Bad Apple!!
Kevin Lin (@linguinelabs in news write-ups) pipes Bad Apple!! through MaleCNS v1.0 and shows a NeuroMechFly body. Open-source; LIF parameters and movement programs are illustrative. Coverage also used the X handle; the repo is the verified source.
Neurons are simulated — leaky integrate-and-fire or better — with documented, tuned parameters.
Dispute this rating →The evidence
No evidence, no level.The integration kernel is a leaky integrate-and-fire model with explicit rest/reset, threshold, refractory count and membrane/synaptic decay constants.
https://github.com/kevinlinxc/badapple-fly/blob/6726892e7ebd93e6c5ffab056ab8db09433c28a6/backend/brain.py#L10-L40The connectome weight array is loaded memory-mapped read-only, so it cannot be modified at run time.
https://github.com/kevinlinxc/badapple-fly/blob/6726892e7ebd93e6c5ffab056ab8db09433c28a6/backend/brain.py#L48-L54Dopamine, serotonin, octopamine and unknown transmitters are assigned zero fast synaptic contribution - a sign convention, not a learning signal.
https://github.com/kevinlinxc/badapple-fly/blob/6726892e7ebd93e6c5ffab056ab8db09433c28a6/scripts/prepare_data.py#L55
The MaleCNS v1.0 graph is simulated as sparse, delay-queued leaky integrate-and-fire neurons with Shiu-style parameters (-52 mV rest and reset, -45 mV threshold, 20 ms membrane and 5 ms synaptic time constants, 2.2 ms refractory, 1.8 ms transmission delay, 0.1 ms step), and identified descending neurons drive a MuJoCo NeuroMechFly body, so simulated neural activity really does produce the output. Not L2: the synapse-count weight array is memory-mapped read-only and appears in the integration loop solely on the right-hand side of the conductance accumulation, so no synapse is ever written during a run; dopamine appears only as a neurotransmitter label that is explicitly given zero fast synaptic contribution. The README is unusually candid that the parameters are invented rather than measured and that the video is spatial stimulation rather than vision.
Record
Nearby
All projects →MaleCNS simulation connected to ViZDoom through modeled visual inputs and a fixed button readout. Includes plasticity experiments and negative validation results; learned survival has not been demonstrated.
Research framework comparing a MaleCNS-constrained sparse recurrent controller with rewired graphs, conventional neural networks, and other controls on simple VizDoom tasks.
Ratings on thelearningfly.com are proposals, not verdicts. Every one of them is arguable in public.