CAPYBRAINCAPYBRAIN studies how capybara-like behavioral drives can become a legible computational system and, eventually, a persistent virtual companion.
Balances novelty and safety so the companion can settle instead of reacting to every change.
The system begins with recognizable capybara patterns: calm regulation, social proximity, vigilance, routines and habitat response. Those concepts become controllable signals, not invented anatomy.
Translate calm, social contact, vigilance, routine and habitat into explicit model inputs.
OPEN METHODStudy how those drives interact inside a fixed, inspectable recurrent network.
OPEN METHODTurn network state into understandable tendencies without claiming emotion or consciousness.
OPEN METHODTest how local memory and repeated care can make a digital presence feel familiar.
OPEN METHODThe lab contains 54 abstract regions connected by fixed, signed weights. Five explicit behavioral drives provide the input. Eight recurrent updates combine new signals with previous state through a visible nonlinear process.
The layout is organized around calm, social proximity, habitat, routine memory and vigilance. It is an engineered model inspired by observable behavior, not a measured capybara connectome. The 3D graph, adjacency matrix and region inspector expose the same computation.
Start with visible behavior and environmental context.
Map five named drives to a normalized input vector.
Run eight recurrent updates through fixed signed links.
Convert the final state into an inspectable response tendency.
Move five behavioral drives, propagate eight updates, rotate the graph and inspect every region.
Open Neural LabCare, routine and local memory change a transparent rule-based companion over time.
Open TerminalCAPYBRAIN does not use or claim a complete capybara connectome. Its 54 regions and links are an engineered functional atlas, not a reconstruction of tissue, neurons or consciousness.
The value of the prototype is inspectability: every input, connection, update and readout can be seen. When future work uses real observations or external data, its source and limitation will be named.