NeuroFly
Closed-loop personal research project coupling FlyWire-based neural activity to a NeuroMechFly body, with recorded simulations and documented assumptions.
Neurons are simulated — leaky integrate-and-fire or better — with documented, tuned parameters.
Dispute this rating →The evidence
No evidence, no level.The neuron model is the published Shiu et al. LIF, with equations and cited biophysical constants.
https://github.com/seven-monarchs/NeuroFly/blob/b59264a34ccc181699dfc4db264c36afc1c5c103/brain_model/model.py#L18-L52Synaptic weights are assigned once from the connectome table at build time and never modified afterwards.
https://github.com/seven-monarchs/NeuroFly/blob/b59264a34ccc181699dfc4db264c36afc1c5c103/brain_model/model.py#L175-L183The closed loop updates Poisson input rates each 25 ms step, not synaptic weights.
https://github.com/seven-monarchs/NeuroFly/blob/b59264a34ccc181699dfc4db264c36afc1c5c103/fly_brain_body_simulation.py#L670-L690
Brian2 LIF network of 138,639 FlyWire v783 neurons using the Shiu et al. (2024) parameter set verbatim (v_0/v_rst -52 mV, v_th -45 mV, t_mbr 20 ms, tau 5 ms, t_rfc 2.2 ms, t_dly 1.8 ms), coupled in closed loop to a NeuroMechFly/MuJoCo body. Membrane voltages change under a documented neuron model, which earns L1. Not L2: syn.w is written once at network construction from the connectome file's 'Excitatory x Connectivity' column and is never touched again; the only thing the 25 ms closed-loop step updates is the PoissonGroup input rate for ascending and visual neurons, which is sensory stimulation, not synaptic change. The injected ascending and visual synapse weights are likewise set once during setup, before the run loop.
Record
Nearby
All projects →Research code for the Shiu et al. connectome-based leaky integrate-and-fire model, including activation/silencing experiments, notebooks, and FlyWire data configuration.
FlyWire whole-brain LIF implementation based on Shiu et al., with multiple simulation backends and benchmarking tools. This repository supplies the neural model; it is not a complete embodied demo package.
Ratings on thelearningfly.com are proposals, not verdicts. Every one of them is arguable in public.