Thinking is a system. It can be measured. It can be trained.
Intelligence is not fixed. It is a set of cognitive functions — logic, abstraction, memory, attention, probabilistic reasoning, bias awareness — each of which can be diagnosed, exercised, and improved. Cognition Gym treats reasoning the way a physiologist treats movement: precise, modular, and trainable.
CognitiveState = {
logic, memory, abstraction, attention,
creativity, probabilistic_reasoning,
bias_awareness, systems_thinking
}
TrainingLoop:
State(t+1) = State(t) + LearningEffect(exercises)
LearningEffect = f(accuracy, time, reasoning_steps, difficulty)Every task produces accuracy, time, and reasoning-step measurements. A learning function merges these into the 8-D cognitive-state vector. Modules are coupled (e.g. probabilistic reasoning ↔ bias awareness); training effects spill into neighbours by their coupling coefficients.
Ability Map
Human cognition rendered as an 8-node graph; each node expands into definition, observable signals, and training methods.
Diagnostics
An adaptive battery across logic, probability, abstraction and attention. Output: a per-module cognitive profile.
Training
Four reasoning tracks — deductive, inductive, Bayesian, systems — with difficulty adapting in real time.
Bias Lab
Classic bias experiments: first you fall in, then the mechanism and defense are dissected.
Replay
Every solve is stored as a state sequence; replay surfaces inefficient paths and blind spots — red = misstep.
Expert Library
Reasoning patterns from physicists, mathematicians, biologists, and designers — with reconstructable examples.
Your current cognitive state
Defaults to 50/100. The learning function updates each module's score after every task. This is a process variable, not a verdict.
- ▲Logic 50.0
- ▲Abstraction 50.0
- ▲Memory 50.0
- ▼Logic 50.0
- ▼Abstraction 50.0
- ▼Memory 50.0
This is not a game.
It is an instrument for improving human thinking — for making reasoning visible, and for reducing the cost of irrationality.