MirrorX

MirrorX asks what happens when a small society of agents can remember, reflect, copy behavior, disagree, and converge.

Research question

Can we make multi-agent cognition visible enough that students can inspect it, change it, and reason about it?

V1 experience

The first site version presents MirrorX as a project brief. It explains the thesis, expected experiment, and learning questions before the simulation exists.

Future experiment

The v3/v4 MirrorX experience becomes a live observatory:

• configure agent count, reflection intensity, randomness, and memory horizon
• run and replay scenarios
• watch belief states evolve over time
• inspect emergence, diversity, and collapse risk
• compare how rule changes alter group behavior

What we expect to learn

• when reflection improves coordination versus creates loops
• whether randomness prevents premature consensus
• how agent cultures form through shared language, shortcuts, and blind spots
• whether students can understand multi-agent cognition when it is visual and interactive