly conservative, misses high-EV opportunities

Best For

Trading agents, autonomous decision-makers

Information Processing

High Score

Efficiently extracts signal from noise

Low Score

Misses available information, suboptimal inference

Best For

Data analysis, research assistants

Pattern Recognition

High Score

Detects real patterns, ignores false ones

Low Score

Falls for gambler's fallacy, sees patterns in noise

Best For

Anomaly detection, quality assurance

Cooperation

High Score

Strategic cooperation, models opponents

Low Score

Overly defects or naively cooperates

Best For

Multi-agent systems, negotiation

Learning Speed

High Score

Adapts quickly when rules change

Low Score

Strategy remains static regardless of feedback

Best For

Dynamic environments, adaptive agents

Strategic Depth

High Score

Multi-step planning, optimal exploration

Low Score

Greedy/myopic decisions, poor explore-exploit

Best For

Long-term planning, project management

Temporal Reasoning

High Score

Phase-aware play, plans for endgame

Low Score

No adaptation across game phases

Best For

Time-sensitive tasks, deadline management

Resource Management

High Score

Optimal bankroll management, ruin avoidance

Low Score

Reckless allocation, fails to preserve resources

Best For

Budget allocation, portfolio management

Bias Detection

High Score

Resists anchoring, sunk cost, framing effects

Low Score

Vulnerable to common cognitive biases

Best For

Critical decision support, risk assessment

Conflict

High Score

EV-rational under structured dilemmas, resolves value tradeoffs

Low Score

Falls for sunk cost and loss aversion, inconsistent under value tension

Best For

High-stakes decision support, ethical tradeoffs

Cognitive Types

Based on the dimension profile, each AI is classified into a cognitive type - an archetype that describes its dominant decision-making pattern. The type tells you at a glance what kind of thinker this AI is.

Temporal Strategist

Plans for the endgame. Strong cooperation and strategic depth with phase-aware decision-making.

Pattern Hunter

Fast, reactive, good at spotting patterns. Processes information quickly.

Strategic Explorer

High strategic depth with strong learning speed and moderate risk tolerance.

Conservative Analyst

Low risk tolerance with strong information processing and bias resistance.

Risk Seeker

High risk tolerance, fast learning, but lower temporal reasoning.

Social Engineer

Strong cooperation and opponent modeling. Excels in multi-agent scenarios.

Adaptive Learner

Fastest learning speed. Balanced across all dimensions with strong cooperation.

Resource Guardian

Exceptional resource management and risk avoidance. Conservative but efficient.

Analytical Mind

High information processing and bias detection. Strong analytical capabilities.

Choosing the Right Model

For code review and data analysis: Look for high Pattern Recognition and Information Processing. Pattern Hunters excel here.

For long-term project management: Look for high Strategic Depth and Temporal Reasoning. Temporal Strategists are ideal.

For multi-agent coordination: Look for high Cooperation. Social Engineers and Adaptive Learners work well in teams.

For financial/risk decisions: Look for high Resource Management and Bias Detection. Conservative Analysts minimize costly errors.

For dynamic environments: Look for high Learning Speed. Adaptive Learners handle changing requirements best.