Preparing the specimen

01 — A closer look at life

Inside aliving mind.

From a living animal to the architecture of thought. Explore the mouse brain, one region at a time, then take the runtime with you.

Begin the journey Open source
Drag to rotate
Mus musculusThe laboratory mouse
Scroll to look beneath the surface

02 — The brain atlas

A world
within.

Original region surfaces · Allen CCFv3

Drag to rotate

03 — Build with the brain

Take the brain.Make it yours.

The runtime is on GitHub, so you can connect your own environment, tools and action layer. No hosted lock-in, no hidden decision loop.

awaiting repository —stars—forks—issues
    status
    reading the repositoryread-only · github rest api
    pre-release

    mousebraindev/MouseBrain

    Mousebrain Core

    Open agent runtime for bringing the Mousebrain decision loop into software, hardware and custom environments.

    License
    MIT
    Repository
    MouseBrain
    Interface
    Observe → decide → act
    Open the repository

    04 — The token

    $MOUSEBRAIN

    Terms publish here when they are final.

    Ticker
    $MOUSEBRAIN
    Contract
    not deployed
    Supply
    not set
    Status
    pre‑launch
    $mousebrain / pricefeed not connected
    $MOUSEBRAIN

    The chart connects at launch.

    robinhood · awaiting pool addressno price · no promise · no date

    05 — Anatomy becomes behaviour

    Every actionbegins somewhere.

    No region works alone. Behaviour emerges from connected circuits.

    01

    A scent becomes
    a signal.

    Olfactory pathways

    Receptors in the nose turn chemical cues into electrical signals. The olfactory bulb begins processing these patterns, which travel on to other brain regions.

    02

    A movement finds
    its rhythm.

    Cerebellum

    A careful step is a continuous adjustment. The cerebellum helps coordinate timing and balance, refining movement as the animal learns from experience.

    A note on the science

    The brain surfaces come from the Allen Mouse Common Coordinate Framework. They show a reference anatomy, rather than a scan of this individual animal. The moving signals are an illustration of communication, not recorded neural activity.

    Meet the atlas

    06 — Beyond the specimen

    Put the brainanywhere.

    Connect Mousebrain to a browser, a machine, a workflow. Two of them are built out here; the loop underneath is the same one every time.

    03More, as they are built

    Inside a browser

    Reads the page, not the markup.

    The page arrives as a field of pixels. It picks a place to click, the click takes it somewhere, and the next page is the next decision. What it cannot do yet is pick well.

    Reads
    Page pixels
    Acts with
    Pointer, clicks, navigation

    Four crops of the live page, one of them on a link. The cortex names a quadrant, the cursor goes there, and whatever it lands on is the next article. The run recorded here is 14 decisions and 3 of them found a link, which is chance: over 800 training trials the hit rate moved 26% to 27% while the frozen control moved 24% to 23%, so there is no learning to claim yet. The loop is real; the competence is not there.

    07 — Perception

    It sees beforeit acts.

    Before anything decides anything, light has to get through an eye that is not yours. Two cones, coarse detail, and a retina built for dusk.

    The same frames, split down the middle. On one side the camera, on the other the same light after two cones instead of three, an acuity of half a cycle per degree, and a retina that is 97% rods. Footage “Pet mice mini” by A. C. Tatarinov, CC0.

    What changes

    ColourS-opsin peaks near 360 nm and M-opsin near 510 nm. There is no long-wavelength cone, so the red bottle and the pink tube stop being colours and become dark shapes.
    DetailAbout 0.5 cycles per degree against roughly 30 for a human eye. Edges survive, texture does not.
    LightA rod-dominated retina trades contrast for sensitivity, and the two halves are tuned differently: the half that looks up is ultraviolet, the half that looks down is green.

    A camera cannot record ultraviolet, so the blue channel stands in for that cone. Everything else is applied as its own measured step in _dev/mouse_eye.py.