A fly nervous system you can experiment with

What happens between seeing something approach and moving away? Fly CNS Activity Lab lets you explore that question by sending activity through a measured fruit-fly connectivity graph.

Open the fly arena Follow the five-minute tour

The idea behind the lab

A nervous system contains neurons that receive input, influence other neurons, and eventually affect muscles. A connectome describes which neurons connect to which others. In this lab, groups of neurons become computational units, and measured synapse counts determine the connections between them.

You can inject a visual signal, watch activity propagate, and silence selected populations to see how the modeled response changes. In the arena, motor readouts also move a small fly body. Its new position changes what it sees next, completing a feedback loop.

The anatomical data comes from Janelia’s MaleCNS project, which spans the central brain, optic lobes, and ventral nerve cord of a male fruit fly, including their connection through the neck. This makes sensory-to-motor routes accessible within one structural dataset. Read Janelia’s project overview.

EnvironmentAn object approaches.
Visual inputSize and expansion produce a signal.
CircuitActivity follows weighted connections.
MovementMotor readouts move the body.

Feedback: the new body position changes the next visual input. These are stages of the software loop, not four directly connected neurons.

What you are actually seeing

Part of the experiment	Where it comes from
Connections between populations	Aggregated, directed MaleCNS synapse counts.
Activity flowing through the graph	A bounded rate equation with assumed synaptic signs and response times.
The fly’s anatomical shape	Adapted display geometry from the Flybody project.
Visual sensing and flight forces	Explicit geometric and motor-to-force rules chosen for this prototype.

This is a simplified computational twin: it combines structural measurements with a model you can intervene in. It has no live connection to a biological fly and has not been calibrated against recorded flight. Its value is making assumptions and their consequences inspectable.

Choose the view that matches your question

View	Use it to ask	What runs
Fly in an arena	How can a circuit, body, and environment affect one another?	526 populations, with geometric visual feedback and illustrative movement.
Circuit experiment	Where does an input spread, and how does silencing change motor activity?	414 populations, with a prescribed visual-feature pulse.
Model & data	What are the equations, sources, limitations, and full-network comparisons?	Documentation and recorded full typed-CNS results. The full model is available in the Python download.

A population pools neurons of the same annotated cell type, side, and, where relevant, nerve-cord segment. The arena’s 526 units represent 9,615 neurons. Those neurons are not simulated separately.

Start with one useful comparison

Run the intact circuit. Open the arena and press Start experiment. Watch the fly, then look at the wing motor activity below.
Remove wing power. Select Silence DLM / DVM · wing power and start the new run.
Ask what remains. In the default model, a jump can remain while forward powered flight disappears. This helps separate the launch pathway from the wing-power readout.

That comparison is an intervention on this model. It suggests what to inspect next in the circuit and in the assumed force rules. It does not, on its own, establish the same causal effect in a living fly.

Try the comparison

Each link opens the relevant settings and waits for you to press Start.

Intact circuit Wing power silenced

A few useful terms

CNS: Central nervous system: here, the brain, optic lobes, and ventral nerve cord.
Synapse: A connection through which one neuron can influence another. The graph records direction and connection counts.
Descending neuron: A neuron carrying signals from the brain toward circuits in the nerve cord. DNp01 is one selected readout.
Motor neuron: A neuron associated with muscle control. TTMn, DLMn/DVMn, and b2 MN are used by the arena’s body adapter.
Silencing: Clamping selected model activity to zero and examining what changes downstream.
au: Arbitrary arena units. Height and speed are illustrative model quantities.

For the exact buttons, graph colors, export options, and troubleshooting, continue to How to use the lab. For the reasoning behind the experiment, start with What can a connectome-driven fly tell us?