Learn by doing
Follow four short tutorials from one simulation to an external extension. Start the tutorials →
Behaviour from simple neural substrates
Compose nodes, bodies, tasks, and reproducible research operations in Julia.
Preparing the scored interval…
A small core with explicit research paths
BrainlessLab connects a node model, body, task, and evaluation protocol. The initial core contains the fixture-validated Falandays implementation and three benchmark tasks: Tracking, Pong, and Wall.
Use simulate for one diagnostic run. Use a typed operation plan when repeated work needs
exact settings, independent trials, and a portable record.
Understand the system
Follow runtime composition through evaluation, operations, experiments, and records. Open the handbook →
Browse research
Review benchmark and experiment protocols with exact record paths and evidence states. See research records →
Explore capabilities
Filter implemented features outside the compact core by family, readiness, and tag. Open Experimental →
Every standard record preserves the submitted and resolved configurations, realised seed streams, raw tables, summaries, provenance, and checksums. A record makes a run reviewable. It does not make an exploratory result confirmed.
A project of the Diverse Intelligences Summer Institute 2026, Geneva, New York.