Purpose

This simulation demonstrates how interactionist agents (the sharks) interact with their environment and with each other.

Remember that interactionist agents know nothing about their world when they get born. They organize their behavior through experience while accomplishing intrinsic drives. They initially tend to explore contrasted colors in the environment. Eventually, they discover that they can eat blue tiles and cuddle with other agents. Once this is discovered, they tend to preferably seek blue tiles and other agents rather than interacting with other kinds of objects. And then they tend to become quieter as they grow older.

In some instances, it may take a while before they learn to cuddle together. You can increase their odds to meet by luring them into the same area, or introducing additional agents. You can experience what it is like to terminate an agent, which, we believe, is still ethically OK :-).

Controls

• The 3D simulation should appear on the top of this page (better with Chrome). If not, please follow a WebGL troubleshooting procedure such as this one.
• Buttons: Play, Pause, Reset: control the simulation.
• Arrows: move the cursor (blue-edged cube) horizontally.
• Action Keys: insert Wall (W) at the cursor location, or Food (F), another agent (E), Yellow tiles (Y), Green tiles (G), or empty this location (SPACE).
• CTRL + arrows: translate the camera.
• ALT + arrows: rotate the camera.
• ALT + Page-Up/Page-Down: zoom in/out (ALT + fn + arrows on a Mac).
• TAB: toggle first-person camera.
• The tabs Ernest n show how Agent n sees its world: 12 pixel visual outcome, 3x3 tactile matrix, other outcomes: eat, cuddle, busmp. The spatial memory represents the agent's representation of its local surrounding environment. The control buttons let you pause, run a step at a time, or terminate this agent.

Credits

Simulation:

• Olivier Voisin (2011) Protocole de test pour agents développementaux et intrinsèquement motivés par une approche interactionniste.
• Source: seca repository on github.

Learning:

• Simon Gay, Olivier Georgeon, Jong Kim (2012). Implementing spatial awareness in an environment-agnostic agent. BRIMS2012 59-66.
• Source: Ernest repository on github.

Updated October 22nd 2021.